Datasets:

---

DanielPFlorian

/

Transformers-Github-Issues

like 0

[]

[]

[ "You can go to the model hub, click on a model and select \"List all files in model\". These are links to the files.\r\n\r\n\r\n" ]

[]

[]

[ "Hi, do you mind copy-pasting the text in your terminal instead of putting images? It'll be easier for us to understand, to debug, and for other users to search for similar issues. Thanks!", "> Hi, do you mind copy-pasting the text in your terminal instead of putting images? It'll be easier for us to understand, to debug, and for other users to search for similar issues. Thanks!\r\n\r\nI've figure out the problem of gpu core. According to https://github.com/pytorch/pytorch/issues/31285, my gpu card is not supported by pytorch. I'm trying to build pytorch from source.\r\nBut it's weird when i using `run_language_modeling.py ` with `--no_cuda `. There is no error message.\r\nThe command i used is \r\n```shell\r\npython3 run_language_modeling.py --output_dir=/home/xxx/gpt_model/transformers/examples/language-modeling/output_dir --model_type=gpt2 --model_name_or_path=gpt2 --per_device_train_batch_size=1 --do_train --train_data_file=/home/xxx/gpt_model/data_info/transformer.data --block_size=512 --save_steps=500 --overwrite_output_dir --no_cuda\r\n\r\n```\r\nThe output message is \r\n```shell\r\n09/09/2020 16:23:23 - WARNING - __main__ - Process rank: -1, device: cpu, n_gpu: 0, distributed training: False, 16-bits training: False\r\n09/09/2020 16:23:23 - INFO - __main__ - Training/evaluation parameters TrainingArguments(output_dir='/home/xxx/gpt_model/transformers/examples/language-modeling/output_dir', overwrite_output_dir=True, do_train=True, do_eval=False, do_predict=False, evaluate_during_training=False, prediction_loss_only=False, per_device_train_batch_size=1, per_device_eval_batch_size=8, per_gpu_train_batch_size=None, per_gpu_eval_batch_size=None, gradient_accumulation_steps=1, learning_rate=5e-05, weight_decay=0.0, adam_beta1=0.9, adam_beta2=0.999, adam_epsilon=1e-08, max_grad_norm=1.0, num_train_epochs=3.0, max_steps=-1, warmup_steps=0, logging_dir='runs/Sep09_16-23-23_TENCENT64.site', logging_first_step=False, logging_steps=500, save_steps=500, save_total_limit=None, no_cuda=True, seed=42, fp16=False, fp16_opt_level='O1', local_rank=-1, tpu_num_cores=None, tpu_metrics_debug=False, debug=False, dataloader_drop_last=False, eval_steps=1000, past_index=-1, run_name=None, disable_tqdm=False, remove_unused_columns=True)\r\n/home/xxx/anaconda3/envs/transformers/lib/python3.6/site-packages/transformers/modeling_auto.py:821: FutureWarning: The class `AutoModelWithLMHead` is deprecated and will be removed in a future version. Please use `AutoModelForCausalLM` for causal language models, `AutoModelForMaskedLM` for masked language models and `AutoModelForSeq2SeqLM` for encoder-decoder models.\r\n FutureWarning,\r\n/home/xxx/anaconda3/envs/transformers/lib/python3.6/site-packages/transformers/tokenization_utils_base.py:1321: FutureWarning: The `max_len` attribute has been deprecated and will be removed in a future version, use `model_max_length` instead.\r\n FutureWarning,\r\n09/09/2020 16:23:31 - INFO - filelock - Lock 140250526322472 acquired on /home/xxx/gpt_model/data_info/cached_lm_GPT2Tokenizer_512_transformer.data.lock\r\n09/09/2020 16:23:32 - INFO - filelock - Lock 140250526322472 released on /home/xxx/gpt_model/data_info/cached_lm_GPT2Tokenizer_512_transformer.data.lock\r\n\r\n\r\n/home/xxx/anaconda3/envs/transformers/lib/python3.6/site-packages/transformers/trainer.py:247: FutureWarning: Passing `prediction_loss_only` as a keyword argument is deprecated and won't be possible in a future version. Use `args.prediction_loss_only` instead.\r\n FutureWarning,\r\n<library>In get_train_dataloader ,tarin_batch_size = 1\r\nEpoch: 0%| | 0/3 \r\n\r\nKilledion: 4%|████▎ | 194/5390 [04:30<2:58:02, 2.06s/it] \r\n```\r\nAnd `echo $?`, the return code is `137`. Thanks a lot.", "The return code 137 means that you have an out of memory error. Do you get the same error if you use `distilgpt2` with a `--block_size=64`? (Just for testing purposes).\r\n\r\nWe've also recently patched a memory error on the `Trainer`, could you install from source to benefit from the fix? You can do so as such:\r\n\r\n`pip install git+https://github.com/huggingface/transformers`", "> pip install git+https://github.com/huggingface/transformers\r\n\r\nYes, I found `Out of memory` log in `/var/log/messages` and it turns out the process use a lot of memory.\r\n```shell\r\nSep 9 17:10:44 centos kernel: Out of memory: Kill process 126138 (python3) score 939 or sacrifice child\r\nSep 9 17:10:44 centos kernel: Killed process 126170 (python3) total-vm:325690732kB, anon-rss:125105968kB, file-rss:0kB\r\n```\r\nAnd this is my machine info (caused in shell with `top` command)\r\n```shell\r\ntop - 17:47:55 up 303 days, 2:35, 4 users, load average: 1.14, 2.98, 3.44\r\nTasks: 468 total, 1 running, 467 sleeping, 0 stopped, 0 zombie\r\n%Cpu(s): 0.0 us, 0.0 sy, 0.0 ni,100.0 id, 0.0 wa, 0.0 hi, 0.0 si, 0.0 st\r\nKiB Mem : 13132300+total, 12383201+free, 3294300 used, 4196688 buff/cache\r\nKiB Swap: 2088956 total, 12 free, 2088944 used. 12565272+avail Mem \r\n\r\n PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND \r\n 2868 root 20 0 3006004 19072 0 S 0.3 0.0 225:15.55 dockerd \r\n```\r\nThen i use `pip uninstall transformers; pip install git+https://github.com/huggingface/transformers` to reinstall the `transformers` library, and run the `run_language_modeling.py` again.\r\n```shell\r\npython3 run_language_modeling.py --output_dir=/home/xxx/gpt_model/transformers/examples/language-modeling/output_dir --model_type=gpt2 --model_name_or_path=distilgpt2 --per_device_train_batch_size=1 --do_train --train_data_file=/home/xxx/gpt_model/data_info/data.txt --block_size=64 --save_steps=500 --overwrite_output_dir --no_cuda\r\n```\r\nThis is the memory info showed by `top` after i run `run_language_modeling.py` with cpu.\r\n```shell\r\ntop - 17:51:40 up 303 days, 2:38, 4 users, load average: 23.42, 9.30, 5.50\r\nTasks: 463 total, 2 running, 461 sleeping, 0 stopped, 0 zombie\r\n%Cpu(s): 34.4 us, 49.5 sy, 0.0 ni, 16.1 id, 0.0 wa, 0.0 hi, 0.0 si, 0.0 st\r\nKiB Mem : 13132300+total, 12142824+free, 5283664 used, 4611096 buff/cache\r\nKiB Swap: 2088956 total, 12 free, 2088944 used. 12365893+avail Mem \r\n\r\n PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND \r\n 25414 xxx 20 0 0.188t 1.945g 83876 R 3356 1.6 47:31.03 python3 \r\n```\r\nIt already runs about two hours and works well. Thanks a lot.", "By the way, how can i use the fine-tuned model please? There are 6 files generated under `output_dir/checkpoint`, which are \r\n`config.json log_history.json optimizer.pt pytorch_model.bin scheduler.pt training_args.bin`. Then how can i use them? Should i just use them as follows? Does the fine-tuned model need to be renamed? Which level of the checkpoint directory should I specify？ Thanks a lot.\r\n```shell\r\ntokenizer = GPT2Tokenizer.from_pretrained('/home/xxx/gpt_model/transformers/examples/language-modeling/output_dir/checkpoint-15000')\r\n``` \r\n", "Cool! Yes, the way you load the model is correct. I'm guessing you want to use the resulting model, so given that you passed the following as `--output_dir`: `--output_dir=/home/xxx/gpt_model/transformers/examples/language-modeling/output_dir`, you should be able to load it like\r\n\r\n```py\r\ntokenizer = GPT2Tokenizer.from_pretrained('/home/xxx/gpt_model/transformers/examples/language-modeling/output_dir')\r\nmodel = GPT2LMHeadModel.from_pretrained('/home/xxx/gpt_model/transformers/examples/language-modeling/output_dir')\r\n\r\n```", "> Cool! Yes, the way you load the model is correct. I'm guessing you want to use the resulting model, so given that you passed the following as `--output_dir`: `--output_dir=/home/xxx/gpt_model/transformers/examples/language-modeling/output_dir`, you should be able to load it like\r\n> \r\n> ```python\r\n> tokenizer = GPT2Tokenizer.from_pretrained('/home/xxx/gpt_model/transformers/examples/language-modeling/output_dir')\r\n> model = GPT2LMHeadModel.from_pretrained('/home/xxx/gpt_model/transformers/examples/language-modeling/output_dir')\r\n> ```\r\n\r\nGot it! \r\nI use `--output_dir`: `--output_dir=/home/xxx/gpt_model/transformers/examples/language-modeling/output_dir` and there are many checkpoint generate under it. Just choose one of them, for example, `checkpoint-14000` and copy `'vocab.json', 'merges.txt'` into `/home/xxx/gpt_model/transformers/examples/language-modeling/output_dir/checkpoint-14000`.\r\n```shell\r\n/home/xxx/gpt_model/transformers/examples/language-modeling/output_dir/checkpoint-14000 >>> wget https://cdn.huggingface.co/distilgpt2-vocab.json -O vocab.json\r\n/home/xxx/gpt_model/transformers/examples/language-modeling/output_dir/checkpoint-14000 >>> wget https://cdn.huggingface.co/distilgpt2-merges.txt -O merges.txt\r\n```\r\nThen i use the [run_generation.py](https://github.com/huggingface/transformers/blob/master/examples/text-generation/run_generation.py) like follows and it generate text as expect.\r\n```shell\r\npython run_generation.py --model_type=gpt2 --model_name_or_path=/home/xxx/gpt_model/transformers/examples/language-modeling/output_dir/checkpoint-14000 --no_cuda\r\n```\r\nThank you for your help! ", "Very cool, glad you got it to work! Let us know if you face any other issues.", "It seems like I still have the same issue. I try to use run_language_modeling.py to train a small-bert model (6 layer) from scratch. The process is killed after about 3 hours of training. The error message is simply \"Killed\" and I observes there's a constant increasing usage of memory. So I think the issue is also OOM\r\n\r\n```\r\ntop - 16:36:33 up 5:29, 1 user, load average: 1.07, 1.19, 1.16\r\nTasks: 353 total, 3 running, 349 sleeping, 0 stopped, 1 zombie\r\n%Cpu(s): 10.4 us, 1.2 sy, 0.0 ni, 88.1 id, 0.1 wa, 0.0 hi, 0.2 si, 0.0 st\r\nMiB Mem : 16010.9 total, 151.5 free, 3386.3 used, 12473.2 buff/cache\r\nMiB Swap: 2048.0 total, 1550.2 free, 497.8 used. 12181.5 avail Mem \r\n\r\n```\r\nIs this normal? I can't think of a reason why it would use this much memory." ]

[ { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "Hi, this is a very nice issue and I plan to work soon (in the coming 2 weeks) on related things (improving the examples to make full use of the Rust tokenization features). I'll re-read this issue (and all the links) to extract all the details and likely come back to you at that time.\r\n\r\nIn the meantime, here are two elements for your project:\r\n- the first is that for the fast tokenizers, the output of the tokenizer (a `BatchEncoding` instance) is actually a special kind of python dict with some super fast alignement methods powered by Rust, including a `char_to_token` alignement method that could maybe make your `align_tokens_to_annos` method a lot simpler and faster. You can read about it here: https://huggingface.co/transformers/main_classes/tokenizer.html#transformers.BatchEncoding.char_to_token\r\n- the second element is that support for sentence piece is almost there in `tokenizers` so we will soon be able to use fast tokenizers for almost all the models.", "Thanks! That's super helpful.\r\nI also found out I can iterate over the batch which is really nice. \r\n\r\nI did find a bug and [opened an issue](https://github.com/huggingface/tokenizers/issues/404) \r\n```python\r\nfrom transformers import BertTokenizerFast,GPT2TokenizerFast\r\ntokenizer = BertTokenizerFast.from_pretrained('bert-base-cased',)\r\n\r\n\r\nfor i in range(1,5):\r\n txt = \"💩\"*i\r\n enc = tokenizer(txt,return_offsets_mapping=True)\r\n token_at_i = enc.char_to_token(i-1)\r\n dec = tokenizer.decode(enc['input_ids'])\r\n \r\n print (f\" I wrote {txt} but got back '{dec}' and char_to_tokens({i-1}) returned {token_at_i}\")\r\n```\r\n```\r\n I wrote 💩 but got back '[CLS] [UNK] [SEP]' and char_to_tokens(0) returned 1\r\n I wrote 💩💩 but got back '[CLS] [UNK] [SEP]' and char_to_tokens(1) returned 1\r\n I wrote 💩💩💩 but got back '[CLS] [UNK] [SEP]' and char_to_tokens(2) returned 1\r\n I wrote 💩💩💩💩 but got back '[CLS] [UNK] [SEP]' and char_to_tokens(3) returned 1\r\n```", "## Progress - But Am I doing this right ? \r\nHi,\r\nI made some progress and got some logic that aligns the tokens to offset annotations. A function call looks like this\r\n```python\r\nbatch,labels = tokenize_with_labels(texts,annotations,tokenizer,label_set=label_set)\r\n```\r\n\r\nAnd then a visualization of the alignment looks like this (which is trying to show annotations across multiple tokens) \r\n\r\n\r\n## Question\r\n\r\nSo I'm trying to get the padding /batching working. I think I've got something good but would love some input on how this might subtly fail. \r\n\r\n```python\r\nbatch,labels = tokenize_with_labels(texts,annotations,tokenizer,label_set=label_set)\r\nbatch.data['labels'] =labels # Put the labels in the BatchEncoding dict\r\n\r\npadded = tokenizer.pad(batch,padding='longest') # Call pad, which ignores the labels and offset_mappings\r\nbatch_size = len(padded['input_ids'][0]) # Get the padded sequence size\r\nfor i in range(len(padded['labels'])): #for each label\r\n ls = padded['labels'][i]\r\n difference = batch_size - len(ls) # How much do we need to pad ? \r\n padded['labels'][i] = padded['labels'][i] +[0] *difference # Pad \r\n padded['offset_mapping'][i]+=[(0,0)]*difference #pad the offset mapping so we can call convert_to_tensors\r\n \r\n \r\ntensors = padded.convert_to_tensors(tensor_type='pt') #convert to a tensor\r\n```\r\n\r\n\r\n", "Hmm I think we should have an option to pad the labels in `tokenizer.pad`.\r\nEither based on the shape of the labels or with an additional flag.\r\nI'll work on the tokenizers this week. Will add this to the stack.", "I think that presupposes that the user has labels aligned to tokens, or that their is one and only one right way to align labels and tokens, which isn't consistent with the original issue. \r\n\r\nWhen that's not the case, then we need to tokenize, then align labels and finally pad. (Also need to deal with overflow, but I haven't gotten that far yet) . Notably, the user may want to use a BIO,BILSO or other schema and needs access to the tokens to modify the labels accordingly. \r\n\r\nSomething that confused me as I've been working on this is that the _pad function [operates explicitly on named attributes of the batch encoding dict](https://github.com/huggingface/transformers/blob/15478c1287a4e7b52c01730ffb0718243d153600/src/transformers/tokenization_utils_base.py#L2642-L2663) whereas as a user I'd expect it to operate on everything in the underlying ```encoding.data``` dict. That however doesn't work because the dict includes offset_mappings which don't tensorize nicely. \r\n\r\n\r\nBecause of the logic involved in alignment, I think that padding of the tokens and labels might be better done outside of the tokenizer, probably with a specialized function / module. \r\nThe upside of padding in one go is the efficiency of doing so in Rust, but I'd speculate that for token classification, the running time would be dominated by training anyway, and the efficiency gains wouldn't justify the API complexity or documentation burdon of doing it all in one place. \r\n\r\nAlso, I think that's a theoretical point because it seems that the padding is [done in python anyway](https://github.com/huggingface/transformers/blob/15478c1287a4e7b52c01730ffb0718243d153600/src/transformers/tokenization_utils_base.py#L2646) ? \r\n\r\nI ended up doing\r\n``` python\r\ndef tokenize_with_labels(\r\n texts: List[str],\r\n raw_labels: List[List[SpanAnnotation]],\r\n tokenizer: PreTrainedTokenizerFast,\r\n label_set: LabelSet, #Basically the alignment strategy\r\n):\r\n batch_encodings = tokenizer(\r\n texts,\r\n return_offsets_mapping=True,\r\n padding=\"longest\",\r\n max_length=256,\r\n truncation=True,\r\n )\r\n batch_labels: IntListList = []\r\n for encoding, annotations in zip(batch_encodings.encodings, raw_labels):\r\n batch_labels.append(label_set.align_labels_to_tokens(encoding, annotations))\r\n return batch_encodings, batch_labels\r\n```\r\n\r\nwhere align_labels_to_tokens operates on already padded tokens.\r\n\r\nI found this the most convenient way to get dynamic batches with a collator \r\n\r\n```python\r\n@dataclass\r\nclass LTCollator:\r\n tokenizer: PreTrainedTokenizerFast\r\n label_set: LabelSet\r\n padding: PaddingStrategy = True\r\n max_length: Optional[int] = None\r\n\r\n def __call__(self, texts_and_labels: Example) -> BatchEncoding:\r\n texts: List[str] = []\r\n annotations: List[List[SpanAnnotation]] = []\r\n for (text, annos) in texts_and_labels:\r\n texts.append(text)\r\n annotations.append(annos)\r\n\r\n batch, labels = tokenize_with_labels(\r\n texts, annotations, self.tokenizer, label_set=self.label_set\r\n )\r\n del batch[\"offset_mapping\"]\r\n batch.data[\"labels\"] = labels # Put the labels in the BatchEncoding dict\r\n tensors = batch.convert_to_tensors(tensor_type=\"pt\") # convert to a tensor\r\n return tensors\r\n```\r\n", "As an example of the end to end flow, (and please No one use this it's a probably buggy work in progress)\r\n```python\r\nfrom typing import Any, Optional, List, Tuple\r\nfrom transformers import (\r\n BertTokenizerFast,\r\n BertModel,\r\n BertForMaskedLM,\r\n BertForTokenClassification,\r\n TrainingArguments,\r\n)\r\nimport torch\r\nfrom transformers import AdamW, Trainer\r\n\r\nfrom dataclasses import dataclass\r\nfrom torch.utils.data import Dataset\r\nimport json\r\n\r\nfrom torch.utils.data.dataloader import DataLoader\r\nfrom transformers import PreTrainedTokenizerFast, DataCollatorWithPadding, BatchEncoding\r\nfrom transformers.tokenization_utils_base import PaddingStrategy\r\n\r\nfrom labelset import LabelSet\r\nfrom token_types import IntListList, SpanAnnotation\r\nfrom tokenize_with_labels import tokenize_with_labels\r\n\r\nExample = Tuple[str, List[List[SpanAnnotation]]]\r\n\r\n\r\n@dataclass\r\nclass LTCollator:\r\n tokenizer: PreTrainedTokenizerFast\r\n label_set: LabelSet\r\n padding: PaddingStrategy = True\r\n max_length: Optional[int] = None\r\n\r\n def __call__(self, texts_and_labels: Example) -> BatchEncoding:\r\n texts: List[str] = []\r\n annotations: List[List[SpanAnnotation]] = []\r\n for (text, annos) in texts_and_labels:\r\n texts.append(text)\r\n annotations.append(annos)\r\n\r\n batch, labels = tokenize_with_labels(\r\n texts, annotations, self.tokenizer, label_set=self.label_set\r\n )\r\n del batch[\"offset_mapping\"]\r\n batch.data[\"labels\"] = labels # Put the labels in the BatchEncoding dict\r\n tensors = batch.convert_to_tensors(tensor_type=\"pt\") # convert to a tensor\r\n return tensors\r\n\r\n\r\nclass LTDataset(Dataset):\r\n def __init__(\r\n self, data: Any, tokenizer: PreTrainedTokenizerFast,\r\n ):\r\n self.tokenizer = tokenizer\r\n for example in data[\"examples\"]:\r\n for a in example[\"annotations\"]:\r\n a[\"label\"] = a[\"tag\"]\r\n self.texts = []\r\n self.annotations = []\r\n for example in data[\"examples\"]:\r\n self.texts.append(example[\"content\"])\r\n self.annotations.append(example[\"annotations\"])\r\n\r\n def __len__(self):\r\n return len(self.texts)\r\n\r\n def __getitem__(self, idx) -> Example:\r\n\r\n return self.texts[idx], self.annotations[idx]\r\n\r\n\r\n@dataclass\r\nclass LTDataControls:\r\n dataset: LTDataset\r\n collator: LTCollator\r\n label_set: LabelSet\r\n\r\n\r\ndef lt_data_factory(\r\n json_path: str, tokenizer: PreTrainedTokenizerFast, max_length=None\r\n):\r\n data = json.load(open(json_path))\r\n dataset = LTDataset(data=data, tokenizer=tokenizer)\r\n tags = list(map(lambda x: x[\"name\"], data[\"schema\"][\"tags\"]))\r\n label_set = LabelSet(tags)\r\n collator = LTCollator(\r\n max_length=max_length, label_set=label_set, tokenizer=tokenizer\r\n )\r\n return LTDataControls(dataset=dataset, label_set=label_set, collator=collator)\r\n\r\n\r\nif __name__ == \"__main__\":\r\n from transformers import BertTokenizerFast, GPT2TokenizerFast\r\n\r\n tokenizer = BertTokenizerFast.from_pretrained(\"bert-base-cased\",)\r\n data_controls = lt_data_factory(\r\n \"/home/tal/Downloads/small_gold_no_paragr_location_types_false_5_annotations.json\",\r\n tokenizer=tokenizer,\r\n max_length=256,\r\n )\r\n dl = DataLoader(\r\n data_controls.dataset, collate_fn=data_controls.collator, batch_size=10\r\n )\r\n model = BertForTokenClassification.from_pretrained(\r\n \"bert-base-cased\", num_labels=len(data_controls.label_set.ids_to_label.values())\r\n )\r\n train = Trainer(\r\n model=model,\r\n data_collator=data_controls.collator,\r\n train_dataset=data_controls.dataset,\r\n args=TrainingArguments(\"/tmp/trainer\", per_device_train_batch_size=2),\r\n )\r\n train.train()\r\n```", "Also, I found [this comment](https://github.com/huggingface/transformers/issues/6860#issuecomment-684118716) by @sgugger about the trainer\r\n>Note that there are multiple frameworks that provide generic training loops. The goal of Trainer (I'm assuming you're talking about it since there is no train.py file) is **not to replace them or compete with them but to provide an easy way to train and finetune Transformers models**. Those models don't take nested inputs, so Trainer does not support this. Those models are expected to return the loss as the first item of their output, so Trainer expects it too.\r\n\r\nI think that sentiment might make sense here, that what I'm looking for is outside the scope of the library. If that's the case I would have preferred it be written in big bold letters, rather than the library trying to cater to this use case\r\n \r\n", "So,\r\nAfter much rabbit hole, I've written a blog post about the [considerations when doing alignment/padding/batching](https://www.lighttag.io/blog/sequence-labeling-with-transformers/) and another [walking through an implementation](https://www.lighttag.io/blog/sequence-labeling-with-transformers/example). \r\n\r\nIt even [comes with a repo](https://github.com/LightTag/sequence-labeling-with-transformers) \r\n\r\nso \r\nIf we have annotated data like this\r\n```python\r\n[{'annotations': [],\r\n 'content': 'No formal drug interaction studies of Aranesp? have been '\r\n 'performed.',\r\n 'metadata': {'original_id': 'DrugDDI.d390.s0'}},\r\n {'annotations': [{'end': 13, 'label': 'drug', 'start': 6, 'tag': 'drug'},\r\n {'end': 60, 'label': 'drug', 'start': 43, 'tag': 'drug'},\r\n {'end': 112, 'label': 'drug', 'start': 105, 'tag': 'drug'},\r\n {'end': 177, 'label': 'drug', 'start': 164, 'tag': 'drug'},\r\n {'end': 194, 'label': 'drug', 'start': 181, 'tag': 'drug'},\r\n {'end': 219, 'label': 'drug', 'start': 211, 'tag': 'drug'},\r\n {'end': 238, 'label': 'drug', 'start': 227, 'tag': 'drug'}],\r\n 'content': 'Since PLETAL is extensively metabolized by cytochrome P-450 '\r\n 'isoenzymes, caution should be exercised when PLETAL is '\r\n 'coadministered with inhibitors of C.P.A. such as ketoconazole '\r\n 'and erythromycin or inhibitors of CYP2C19 such as omeprazole.',\r\n 'metadata': {'original_id': 'DrugDDI.d452.s0'}},\r\n {'annotations': [{'end': 58, 'label': 'drug', 'start': 47, 'tag': 'drug'},\r\n {'end': 75, 'label': 'drug', 'start': 62, 'tag': 'drug'},\r\n {'end': 135, 'label': 'drug', 'start': 124, 'tag': 'drug'},\r\n {'end': 164, 'label': 'drug', 'start': 152, 'tag': 'drug'}],\r\n 'content': 'Pharmacokinetic studies have demonstrated that omeprazole and '\r\n 'erythromycin significantly increased the systemic exposure of '\r\n 'cilostazol and/or its major metabolites.',\r\n 'metadata': {'original_id': 'DrugDDI.d452.s1'}}]\r\n```\r\nWe can do this\r\n```python\r\nfrom sequence_aligner.labelset import LabelSet\r\nfrom sequence_aligner.dataset import TrainingDataset\r\nfrom sequence_aligner.containers import TraingingBatch\r\nimport json\r\nraw = json.load(open('./data/ddi_train.json'))\r\nfor example in raw:\r\n for annotation in example['annotations']:\r\n #We expect the key of label to be label but the data has tag\r\n annotation['label'] = annotation['tag']\r\n\r\nfrom torch.utils.data import DataLoader\r\nfrom transformers import BertForTokenClassification,AdamW\r\nmodel = BertForTokenClassification.from_pretrained(\r\n \"bert-base-cased\", num_labels=len(dataset.label_set.ids_to_label.values())\r\n)\r\noptimizer = AdamW(model.parameters(), lr=5e-6)\r\n\r\ndataloader = DataLoader(\r\n dataset,\r\n collate_fn=TraingingBatch,\r\n batch_size=4,\r\n shuffle=True,\r\n)\r\nfor num, batch in enumerate(dataloader):\r\n loss, logits = model(\r\n input_ids=batch.input_ids,\r\n attention_mask=batch.attention_masks,\r\n labels=batch.labels,\r\n )\r\n loss.backward()\r\n optimizer.step()\r\n\r\n\r\n-------------------------------\r\n\r\nI think most of this is out of scope for the transformers library itself, so am all for closing this issue if no one objects", "(I attempted to fix the links above, let me know if this is correct @talolard)", "> (I attempted to fix the links above, let me know if this is correct @talolard)\r\n\r\nLinks seem kosher, thanks", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n" ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/7018?src=pr&el=h1) Report\n> Merging [#7018](https://codecov.io/gh/huggingface/transformers/pull/7018?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/d6c08b07a087e83915b4b3156bbf464cebc7b9b5?el=desc) will **increase** coverage by `0.26%`.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/7018?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #7018 +/- ##\n==========================================\n+ Coverage 78.74% 79.01% +0.26% \n==========================================\n Files 168 164 -4 \n Lines 32172 30987 -1185 \n==========================================\n- Hits 25335 24483 -852 \n+ Misses 6837 6504 -333 \n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/7018?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/modeling\\_tf\\_longformer.py](https://codecov.io/gh/huggingface/transformers/pull/7018/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9sb25nZm9ybWVyLnB5) | `16.37% <0.00%> (-82.31%)` | :arrow_down: |\n| [src/transformers/modeling\\_tf\\_electra.py](https://codecov.io/gh/huggingface/transformers/pull/7018/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9lbGVjdHJhLnB5) | `25.13% <0.00%> (-73.83%)` | :arrow_down: |\n| [src/transformers/modeling\\_longformer.py](https://codecov.io/gh/huggingface/transformers/pull/7018/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19sb25nZm9ybWVyLnB5) | `19.71% <0.00%> (-72.34%)` | :arrow_down: |\n| [src/transformers/configuration\\_longformer.py](https://codecov.io/gh/huggingface/transformers/pull/7018/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9jb25maWd1cmF0aW9uX2xvbmdmb3JtZXIucHk=) | `75.00% <0.00%> (-25.00%)` | :arrow_down: |\n| [src/transformers/utils/logging.py](https://codecov.io/gh/huggingface/transformers/pull/7018/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy91dGlscy9sb2dnaW5nLnB5) | `75.00% <0.00%> (-10.90%)` | :arrow_down: |\n| [src/transformers/modeling\\_t5.py](https://codecov.io/gh/huggingface/transformers/pull/7018/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190NS5weQ==) | `76.70% <0.00%> (-6.07%)` | :arrow_down: |\n| [src/transformers/tokenization\\_albert.py](https://codecov.io/gh/huggingface/transformers/pull/7018/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy90b2tlbml6YXRpb25fYWxiZXJ0LnB5) | `87.50% <0.00%> (-5.77%)` | :arrow_down: |\n| [src/transformers/testing\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/7018/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy90ZXN0aW5nX3V0aWxzLnB5) | `65.89% <0.00%> (-3.50%)` | :arrow_down: |\n| [...rc/transformers/data/datasets/language\\_modeling.py](https://codecov.io/gh/huggingface/transformers/pull/7018/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9kYXRhL2RhdGFzZXRzL2xhbmd1YWdlX21vZGVsaW5nLnB5) | `90.80% <0.00%> (-2.14%)` | :arrow_down: |\n| [src/transformers/data/data\\_collator.py](https://codecov.io/gh/huggingface/transformers/pull/7018/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9kYXRhL2RhdGFfY29sbGF0b3IucHk=) | `91.90% <0.00%> (-0.93%)` | :arrow_down: |\n| ... and [40 more](https://codecov.io/gh/huggingface/transformers/pull/7018/diff?src=pr&el=tree-more) | |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/7018?src=pr&el=continue).\n> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta)\n> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`\n> Powered by [Codecov](https://codecov.io/gh/huggingface/transformers/pull/7018?src=pr&el=footer). Last update [d6c08b0...6d5adc4](https://codecov.io/gh/huggingface/transformers/pull/7018?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n" ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/7017?src=pr&el=h1) Report\n> Merging [#7017](https://codecov.io/gh/huggingface/transformers/pull/7017?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/5c4eb4b1ac45291e89c1be0fb1fdacd841b19a47?el=desc) will **decrease** coverage by `0.15%`.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/7017?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #7017 +/- ##\n==========================================\n- Coverage 80.39% 80.23% -0.16% \n==========================================\n Files 164 164 \n Lines 30986 30986 \n==========================================\n- Hits 24910 24863 -47 \n- Misses 6076 6123 +47 \n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/7017?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/modeling\\_tf\\_longformer.py](https://codecov.io/gh/huggingface/transformers/pull/7017/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9sb25nZm9ybWVyLnB5) | `16.37% <0.00%> (-82.31%)` | :arrow_down: |\n| [src/transformers/modeling\\_tf\\_electra.py](https://codecov.io/gh/huggingface/transformers/pull/7017/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9lbGVjdHJhLnB5) | `25.13% <0.00%> (-73.83%)` | :arrow_down: |\n| [src/transformers/modeling\\_gpt2.py](https://codecov.io/gh/huggingface/transformers/pull/7017/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19ncHQyLnB5) | `79.03% <0.00%> (-7.80%)` | :arrow_down: |\n| [src/transformers/modeling\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/7017/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ191dGlscy5weQ==) | `84.14% <0.00%> (-3.07%)` | :arrow_down: |\n| [src/transformers/generation\\_tf\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/7017/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9nZW5lcmF0aW9uX3RmX3V0aWxzLnB5) | `83.45% <0.00%> (-1.76%)` | :arrow_down: |\n| [src/transformers/data/data\\_collator.py](https://codecov.io/gh/huggingface/transformers/pull/7017/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9kYXRhL2RhdGFfY29sbGF0b3IucHk=) | `91.90% <0.00%> (-0.81%)` | :arrow_down: |\n| [src/transformers/modeling\\_tf\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/7017/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl91dGlscy5weQ==) | `86.97% <0.00%> (-0.33%)` | :arrow_down: |\n| [src/transformers/pipelines.py](https://codecov.io/gh/huggingface/transformers/pull/7017/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9waXBlbGluZXMucHk=) | `81.00% <0.00%> (+0.24%)` | :arrow_up: |\n| [src/transformers/configuration\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/7017/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9jb25maWd1cmF0aW9uX3V0aWxzLnB5) | `96.64% <0.00%> (+0.67%)` | :arrow_up: |\n| [src/transformers/tokenization\\_auto.py](https://codecov.io/gh/huggingface/transformers/pull/7017/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy90b2tlbml6YXRpb25fYXV0by5weQ==) | `97.87% <0.00%> (+2.12%)` | :arrow_up: |\n| ... and [7 more](https://codecov.io/gh/huggingface/transformers/pull/7017/diff?src=pr&el=tree-more) | |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/7017?src=pr&el=continue).\n> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta)\n> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`\n> Powered by [Codecov](https://codecov.io/gh/huggingface/transformers/pull/7017?src=pr&el=footer). Last update [5c4eb4b...4de36a2](https://codecov.io/gh/huggingface/transformers/pull/7017?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n" ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/7016?src=pr&el=h1) Report\n> Merging [#7016](https://codecov.io/gh/huggingface/transformers/pull/7016?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/d31031f603043281d4fbac6cbdcfb6497fd500ab?el=desc) will **decrease** coverage by `3.56%`.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/7016?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #7016 +/- ##\n==========================================\n- Coverage 80.03% 76.47% -3.57% \n==========================================\n Files 161 161 \n Lines 30120 30120 \n==========================================\n- Hits 24108 23033 -1075 \n- Misses 6012 7087 +1075 \n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/7016?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/modeling\\_longformer.py](https://codecov.io/gh/huggingface/transformers/pull/7016/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19sb25nZm9ybWVyLnB5) | `92.02% <ø> (ø)` | |\n| [src/transformers/configuration\\_reformer.py](https://codecov.io/gh/huggingface/transformers/pull/7016/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9jb25maWd1cmF0aW9uX3JlZm9ybWVyLnB5) | `21.62% <0.00%> (-78.38%)` | :arrow_down: |\n| [src/transformers/modeling\\_reformer.py](https://codecov.io/gh/huggingface/transformers/pull/7016/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19yZWZvcm1lci5weQ==) | `16.87% <0.00%> (-77.64%)` | :arrow_down: |\n| [src/transformers/modeling\\_tf\\_mobilebert.py](https://codecov.io/gh/huggingface/transformers/pull/7016/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9tb2JpbGViZXJ0LnB5) | `24.55% <0.00%> (-72.36%)` | :arrow_down: |\n| [src/transformers/configuration\\_mobilebert.py](https://codecov.io/gh/huggingface/transformers/pull/7016/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9jb25maWd1cmF0aW9uX21vYmlsZWJlcnQucHk=) | `26.47% <0.00%> (-70.59%)` | :arrow_down: |\n| [src/transformers/modeling\\_mobilebert.py](https://codecov.io/gh/huggingface/transformers/pull/7016/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19tb2JpbGViZXJ0LnB5) | `23.49% <0.00%> (-65.97%)` | :arrow_down: |\n| [src/transformers/modeling\\_transfo\\_xl\\_utilities.py](https://codecov.io/gh/huggingface/transformers/pull/7016/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190cmFuc2ZvX3hsX3V0aWxpdGllcy5weQ==) | `52.98% <0.00%> (-13.44%)` | :arrow_down: |\n| [src/transformers/modeling\\_transfo\\_xl.py](https://codecov.io/gh/huggingface/transformers/pull/7016/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190cmFuc2ZvX3hsLnB5) | `67.10% <0.00%> (-12.67%)` | :arrow_down: |\n| [src/transformers/tokenization\\_roberta.py](https://codecov.io/gh/huggingface/transformers/pull/7016/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy90b2tlbml6YXRpb25fcm9iZXJ0YS5weQ==) | `87.67% <0.00%> (-10.96%)` | :arrow_down: |\n| [src/transformers/tokenization\\_utils\\_base.py](https://codecov.io/gh/huggingface/transformers/pull/7016/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy90b2tlbml6YXRpb25fdXRpbHNfYmFzZS5weQ==) | `86.72% <0.00%> (-7.32%)` | :arrow_down: |\n| ... and [14 more](https://codecov.io/gh/huggingface/transformers/pull/7016/diff?src=pr&el=tree-more) | |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/7016?src=pr&el=continue).\n> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta)\n> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`\n> Powered by [Codecov](https://codecov.io/gh/huggingface/transformers/pull/7016?src=pr&el=footer). Last update [d31031f...2f15fa5](https://codecov.io/gh/huggingface/transformers/pull/7016?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n" ]

[]

[]

[ "That is 100% correct - thanks for notifying us :-) PR is linked.", "Also what is the 1, 4, 21 in that example? Can someone answer this?", "Just random indices that will be attended and will attend globally", "So if we are passing a sentence during inferencing `I love Hugging face` and I give value `attention_mask[:, [0,-1]] = 2` then ` <s> and </s>` will be attended globally?? And then I can fetch `<s>` embedding to get embedding of the sentence?" ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/7014?src=pr&el=h1) Report\n> Merging [#7014](https://codecov.io/gh/huggingface/transformers/pull/7014?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/ce37be9d94da57897cce9c49b3421e6a8a927d4a?el=desc) will **increase** coverage by `2.40%`.\n> The diff coverage is `20.00%`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/7014?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #7014 +/- ##\n==========================================\n+ Coverage 77.60% 80.01% +2.40% \n==========================================\n Files 161 161 \n Lines 30120 30124 +4 \n==========================================\n+ Hits 23374 24103 +729 \n+ Misses 6746 6021 -725 \n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/7014?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/modeling\\_pegasus.py](https://codecov.io/gh/huggingface/transformers/pull/7014/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19wZWdhc3VzLnB5) | `69.23% <20.00%> (-30.77%)` | :arrow_down: |\n| [src/transformers/modeling\\_tf\\_electra.py](https://codecov.io/gh/huggingface/transformers/pull/7014/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9lbGVjdHJhLnB5) | `25.13% <0.00%> (-73.83%)` | :arrow_down: |\n| [src/transformers/modeling\\_marian.py](https://codecov.io/gh/huggingface/transformers/pull/7014/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19tYXJpYW4ucHk=) | `60.00% <0.00%> (-30.00%)` | :arrow_down: |\n| [src/transformers/activations.py](https://codecov.io/gh/huggingface/transformers/pull/7014/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9hY3RpdmF0aW9ucy5weQ==) | `85.00% <0.00%> (-5.00%)` | :arrow_down: |\n| [src/transformers/modeling\\_bart.py](https://codecov.io/gh/huggingface/transformers/pull/7014/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19iYXJ0LnB5) | `93.90% <0.00%> (-0.51%)` | :arrow_down: |\n| [src/transformers/modeling\\_tf\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/7014/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl91dGlscy5weQ==) | `86.97% <0.00%> (-0.33%)` | :arrow_down: |\n| [src/transformers/tokenization\\_utils\\_base.py](https://codecov.io/gh/huggingface/transformers/pull/7014/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy90b2tlbml6YXRpb25fdXRpbHNfYmFzZS5weQ==) | `93.76% <0.00%> (-0.28%)` | :arrow_down: |\n| [src/transformers/generation\\_tf\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/7014/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9nZW5lcmF0aW9uX3RmX3V0aWxzLnB5) | `86.71% <0.00%> (+2.50%)` | :arrow_up: |\n| [src/transformers/modeling\\_lxmert.py](https://codecov.io/gh/huggingface/transformers/pull/7014/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19seG1lcnQucHk=) | `90.76% <0.00%> (+20.74%)` | :arrow_up: |\n| ... and [3 more](https://codecov.io/gh/huggingface/transformers/pull/7014/diff?src=pr&el=tree-more) | |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/7014?src=pr&el=continue).\n> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta)\n> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`\n> Powered by [Codecov](https://codecov.io/gh/huggingface/transformers/pull/7014?src=pr&el=footer). Last update [ce37be9...f841173](https://codecov.io/gh/huggingface/transformers/pull/7014?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n" ]

[]

[]

[]

[ { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n" ]

[]

[]

[ "@patil-suraj \r\nCan you help me with this?" ]

[]

[]

[ "Upon further updates, I think the issue was with using wrong AutoModel function. I replaced it with AutoModelforSequenceClassification at the time of download and it worked." ]

[]

[]

[ "Pinging @sshleifer, the summarization master", "See https://discuss.huggingface.co/t/summarization-on-long-documents/920/2, and feel free to reply there!\r\nI don't have a code snippet but feel free to contribute one to that discussion! " ]

[]

[]

[ "Hey @dakshvar22,\r\n\r\nThanks for posting this Diverse Beam Search paper. Looks cool :-) \r\nAt the moment, I won't find the time to take a deeper look, but feel free to open a PR to add it to our generate() method and I'm happy to discuss possible design choices :-) ", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n", "My understanding is that this should have been closed by https://github.com/huggingface/transformers/pull/9006?" ]

[ { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n" ]

[]

[ { "login": "sgugger", "id": 35901082, "node_id": "MDQ6VXNlcjM1OTAxMDgy", "avatar_url": "https://avatars.githubusercontent.com/u/35901082?v=4", "gravatar_id": "", "url": "https://api.github.com/users/sgugger", "html_url": "https://github.com/sgugger", "followers_url": "https://api.github.com/users/sgugger/followers", "following_url": "https://api.github.com/users/sgugger/following{/other_user}", "gists_url": "https://api.github.com/users/sgugger/gists{/gist_id}", "starred_url": "https://api.github.com/users/sgugger/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/sgugger/subscriptions", "organizations_url": "https://api.github.com/users/sgugger/orgs", "repos_url": "https://api.github.com/users/sgugger/repos", "events_url": "https://api.github.com/users/sgugger/events{/privacy}", "received_events_url": "https://api.github.com/users/sgugger/received_events", "type": "User", "site_admin": false } ]

[ "Hello! On what version of `transformers` are you running?", "> Hello! On what version of `transformers` are you running?\r\n\r\nHi @LysandreJik, It's 3.1.0", "@TikaToka Faced the same problem. Just installed everything again from master and it worked. \r\n", "> @TikaToka Faced the same problem. Just installed everything again from master and it worked.\r\n\r\nI'm still having a same issue :(", "You need an install from source to use the current examples (as stated in their [README](https://github.com/huggingface/transformers/tree/master/examples)). In colab you can do so by executing a cell with\r\n```\r\n! pip install git+git://github.com/huggingface/transformers/\r\n```\r\n\r\nAlternatively, you can find the version of the example that work with 3.1.0 [here](https://github.com/huggingface/transformers/tree/v3.1.0/examples/language-modeling).", "\r\n\r\n\r\n> @TikaToka Faced the same problem. Just installed everything again from master and it worked.\r\n\r\nThis Worked Thanks! I made a mistake while installing ", "> You need an install from source to use the current examples (as stated in their [README](https://github.com/huggingface/transformers/tree/master/examples)). In colab you can do so by executing a cell with\r\n> \r\n> ```\r\n> ! pip install git+git://github.com/huggingface/transformers/\r\n> ```\r\n> \r\n> Alternatively, you can find the version of the example that work with 3.1.0 [here](https://github.com/huggingface/transformers/tree/v3.1.0/examples/language-modeling).\r\n\r\nThank you for specific explanation! this worked!", "I still get the error:\r\nTypeError: __init__() got an unexpected keyword argument 'cache_dir'\r\nwhen running the latest version for transformers (3.1.0). \r\nI'm also running on Colab environment:\r\nCommand:\r\n!pip3 install transformers\r\n(also tried #! pip3 install git+git://github.com/huggingface/transformers/)\r\n\r\n!wget https://raw.githubusercontent.com/huggingface/transformers/master/examples/language-modeling/run_language_modeling.py\r\n\r\n%%bash\r\nexport TRAIN_FILE=train_path\r\nexport TEST_FILE=valid_path\r\nexport MODEL_NAME=gpt2\r\nexport OUTPUT_DIR=output\r\n\r\npython run_language_modeling.py \\\r\n --output_dir=output \\\r\n --model_type=gpt2 \\\r\n --model_name_or_path=gpt2 \\\r\n --do_train \\\r\n --train_data_file=$TRAIN_FILE \\\r\n --do_eval \\\r\n --eval_data_file=$TEST_FILE \\\r\n --cache_dir=None\r\n\r\nOutput:\r\nTraceback (most recent call last):\r\n File \"run_language_modeling.py\", line 313, in <module>\r\n main()\r\n File \"run_language_modeling.py\", line 242, in main\r\n get_dataset(data_args, tokenizer=tokenizer, cache_dir=model_args.cache_dir) if training_args.do_train else None\r\n File \"run_language_modeling.py\", line 143, in get_dataset\r\n cache_dir=cache_dir,\r\nTypeError: __init__() got an unexpected keyword argument 'cache_dir'" ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/7005?src=pr&el=h1) Report\n> Merging [#7005](https://codecov.io/gh/huggingface/transformers/pull/7005?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/c18f5916a03d0a161d003e95ffff8120d8addc0c?el=desc) will **decrease** coverage by `0.59%`.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/7005?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #7005 +/- ##\n==========================================\n- Coverage 80.63% 80.03% -0.60% \n==========================================\n Files 161 161 \n Lines 30123 30123 \n==========================================\n- Hits 24289 24109 -180 \n- Misses 5834 6014 +180 \n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/7005?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/modeling\\_tf\\_electra.py](https://codecov.io/gh/huggingface/transformers/pull/7005/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9lbGVjdHJhLnB5) | `25.13% <0.00%> (-73.83%)` | :arrow_down: |\n| [src/transformers/data/data\\_collator.py](https://codecov.io/gh/huggingface/transformers/pull/7005/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9kYXRhL2RhdGFfY29sbGF0b3IucHk=) | `91.90% <0.00%> (-0.81%)` | :arrow_down: |\n| [src/transformers/modeling\\_tf\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/7005/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl91dGlscy5weQ==) | `86.97% <0.00%> (-0.33%)` | :arrow_down: |\n| [src/transformers/file\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/7005/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9maWxlX3V0aWxzLnB5) | `82.66% <0.00%> (+0.25%)` | :arrow_up: |\n| [src/transformers/modeling\\_tf\\_flaubert.py](https://codecov.io/gh/huggingface/transformers/pull/7005/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9mbGF1YmVydC5weQ==) | `88.34% <0.00%> (+63.80%)` | :arrow_up: |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/7005?src=pr&el=continue).\n> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta)\n> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`\n> Powered by [Codecov](https://codecov.io/gh/huggingface/transformers/pull/7005?src=pr&el=footer). Last update [c18f591...0d1e538](https://codecov.io/gh/huggingface/transformers/pull/7005?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n" ]

[]

[]

[ "@sshleifer : Does it make sense to still add this? If yes, I can rebase and update based on Suraj's comments earlier. If not, will close.", "Let's hold off for now. I think passing --eval_num_beams=1 is close enough to equivalent.\r\nThanks for trying!" ]

[]

[]

[ "It's my issue" ]

[]

[]

[ "Hello @stas00 ,\r\n\r\nCan you show us a sample of your code ?\r\n\r\nDid you explicitely add special tokens to your tokenizer ?\r\n\r\nFrom my understanding this warning appears when the method _sanitize_special_tokens_ of the tokenizer returns a strictly positive integer.\r\n\r\nThe docstring of the method is:\r\n```\r\n\"\"\"\r\n Make sure that all the special tokens attributes of the tokenizer (:obj:`tokenizer.mask_token`,\r\n :obj:`tokenizer.cls_token`, etc.) are in the vocabulary.\r\n Add the missing ones to the vocabulary if needed.\r\n Return:\r\n :obj:`int`: The number of tokens added in the vocaulary during the operation.\r\n\"\"\"\r\n```\r\n\r\nSo this warning appears when you add special tokens to the vocabulary **after** loading the tokenizer. If you use a model trained on the first version of the tokenizer (before adding the new tokens), you might feed it tokens it has not been trained on, which would lead to a random embedding and worse performance.\r\n\r\nIf you load your model and your tokenizer with the same training, for example:\r\n```\r\ntokenizer = BertTokenizer.from_pretrained('bert-base-uncased')\r\nmodel = BertModel.from_pretrained('bert-base-uncased')\r\n```\r\nthen I suggest not providing special tokens, as the basic ones are already present.\r\n\r\nPlease let me know if it helps.", "You're absolutely correct, the tokenizer was adding special tokens (copied from another tokenizer, but wasn't really needing them), so I removed them now and the warning is gone.\r\n\r\nAnd, yes, I forgot to add the code - wasn't my best!\r\n\r\nMuch appreciating your follow up, @nassim-yagoub ", "The same warning also happens with this code:\r\n```\r\nimport torch\r\nfrom transformers import AutoModel, AutoTokenizer\r\n\r\ntokenizer = AutoTokenizer.from_pretrained('vinai/bertweet-base')\r\nmodel = AutoModel.from_pretrained('vinai/bertweet-base')\r\n```\r\n```\r\npython --version\r\nPython 3.9.13\r\n```\r\nI do not understand where tokens are added to the vocabulary after loading the tokenizer.", "@sbocconi Any idea how to solve the warnings in the case of BERTTweet?\r\n", "No unfortunately @codepujan, it has been a while I have not used this functionality", "This worked for me:\r\n\r\ntransformers.logging.set_verbosity_error()" ]

[]

[]

[]

[]

[]

[ "I realized that in QA, word/token embeddings are used, while I was looking for multi-sentence level embeddings. Pl. ignore my question." ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6999?src=pr&el=h1) Report\n> Merging [#6999](https://codecov.io/gh/huggingface/transformers/pull/6999?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/90ec78b5140251f093f658ebd4d2925e8c03f5e6?el=desc) will **decrease** coverage by `1.44%`.\n> The diff coverage is `100.00%`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/6999?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #6999 +/- ##\n==========================================\n- Coverage 80.58% 79.14% -1.45% \n==========================================\n Files 161 161 \n Lines 30123 30123 \n==========================================\n- Hits 24276 23841 -435 \n- Misses 5847 6282 +435 \n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/6999?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/trainer.py](https://codecov.io/gh/huggingface/transformers/pull/6999/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy90cmFpbmVyLnB5) | `54.95% <100.00%> (ø)` | |\n| [src/transformers/modeling\\_tf\\_longformer.py](https://codecov.io/gh/huggingface/transformers/pull/6999/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9sb25nZm9ybWVyLnB5) | `16.37% <0.00%> (-82.31%)` | :arrow_down: |\n| [src/transformers/modeling\\_lxmert.py](https://codecov.io/gh/huggingface/transformers/pull/6999/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19seG1lcnQucHk=) | `70.01% <0.00%> (-20.75%)` | :arrow_down: |\n| [src/transformers/modeling\\_roberta.py](https://codecov.io/gh/huggingface/transformers/pull/6999/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19yb2JlcnRhLnB5) | `77.73% <0.00%> (-19.35%)` | :arrow_down: |\n| [src/transformers/generation\\_tf\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6999/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9nZW5lcmF0aW9uX3RmX3V0aWxzLnB5) | `84.21% <0.00%> (-2.26%)` | :arrow_down: |\n| [src/transformers/tokenization\\_utils\\_base.py](https://codecov.io/gh/huggingface/transformers/pull/6999/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy90b2tlbml6YXRpb25fdXRpbHNfYmFzZS5weQ==) | `94.03% <0.00%> (+0.27%)` | :arrow_up: |\n| [src/transformers/modeling\\_bart.py](https://codecov.io/gh/huggingface/transformers/pull/6999/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19iYXJ0LnB5) | `94.41% <0.00%> (+0.50%)` | :arrow_up: |\n| [src/transformers/activations.py](https://codecov.io/gh/huggingface/transformers/pull/6999/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9hY3RpdmF0aW9ucy5weQ==) | `90.00% <0.00%> (+5.00%)` | :arrow_up: |\n| [src/transformers/data/data\\_collator.py](https://codecov.io/gh/huggingface/transformers/pull/6999/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9kYXRhL2RhdGFfY29sbGF0b3IucHk=) | `92.30% <0.00%> (+5.66%)` | :arrow_up: |\n| [src/transformers/modeling\\_marian.py](https://codecov.io/gh/huggingface/transformers/pull/6999/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19tYXJpYW4ucHk=) | `90.00% <0.00%> (+30.00%)` | :arrow_up: |\n| ... and [1 more](https://codecov.io/gh/huggingface/transformers/pull/6999/diff?src=pr&el=tree-more) | |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/6999?src=pr&el=continue).\n> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta)\n> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`\n> Powered by [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6999?src=pr&el=footer). Last update [90ec78b...240b7da](https://codecov.io/gh/huggingface/transformers/pull/6999?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n", "I think this was touched by @jysohn23 recently so pinging him here", "I think the problem is not that we don't call `.item()` but that we don't call `.detach()`, which means some variables are kept forever for a backward pass (that is never called).\r\nThe `.item()` are removed because it's needed for faster TPU training.", "This ends up calling `.item()` every step which ends up hurting performance by like 2X. What @sgugger sounds promising. Can we try that out instead?", "I have updated it to create a detached tensor instead of calling item().", "Can you confirm it fixes the memory leak? This is the right fix IMO (@LysandreJik this might be the fix for the TPU memory leak we have in another issue too.)", "> Can you confirm it fixes the memory leak? This is the right fix IMO (@LysandreJik this might be the fix for the TPU memory leak we have in another issue too.)\r\n\r\nGreat! Yes, I have tested to make sure that this fixes the leak." ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6998?src=pr&el=h1) Report\n> Merging [#6998](https://codecov.io/gh/huggingface/transformers/pull/6998?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/0054a48cdd64e7309184a64b399ab2c58d75d4e5?el=desc) will **increase** coverage by `0.19%`.\n> The diff coverage is `20.00%`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/6998?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #6998 +/- ##\n==========================================\n+ Coverage 80.53% 80.72% +0.19% \n==========================================\n Files 168 168 \n Lines 32179 32197 +18 \n==========================================\n+ Hits 25915 25991 +76 \n+ Misses 6264 6206 -58 \n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/6998?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/trainer\\_tf.py](https://codecov.io/gh/huggingface/transformers/pull/6998/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy90cmFpbmVyX3RmLnB5) | `12.46% <20.00%> (+0.24%)` | :arrow_up: |\n| [src/transformers/modeling\\_tf\\_flaubert.py](https://codecov.io/gh/huggingface/transformers/pull/6998/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9mbGF1YmVydC5weQ==) | `24.53% <0.00%> (-63.81%)` | :arrow_down: |\n| [src/transformers/modeling\\_marian.py](https://codecov.io/gh/huggingface/transformers/pull/6998/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19tYXJpYW4ucHk=) | `60.00% <0.00%> (-30.00%)` | :arrow_down: |\n| [src/transformers/modeling\\_openai.py](https://codecov.io/gh/huggingface/transformers/pull/6998/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19vcGVuYWkucHk=) | `72.25% <0.00%> (-10.00%)` | :arrow_down: |\n| [src/transformers/activations.py](https://codecov.io/gh/huggingface/transformers/pull/6998/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9hY3RpdmF0aW9ucy5weQ==) | `85.00% <0.00%> (-5.00%)` | :arrow_down: |\n| [src/transformers/modeling\\_bart.py](https://codecov.io/gh/huggingface/transformers/pull/6998/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19iYXJ0LnB5) | `93.93% <0.00%> (-0.51%)` | :arrow_down: |\n| [src/transformers/generation\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6998/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9nZW5lcmF0aW9uX3V0aWxzLnB5) | `96.92% <0.00%> (-0.28%)` | :arrow_down: |\n| [src/transformers/tokenization\\_utils\\_base.py](https://codecov.io/gh/huggingface/transformers/pull/6998/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy90b2tlbml6YXRpb25fdXRpbHNfYmFzZS5weQ==) | `93.77% <0.00%> (-0.28%)` | :arrow_down: |\n| [src/transformers/file\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6998/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9maWxlX3V0aWxzLnB5) | `82.41% <0.00%> (-0.26%)` | :arrow_down: |\n| [src/transformers/modeling\\_tf\\_xlm.py](https://codecov.io/gh/huggingface/transformers/pull/6998/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl94bG0ucHk=) | `93.26% <0.00%> (+4.84%)` | :arrow_up: |\n| ... and [2 more](https://codecov.io/gh/huggingface/transformers/pull/6998/diff?src=pr&el=tree-more) | |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/6998?src=pr&el=continue).\n> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta)\n> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`\n> Powered by [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6998?src=pr&el=footer). Last update [0054a48...b5c5bdc](https://codecov.io/gh/huggingface/transformers/pull/6998?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n", "> \r\n> \r\n> Thanks! It still needs some refactoring but in overall it looks ok. Can you also try the examples scripts for sequence classification and multiple choices to see if there is a drop in performance or not and be sure to do not bring any inconsistency.\r\n> \r\n> Optionaly if you have the possibility to run it, also the examples scripts, on a multi-gpu environment and check the same thing it would be appreciated, otherwise, no worries I will do it before merging :)\r\n\r\nI will try to (and learn to) run the other scripts later today, but I can only run on a single-gpu environment.", "No problem I can do it in multiple gpu env.", "The trainer will still be task-agnostic, the goal is just to add a new parameter to the training_step function (or possibly a class field) to handle the value that will be used to compute the scaled loss instead. It should work for all the tasks.\r\n\r\nThis computation cannot be done in the loss, because the loss computation is done over a per replica batch size and not over the global batch size.\r\n\r\nI'm not in favor of having different trainers. I don't mind having few differences between the two trainers as long as the external usage is the same.", "I don't understand your last comment. Having users subclass Trainer for specific behavior is what is indicated in the documentation. We cannot offer everything any user can think of in the training loop so this the way of customizing one. There is a `Seq2SeqTrainer` in preparation on the PyTorch side for instance, for code that is relevant to this only.\r\n\r\nOpening the door to have task-specific components in the main tf_trainer file will make it unreadable in a few months, when every user will have added their own, and then users that rely on a custom Trainers won't use that class anymore, because they won't understand it.\r\n\r\nTagging @julien-c for his advice.", "@sgugger , if fact, I also doubt that the pytorch trainer, while working with token level tasks, have some inaccurate loss computation - when we have distributed training and/or gradient accumulation. For example, [in DistilBertForTokenClassification](https://github.com/huggingface/transformers/blob/master/src/transformers/modeling_distilbert.py#L820), we use\r\n\r\n loss_fct = CrossEntropyLoss()\r\n\r\nand \r\n \r\n loss = loss_fct(active_logits, active_labels)\r\n\r\n[CrossEntropyLoss](https://pytorch.org/docs/stable/generated/torch.nn.CrossEntropyLoss.html#crossentropyloss) has default reduction `mean`, so basically we compute the averaged loss over the tokens (label !=-100) on each single batch. Then we accumulate it `gradient_accumulation_steps`, and finally average again by dividing `gradient_accumulation_steps`, see [here](https://github.com/huggingface/transformers/blob/master/src/transformers/trainer.py#L1041).\r\n\r\nHowever, this is no longer the same as the `per example losses over the global batch divided the number of tokens on that global batch`. In practice, it may not harm the training, but theoretically, it is not what exactly the gradient accumulation mean to be.\r\n\r\nBut this should be discussed in another thread, not here.", "> pening the door to have task-specific components in the main tf_trainer file will make it unreadable in a few months, when every user will have added their own, and then users that rely on a custom Trainers won't use that class anymore, because they won't understand it.\r\n\r\nThe current draft code requires a task to compute the number of `instances` in each `example`. For token level tasks, it will be the number of tokens with labels != -100. Fore sentence level, in general, it will just be the number of sentences. The tf trainer will just use this information for further calculation, but it doesn't need to know what the tasks are.", "@jplu , this is not a finalized version. However, I want to have some feedback from you about if this approach is OK. Thanks.", "For the sake of simplicity, I suspect this would be best left as a user-space extension/custom loss (via a subclass of TFTrainer for instance)\r\n\r\nIf needed we can add/improve extension points in TFTrainer to make it easier.\r\n\r\nWhat do you think?", "@julien-c \r\n\r\nMy thoughts are:\r\n\r\n- Hugging Face's ` transformers` has `run_ner_tf.py` which is assumed to give the correct training/evaluation losses\r\n- (maybe I misunderstand the roles of scripts in examples dir?)\r\n- If they are meant to be correct, and let's say the right way for token level tasks is to count the number of tokens (not ignored), then the current script won't give the correct results. In this case, the majority of users who use it won't know they are supposed to subclass TFTrainer class\r\n- Even they are aware of the necessity of counting tokens rather than examples, in TF, it is not easy to make it right - if it has to work correctly in a distributed strategy and/or gradient accumulation. It is not just about modifying the loss calculation - the token counting has to be in a global batch, not a batch already distributed to a single replica. Then in a single replica, the per example losses (on that small batch) has to be divided by the number of tokens computed in the big batch before being distributed.\r\n- If the scripts in examples dir have only the purpose of demonstration of the library's usage - leaving users to customize trainers is fine, and adding/improving extension points seems good to me (although I don't know what it looks like for now). Maybe it is also good to have some warning and a brief tutorial to let users know how to do things.\r\n\r\nPlease let me know the team's decision about this issue (how to continue or if to close it). Thanks.", "@chiapas Now it looks great! I really like it. Did you try it in context of single and multi-replica?\r\n\r\n@julien-c @sgugger I think there is a misunderstanding here, this PR is not to add a new feature or any refactoring, this is a **bugfix** in the loss computation, that means that anybody that are currently using the current version of the trainer get a wrong loss value for token classifications class. And as @chiapas said the way it is computed in the PyTorch trainer might need a fix as well. There are only few changes and do not impact the readability for TF users.", "> \r\n> \r\n> @chiapas Now it looks great! I really like it. Did you try it in context of single and multi-replica?\r\n\r\n@jplu Thanks. I haven't tried testing yet. I preferred to have your feedback about this new way of fix before finalizing it and testing. By the way, for multi-replica, I can only run on Kaggle or Google colab. I will let you know once the test is done.\r\n\r\n", "This is ok no worries, with TPU is fine as well 👍 ", "@chiapas As far as I can say this PR should also fix the issue #6969 right? As we compute the total number of example at every step.", "> \r\n> \r\n> @chiapas As far as I can say this PR should also fix the issue #6969 right? As we compute the total number of example at every step.\r\n\r\nYes. However, after I opened up that issue and work on this PR, I found that we have\r\n\r\n ds = (\r\n self.train_dataset.repeat()\r\n .shuffle(self.num_train_examples, seed=self.args.seed)\r\n .batch(self.total_train_batch_size, drop_remainder=self.args.dataloader_drop_last)\r\n .prefetch(tf.data.experimental.AUTOTUNE)\r\n )\r\n\r\nSince the dataset is repeated first, it has no ending, and the drop_remainder has no actual effect (other than set the batch shpae from None to a fixed number along batch dimension), so issue #6969 will never occur. However, in this case, `args.dataloader_drop_last` is somehow confusing.", "Indeed, having the `repeat` has the advantage to avoid the potential last partial batch in each epoch, so users don't need to think about scaling the gradients based on the actual batch size and makes the `dataloader_drop_last` useless.", "> \r\n> \r\n> Indeed, having the `repeat` has the advantage to avoid the potential last partial batch in each epoch, so users don't need to think about scaling the gradients based on the actual batch size and makes the `dataloader_drop_last` useless.\r\n\r\nYes, I am ok with this. BTW, `dataloader_drop_last` might still have an effect - if `True`, the batch dimension will be fixed in the compiled graph. If it is set to `False`, even we repeat the dataset first, the batch dimension will still be `None` . In this case, while working with TPU with the way we do gradient accumulation int trainer_tf.py, we will get an error message that complains TPU can't handle slice or shape - I don't remember the exact message, but I can reproduce one quickly.", "The default value of `drop_reminder` is False, which result in an unknown batch size because the last batch may not be full, this is exactly why `drop_reminder` on TPU has to be set to `True` if no repeat is applied otherwise we can let it to False.\r\n\r\nSmall example:\r\n\r\n```python\r\n>>> dataset = tf.data.Dataset.range(100)\r\n>>> dataset.batch(4)\r\n<BatchDataset shapes: (None,), types: tf.int64>\r\n>>> dataset = tf.data.Dataset.range(100)\r\n>>> dataset.batch(4, drop_remainder=True)\r\n<BatchDataset shapes: (4,), types: tf.int64>", "> \r\n> \r\n> The default value of `drop_reminder` is False, which result in an unknown batch size because the last batch may not be full, this is exactly why `drop_reminder` on TPU has to be set to `True` if no repeat is applied otherwise we can let it to False.\r\n\r\n otherwise we can let it to False\r\n\r\nWhat I am saying is: We can't let it to `False` even repeat is applied in `trainer_tf.py`. In general, if `repeat` is used, we don't have to drop. But due to the gradient accumulation implementation, if TPU is used and, if we set `drop_remainde=False`, even the `repeat` is applied, we will still get\r\n\r\n\t<PrefetchDataset shapes: ((None, 512, 512, 3), (None,)), types: (tf.float32, tf.int32)>\r\n\r\n\r\n\tNotFoundError: 3 root error(s) found.\r\n\t (0) Not found: {{function_node __inference_train_step_1_epoch_192920}} No proto found for key <<NO PROGRAM AS COMPILATION FAILED>>\r\n\t\t [[{{node TPUVariableReshard/reshard/_16819633198340046116/_31}}]]\r\n\t (1) Not found: {{function_node __inference_train_step_1_epoch_192920}} No proto found for key <<NO PROGRAM AS COMPILATION FAILED>>\r\n\t\t [[{{node TPUVariableReshard/reshard/_17949385379616849075/_19}}]]\r\n\t (2) Unimplemented: {{function_node __inference_train_step_1_epoch_192920}} Compilation failure: Dynamic input dimension to reshape that is both splitted and combined is not supported: output: f32[0,512,512,3], input: f32[<=0,512,512,3], input_dim: 0\r\n\t\t [[{{node strided_slice_2}}]]\r\n\t\t [[while/body/_1/while]]\r\n\t\tTPU compilation failed\r\n\t\t [[tpu_compile_succeeded_assert/_17625000101377989734/_5]]\r\n\t0 successful operations.\r\n\t6 derived errors ignored.\r\n\r\nYou can test it if you want on\r\n\r\nhttps://www.kaggle.com/yihdarshieh/tpu-gradient-accumulation?scriptVersionId=41324202\r\n\r\nby changing one line in \r\n\r\n def get_training_dataset(batch_size):", "Hummm nice catch, I haven't tested this case with gradient accumulation, thanks!", "@chiapas for me it looks ok, do you want to add anything else? If no can you switch the PR to open, in order to be able for us to merge it.", "> \r\n> \r\n> @chiapas for me it looks ok, do you want to add anything else? If no can you switch the PR to open, in order to be able for us to merge it.\r\n\r\n@jplu , I haven't done any test yet - I pushed the code immediately after writting it (in order to have your feedback), so from my side, I feel more comfortable to check a few things before merged to master. Unless you have done some testings and are eager to merge, maybe wait a bit please? I think tomorrow at some point would be ready. ", "I have tested it on single and 4 GPU training with the example script for NER and was ok, but take the time you need the more we can test the better. I will be happy to know how it works on TPU as well, even through a colab.", "@jplu I had to fix bugs, and now I have a working version - I checked some intermediate values to make sure the calculation of number of `instances` is correct and sent to replica(s).\r\n\r\nDue to the bugs I found - I think there might be a chance that your test yesterday didn't use my code - probably forgot to uninstall the original `transformers` and reinstall `transformers` that is based on my version? I have this doubt because those bugs would throw errors, and the training wouldn't be successful. Sorry about this. If possible, could you test it again with the latest version on multiple GPU env.? Thanks.\r\n\r\nOtherwise, I tested with CPU and 1 GPU on colab. It works fine. You can check here\r\n\r\n https://colab.research.google.com/drive/148whpTObbF53qU_ec0bVkyVWUsLlwppn?usp=sharing\r\n\r\nHowever, using TPU with tf 2.2 or 2.3, I had different errors, for which I think irrelevant to this PR code See below for the error messages. We can probably open a bug report and fix it later.\r\n\r\nAlso, the CI is not green because of some problems from the master branch. From my side, the code is ready to be merged (if you can test on multiple GPU again would be better). Thanks.\r\n\r\n Traceback (most recent call last):\r\n File \"run_tf_ner.py\", line 299, in <module>\r\n main()\r\n File \"run_tf_ner.py\", line 128, in main\r\n training_args.n_replicas,\r\n File \"/content/transformers/examples/token-classification/transformers/src/transformers/file_utils.py\", line 926, in wrapper\r\n return func(*args, **kwargs)\r\n File \"/content/transformers/examples/token-classification/transformers/src/transformers/training_args_tf.py\", line 161, in n_replicas\r\n return self._setup_strategy.num_replicas_in_sync\r\n File \"/content/transformers/examples/token-classification/transformers/src/transformers/file_utils.py\", line 904, in __get__\r\n cached = self.fget(obj)\r\n File \"/content/transformers/examples/token-classification/transformers/src/transformers/file_utils.py\", line 926, in wrapper\r\n return func(*args, **kwargs)\r\n File \"/content/transformers/examples/token-classification/transformers/src/transformers/training_args_tf.py\", line 132, in _setup_strategy\r\n tf.tpu.experimental.initialize_tpu_system(tpu)\r\n File \"/usr/local/lib/python3.6/dist-packages/tensorflow/python/tpu/tpu_strategy_util.py\", line 103, in initialize_tpu_system\r\n serialized_topology = output.numpy()\r\n File \"/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/ops.py\", line 961, in numpy\r\n maybe_arr = self._numpy() # pylint: disable=protected-access\r\n File \"/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/ops.py\", line 929, in _numpy\r\n six.raise_from(core._status_to_exception(e.code, e.message), None)\r\n File \"<string>\", line 3, in raise_from\r\n tensorflow.python.framework.errors_impl.InvalidArgumentError: NodeDef expected inputs 'string' do not match 0 inputs specified; Op<name=_Send; signature=tensor:T -> ; attr=T:type; attr=tensor_name:string; attr=send_device:string ..........\r\n\r\nand TPU with tf 2.3 gives different error\r\n\r\n Traceback (most recent call last):\r\n File \"run_tf_ner.py\", line 299, in <module>\r\n main()\r\n File \"run_tf_ner.py\", line 230, in main\r\n trainer.train()\r\n File \"/content/transformers/src/transformers/trainer_tf.py\", line 474, in train\r\n train_ds = self.get_train_tfdataset()\r\n File \"/content/transformers/src/transformers/trainer_tf.py\", line 137, in get_train_tfdataset\r\n self.num_train_examples = tf.data.experimental.cardinality(self.train_dataset).numpy()\r\n File \"/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/ops.py\", line 1063, in numpy\r\n maybe_arr = self._numpy() # pylint: disable=protected-access\r\n File \"/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/ops.py\", line 1031, in _numpy\r\n six.raise_from(core._status_to_exception(e.code, e.message), None) # pylint: disable=protected-access\r\n File \"<string>\", line 3, in raise_from\r\n tensorflow.python.framework.errors_impl.UnimplementedError: File system scheme '[local]' not implemented (file: 'runs/Sep11_12-45-28_1d2f5ee8ee35')\r\n Encountered when executing an operation using EagerExecutor. This error cancels all future operations and poisons their output tensors.\r\n Error in atexit._run_exitfuncs:\r\n Traceback (most recent call last):\r\n File \"/usr/local/lib/python3.6/dist-packages/tensorflow/python/distribute/tpu_strategy.py\", line 540, in async_wait\r\n context.async_wait()\r\n File \"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/context.py\", line 2319, in async_wait\r\n context().sync_executors()\r\n File \"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/context.py\", line 658, in sync_executors\r\n pywrap_tfe.TFE_ContextSyncExecutors(self._context_handle)\r\n tensorflow.python.framework.errors_impl.UnimplementedError: File system scheme '[local]' not implemented (file: 'runs/Sep11_12-45-28_1d2f5ee8ee35')\r\n Encountered when executing an operation using EagerExecutor. This error cancels all future operations and poisons their output tensors.\r\n 2020-09-11 12:46:28.864320: W ./tensorflow/core/distributed_runtime/eager/destroy_tensor_handle_node.h:57] Ignoring an error encountered when deleting remote tensors handles: Invalid argument: Unable to find the relevant tensor remote_handle: Op ID: 9416, Output num: 0\r\n Additional GRPC error information from remote target /job:worker/replica:0/task:0:\r\n :{\"created\":\"@1599828388.860894650\",\"description\":\"Error received from peer ipv4:10.42.193.26:8470\",\"file\":\"external/com_github_grpc_grpc/src/core/lib/surface/call.cc\",\"file_line\":1056,\"grpc_message\":\"Unable to find the relevant tensor remote_handle: Op ID: 9416, Output num: 0\",\"grpc_status\":3}", ">Due to the bugs I found - I think there might be a chance that your test yesterday didn't use my code - probably forgot to uninstall the original transformers and reinstall transformers that is based on my version? I have this doubt because those bugs would throw errors, and the training wouldn't be successful. Sorry about this. If possible, could you test it again with the latest version on multiple GPU env.? Thanks.\r\n\r\nAh! Might be possible I forgot to install your version. I will re-test it, to be sure.\r\n\r\n> However, using TPU with tf 2.2 or 2.3, I had different errors, for which I think irrelevant to this PR code See below for the error messages. We can probably open a bug report and fix it later.\r\n\r\nThe second error you get means that you cannot load data from localhost, the files have to be hosted on a GCS to make it works.\r\n\r\n> Also, the CI is not green because of some problems from the master branch. \r\n\r\nCan you try to rebase on the current master and see if the CI error still occurs?\r\n\r\n> From my side, the code is ready to be merged (if you can test on multiple GPU again would be better). Thanks.\r\n\r\nI will test that ASAP today and will let you know.", "I have been able to run a NER task over 4 GPUs. Without gradient accumulation:\r\n\r\n```\r\n***** Running training *****\r\n Num examples = 24000\r\n Num Epochs = 3\r\n Instantaneous batch size per device = 32\r\n Total train batch size (w. parallel, distributed & accumulation) = 128\r\n Gradient Accumulation steps = 1\r\n Steps per epoch = 188\r\n Total optimization steps = 564\r\n{'loss': 18.6387, 'learning_rate': 4.9113474e-05, 'epoch': 0.05851063829787234, 'step': 10}\r\n{'loss': 14.222743, 'learning_rate': 4.822695e-05, 'epoch': 0.11170212765957446, 'step': 20}\r\n{'loss': 12.079448, 'learning_rate': 4.734042e-05, 'epoch': 0.16489361702127658, 'step': 30}\r\n{'loss': 10.455857, 'learning_rate': 4.6453897e-05, 'epoch': 0.21808510638297873, 'step': 40}\r\n{'loss': 9.295527, 'learning_rate': 4.5567373e-05, 'epoch': 0.2712765957446808, 'step': 50}\r\n{'loss': 8.374706, 'learning_rate': 4.468085e-05, 'epoch': 0.324468085106383, 'step': 60}\r\n{'loss': 7.674021, 'learning_rate': 4.379432e-05, 'epoch': 0.3776595744680851, 'step': 70}\r\n{'loss': 7.106626, 'learning_rate': 4.29078e-05, 'epoch': 0.4308510638297872, 'step': 80}\r\n{'loss': 6.6170573, 'learning_rate': 4.202128e-05, 'epoch': 0.48404255319148937, 'step': 90}\r\n{'loss': 6.226298, 'learning_rate': 4.113475e-05, 'epoch': 0.5372340425531915, 'step': 100}\r\n{'loss': 5.859087, 'learning_rate': 4.0248226e-05, 'epoch': 0.5904255319148937, 'step': 110}\r\n{'loss': 5.560567, 'learning_rate': 3.93617e-05, 'epoch': 0.6436170212765957, 'step': 120}\r\n{'loss': 5.2810636, 'learning_rate': 3.8475173e-05, 'epoch': 0.6968085106382979, 'step': 130}\r\n{'loss': 5.040142, 'learning_rate': 3.758865e-05, 'epoch': 0.75, 'step': 140}\r\n{'loss': 4.830164, 'learning_rate': 3.6702124e-05, 'epoch': 0.8031914893617021, 'step': 150}\r\n{'loss': 4.6353145, 'learning_rate': 3.5815603e-05, 'epoch': 0.8563829787234043, 'step': 160}\r\n{'loss': 4.446635, 'learning_rate': 3.492908e-05, 'epoch': 0.9095744680851063, 'step': 170}\r\n{'loss': 4.300565, 'learning_rate': 3.4042554e-05, 'epoch': 0.9627659574468085, 'step': 180}\r\n2020-09-11 23:08:59.187632: I tensorflow/core/kernels/data/shuffle_dataset_op.cc:172] Filling up shuffle buffer (this may take a while): 23119 of 24000\r\n2020-09-11 23:08:59.564647: I tensorflow/core/kernels/data/shuffle_dataset_op.cc:221] Shuffle buffer filled.\r\n{'loss': 1.8294506, 'learning_rate': 3.3156026e-05, 'epoch': 1.0106382978723405, 'step': 190}\r\n{'loss': 1.576384, 'learning_rate': 3.22695e-05, 'epoch': 1.0638297872340425, 'step': 200}\r\n{'loss': 1.4572238, 'learning_rate': 3.1382977e-05, 'epoch': 1.1170212765957448, 'step': 210}\r\n{'loss': 1.4322406, 'learning_rate': 3.0496454e-05, 'epoch': 1.1702127659574468, 'step': 220}\r\n{'loss': 1.3880556, 'learning_rate': 2.960993e-05, 'epoch': 1.2234042553191489, 'step': 230}\r\n{'loss': 1.3613675, 'learning_rate': 2.8723403e-05, 'epoch': 1.2765957446808511, 'step': 240}\r\n{'loss': 1.3514798, 'learning_rate': 2.7836879e-05, 'epoch': 1.3297872340425532, 'step': 250}\r\n{'loss': 1.3266419, 'learning_rate': 2.6950353e-05, 'epoch': 1.3829787234042552, 'step': 260}\r\n{'loss': 1.3012911, 'learning_rate': 2.606383e-05, 'epoch': 1.4361702127659575, 'step': 270}\r\n{'loss': 1.2993147, 'learning_rate': 2.5177305e-05, 'epoch': 1.4893617021276595, 'step': 280}\r\n{'loss': 1.2913059, 'learning_rate': 2.4290777e-05, 'epoch': 1.5425531914893615, 'step': 290}\r\n{'loss': 1.2822802, 'learning_rate': 2.3404255e-05, 'epoch': 1.5957446808510638, 'step': 300}\r\n{'loss': 1.2839314, 'learning_rate': 2.2517732e-05, 'epoch': 1.648936170212766, 'step': 310}\r\n{'loss': 1.2641081, 'learning_rate': 2.1631204e-05, 'epoch': 1.702127659574468, 'step': 320}\r\n{'loss': 1.2524884, 'learning_rate': 2.0744681e-05, 'epoch': 1.7553191489361701, 'step': 330}\r\n{'loss': 1.2450953, 'learning_rate': 1.9858155e-05, 'epoch': 1.8085106382978724, 'step': 340}\r\n{'loss': 1.2448001, 'learning_rate': 1.897163e-05, 'epoch': 1.8617021276595744, 'step': 350}\r\n{'loss': 1.2407304, 'learning_rate': 1.8085108e-05, 'epoch': 1.9148936170212765, 'step': 360}\r\n{'loss': 1.2282307, 'learning_rate': 1.719858e-05, 'epoch': 1.9680851063829787, 'step': 370}\r\n2020-09-11 23:14:15.677977: I tensorflow/core/kernels/data/shuffle_dataset_op.cc:172] Filling up shuffle buffer (this may take a while): 23228 of 24000\r\n2020-09-11 23:14:16.010139: I tensorflow/core/kernels/data/shuffle_dataset_op.cc:221] Shuffle buffer filled.\r\n{'loss': 0.85539556, 'learning_rate': 1.6312057e-05, 'epoch': 2.021276595744681, 'step': 380}\r\n{'loss': 0.94433624, 'learning_rate': 1.542553e-05, 'epoch': 2.074468085106383, 'step': 390}\r\n{'loss': 0.93659353, 'learning_rate': 1.4539006e-05, 'epoch': 2.127659574468085, 'step': 400}\r\n{'loss': 0.8978215, 'learning_rate': 1.365248e-05, 'epoch': 2.1808510638297873, 'step': 410}\r\n{'loss': 0.9126247, 'learning_rate': 1.27659605e-05, 'epoch': 2.2340425531914896, 'step': 420}\r\n{'loss': 0.9438525, 'learning_rate': 1.1879432e-05, 'epoch': 2.2872340425531914, 'step': 430}\r\n{'loss': 0.9655043, 'learning_rate': 1.0992908e-05, 'epoch': 2.3404255319148937, 'step': 440}\r\n{'loss': 0.9694119, 'learning_rate': 1.0106382e-05, 'epoch': 2.393617021276596, 'step': 450}\r\n{'loss': 0.95613927, 'learning_rate': 9.219858e-06, 'epoch': 2.4468085106382977, 'step': 460}\r\n{'loss': 0.9483009, 'learning_rate': 8.333333e-06, 'epoch': 2.5, 'step': 470}\r\n{'loss': 0.93453395, 'learning_rate': 7.4468076e-06, 'epoch': 2.5531914893617023, 'step': 480}\r\n{'loss': 0.92573655, 'learning_rate': 6.5602835e-06, 'epoch': 2.6063829787234045, 'step': 490}\r\n{'loss': 0.9156919, 'learning_rate': 5.67376e-06, 'epoch': 2.6595744680851063, 'step': 500}\r\n{'loss': 0.9098517, 'learning_rate': 4.787233e-06, 'epoch': 2.7127659574468086, 'step': 510}\r\n{'loss': 0.9110744, 'learning_rate': 3.9007095e-06, 'epoch': 2.7659574468085104, 'step': 520}\r\n{'loss': 0.89820355, 'learning_rate': 3.014183e-06, 'epoch': 2.8191489361702127, 'step': 530}\r\n{'loss': 0.8982555, 'learning_rate': 2.1276592e-06, 'epoch': 2.872340425531915, 'step': 540}\r\n{'loss': 0.8997822, 'learning_rate': 1.2411356e-06, 'epoch': 2.925531914893617, 'step': 550}\r\n{'loss': 0.89611685, 'learning_rate': 3.5460886e-07, 'epoch': 2.978723404255319, 'step': 560}\r\nTraining took: 0:17:32.846350\r\nSaving model in /home/jplu/model\r\nConfiguration saved in /home/jplu/model/config.json\r\nModel weights saved in /home/jplu/model/tf_model.h5\r\n***** Running Evaluation *****\r\n Num examples = 2200\r\n Batch size = 32\r\n{'eval_loss': 1.4825085626132246, 'eval_precision': 0.8298914945747288, 'eval_recall': 0.8713708195516354, 'eval_f1': 0.8501254930082466, 'epoch': 3.0, 'step': 564}\r\n```\r\n\r\nWith gradient accumulation:\r\n```\r\n***** Running training *****\r\n Num examples = 24000\r\n Num Epochs = 3\r\n Instantaneous batch size per device = 32\r\n Total train batch size (w. parallel, distributed & accumulation) = 256\r\n Gradient Accumulation steps = 2\r\n Steps per epoch = 94\r\n Total optimization steps = 282\r\n{'loss': 17.029345, 'learning_rate': 4.822695e-05, 'epoch': 0.11702127659574468, 'step': 10}\r\n{'loss': 13.583568, 'learning_rate': 4.6453897e-05, 'epoch': 0.22340425531914893, 'step': 20}\r\n{'loss': 11.413903, 'learning_rate': 4.468085e-05, 'epoch': 0.32978723404255317, 'step': 30}\r\n{'loss': 9.977048, 'learning_rate': 4.29078e-05, 'epoch': 0.43617021276595747, 'step': 40}\r\n{'loss': 8.904137, 'learning_rate': 4.113475e-05, 'epoch': 0.5425531914893617, 'step': 50}\r\n{'loss': 8.056796, 'learning_rate': 3.93617e-05, 'epoch': 0.648936170212766, 'step': 60}\r\n{'loss': 7.339738, 'learning_rate': 3.758865e-05, 'epoch': 0.7553191489361702, 'step': 70}\r\n{'loss': 6.7678766, 'learning_rate': 3.5815603e-05, 'epoch': 0.8617021276595744, 'step': 80}\r\n{'loss': 6.2809086, 'learning_rate': 3.4042554e-05, 'epoch': 0.9680851063829787, 'step': 90}\r\n2020-09-11 23:36:37.142279: I tensorflow/core/kernels/data/shuffle_dataset_op.cc:172] Filling up shuffle buffer (this may take a while): 23596 of 24000\r\n2020-09-11 23:36:37.311979: I tensorflow/core/kernels/data/shuffle_dataset_op.cc:221] Shuffle buffer filled.\r\n{'loss': 2.2556372, 'learning_rate': 3.22695e-05, 'epoch': 1.0638297872340425, 'step': 100}\r\n{'loss': 2.0573573, 'learning_rate': 3.0496454e-05, 'epoch': 1.1702127659574468, 'step': 110}\r\n{'loss': 1.9544038, 'learning_rate': 2.8723403e-05, 'epoch': 1.2765957446808511, 'step': 120}\r\n{'loss': 1.8848253, 'learning_rate': 2.6950353e-05, 'epoch': 1.3829787234042552, 'step': 130}\r\n{'loss': 1.837054, 'learning_rate': 2.5177305e-05, 'epoch': 1.4893617021276595, 'step': 140}\r\n{'loss': 1.7885295, 'learning_rate': 2.3404255e-05, 'epoch': 1.5957446808510638, 'step': 150}\r\n{'loss': 1.7535, 'learning_rate': 2.1631204e-05, 'epoch': 1.702127659574468, 'step': 160}\r\n{'loss': 1.7068337, 'learning_rate': 1.9858155e-05, 'epoch': 1.8085106382978724, 'step': 170}\r\n{'loss': 1.6874169, 'learning_rate': 1.8085108e-05, 'epoch': 1.9148936170212765, 'step': 180}\r\n2020-09-11 23:41:05.123380: I tensorflow/core/kernels/data/shuffle_dataset_op.cc:172] Filling up shuffle buffer (this may take a while): 23596 of 24000\r\n2020-09-11 23:36:05.332080: I tensorflow/core/kernels/data/shuffle_dataset_op.cc:221] Shuffle buffer filled.\r\n{'loss': 1.189175, 'learning_rate': 1.6312057e-05, 'epoch': 2.021276595744681, 'step': 190}\r\n{'loss': 1.2820004, 'learning_rate': 1.4539006e-05, 'epoch': 2.127659574468085, 'step': 200}\r\n{'loss': 1.2558761, 'learning_rate': 1.27659605e-05, 'epoch': 2.2340425531914896, 'step': 210}\r\n{'loss': 1.2865627, 'learning_rate': 1.0992908e-05, 'epoch': 2.3404255319148937, 'step': 220}\r\n{'loss': 1.2853887, 'learning_rate': 9.219858e-06, 'epoch': 2.4468085106382977, 'step': 230}\r\n{'loss': 1.2650005, 'learning_rate': 7.4468076e-06, 'epoch': 2.5531914893617023, 'step': 240}\r\n{'loss': 1.2414478, 'learning_rate': 5.67376e-06, 'epoch': 2.6595744680851063, 'step': 250}\r\n{'loss': 1.243986, 'learning_rate': 3.9007095e-06, 'epoch': 2.7659574468085104, 'step': 260}\r\n{'loss': 1.2295542, 'learning_rate': 2.1276592e-06, 'epoch': 2.872340425531915, 'step': 270}\r\n{'loss': 1.2296557, 'learning_rate': 3.5460886e-07, 'epoch': 2.978723404255319, 'step': 280}\r\nTraining took: 0:19:10.070496\r\nSaving model in /home/jplu/model\r\nConfiguration saved in /home/jplu/model/config.json\r\nModel weights saved in /home/jplu/model/tf_model.h5\r\n***** Running Evaluation *****\r\n Num examples = 2200\r\n Batch size = 32\r\n{'eval_loss': 1.5663623533387114, 'eval_precision': 0.8107354478912513, 'eval_recall': 0.8548327820654171, 'eval_f1': 0.8322003577817532, 'epoch': 3.0, 'step': 282}\r\n```\r\nLooks ok to me. I have also tested for text classification and question answering without any error.\r\n", "> \r\n> \r\n> I have been able to run a NER task over 4 GPUs. Without gradient accumulation:\r\n> \r\n> ```\r\n> ***** Running training *****\r\n> Num examples = 24000\r\n> Num Epochs = 3\r\n> Instantaneous batch size per device = 32\r\n> Total train batch size (w. parallel, distributed & accumulation) = 128\r\n> Gradient Accumulation steps = 1\r\n> Steps per epoch = 188\r\n> Total optimization steps = 564\r\n> {'loss': 18.6387, 'learning_rate': 4.9113474e-05, 'epoch': 0.05851063829787234, 'step': 10}\r\n> {'loss': 14.222743, 'learning_rate': 4.822695e-05, 'epoch': 0.11170212765957446, 'step': 20}\r\n> {'loss': 12.079448, 'learning_rate': 4.734042e-05, 'epoch': 0.16489361702127658, 'step': 30}\r\n> {'loss': 10.455857, 'learning_rate': 4.6453897e-05, 'epoch': 0.21808510638297873, 'step': 40}\r\n> {'loss': 9.295527, 'learning_rate': 4.5567373e-05, 'epoch': 0.2712765957446808, 'step': 50}\r\n> {'loss': 8.374706, 'learning_rate': 4.468085e-05, 'epoch': 0.324468085106383, 'step': 60}\r\n> {'loss': 7.674021, 'learning_rate': 4.379432e-05, 'epoch': 0.3776595744680851, 'step': 70}\r\n> {'loss': 7.106626, 'learning_rate': 4.29078e-05, 'epoch': 0.4308510638297872, 'step': 80}\r\n> {'loss': 6.6170573, 'learning_rate': 4.202128e-05, 'epoch': 0.48404255319148937, 'step': 90}\r\n> {'loss': 6.226298, 'learning_rate': 4.113475e-05, 'epoch': 0.5372340425531915, 'step': 100}\r\n> {'loss': 5.859087, 'learning_rate': 4.0248226e-05, 'epoch': 0.5904255319148937, 'step': 110}\r\n> {'loss': 5.560567, 'learning_rate': 3.93617e-05, 'epoch': 0.6436170212765957, 'step': 120}\r\n> {'loss': 5.2810636, 'learning_rate': 3.8475173e-05, 'epoch': 0.6968085106382979, 'step': 130}\r\n> {'loss': 5.040142, 'learning_rate': 3.758865e-05, 'epoch': 0.75, 'step': 140}\r\n> {'loss': 4.830164, 'learning_rate': 3.6702124e-05, 'epoch': 0.8031914893617021, 'step': 150}\r\n> {'loss': 4.6353145, 'learning_rate': 3.5815603e-05, 'epoch': 0.8563829787234043, 'step': 160}\r\n> {'loss': 4.446635, 'learning_rate': 3.492908e-05, 'epoch': 0.9095744680851063, 'step': 170}\r\n> {'loss': 4.300565, 'learning_rate': 3.4042554e-05, 'epoch': 0.9627659574468085, 'step': 180}\r\n> 2020-09-11 23:08:59.187632: I tensorflow/core/kernels/data/shuffle_dataset_op.cc:172] Filling up shuffle buffer (this may take a while): 23119 of 24000\r\n> 2020-09-11 23:08:59.564647: I tensorflow/core/kernels/data/shuffle_dataset_op.cc:221] Shuffle buffer filled.\r\n> {'loss': 1.8294506, 'learning_rate': 3.3156026e-05, 'epoch': 1.0106382978723405, 'step': 190}\r\n> {'loss': 1.576384, 'learning_rate': 3.22695e-05, 'epoch': 1.0638297872340425, 'step': 200}\r\n> {'loss': 1.4572238, 'learning_rate': 3.1382977e-05, 'epoch': 1.1170212765957448, 'step': 210}\r\n> {'loss': 1.4322406, 'learning_rate': 3.0496454e-05, 'epoch': 1.1702127659574468, 'step': 220}\r\n> {'loss': 1.3880556, 'learning_rate': 2.960993e-05, 'epoch': 1.2234042553191489, 'step': 230}\r\n> {'loss': 1.3613675, 'learning_rate': 2.8723403e-05, 'epoch': 1.2765957446808511, 'step': 240}\r\n> {'loss': 1.3514798, 'learning_rate': 2.7836879e-05, 'epoch': 1.3297872340425532, 'step': 250}\r\n> {'loss': 1.3266419, 'learning_rate': 2.6950353e-05, 'epoch': 1.3829787234042552, 'step': 260}\r\n> {'loss': 1.3012911, 'learning_rate': 2.606383e-05, 'epoch': 1.4361702127659575, 'step': 270}\r\n> {'loss': 1.2993147, 'learning_rate': 2.5177305e-05, 'epoch': 1.4893617021276595, 'step': 280}\r\n> {'loss': 1.2913059, 'learning_rate': 2.4290777e-05, 'epoch': 1.5425531914893615, 'step': 290}\r\n> {'loss': 1.2822802, 'learning_rate': 2.3404255e-05, 'epoch': 1.5957446808510638, 'step': 300}\r\n> {'loss': 1.2839314, 'learning_rate': 2.2517732e-05, 'epoch': 1.648936170212766, 'step': 310}\r\n> {'loss': 1.2641081, 'learning_rate': 2.1631204e-05, 'epoch': 1.702127659574468, 'step': 320}\r\n> {'loss': 1.2524884, 'learning_rate': 2.0744681e-05, 'epoch': 1.7553191489361701, 'step': 330}\r\n> {'loss': 1.2450953, 'learning_rate': 1.9858155e-05, 'epoch': 1.8085106382978724, 'step': 340}\r\n> {'loss': 1.2448001, 'learning_rate': 1.897163e-05, 'epoch': 1.8617021276595744, 'step': 350}\r\n> {'loss': 1.2407304, 'learning_rate': 1.8085108e-05, 'epoch': 1.9148936170212765, 'step': 360}\r\n> {'loss': 1.2282307, 'learning_rate': 1.719858e-05, 'epoch': 1.9680851063829787, 'step': 370}\r\n> 2020-09-11 23:14:15.677977: I tensorflow/core/kernels/data/shuffle_dataset_op.cc:172] Filling up shuffle buffer (this may take a while): 23228 of 24000\r\n> 2020-09-11 23:14:16.010139: I tensorflow/core/kernels/data/shuffle_dataset_op.cc:221] Shuffle buffer filled.\r\n> {'loss': 0.85539556, 'learning_rate': 1.6312057e-05, 'epoch': 2.021276595744681, 'step': 380}\r\n> {'loss': 0.94433624, 'learning_rate': 1.542553e-05, 'epoch': 2.074468085106383, 'step': 390}\r\n> {'loss': 0.93659353, 'learning_rate': 1.4539006e-05, 'epoch': 2.127659574468085, 'step': 400}\r\n> {'loss': 0.8978215, 'learning_rate': 1.365248e-05, 'epoch': 2.1808510638297873, 'step': 410}\r\n> {'loss': 0.9126247, 'learning_rate': 1.27659605e-05, 'epoch': 2.2340425531914896, 'step': 420}\r\n> {'loss': 0.9438525, 'learning_rate': 1.1879432e-05, 'epoch': 2.2872340425531914, 'step': 430}\r\n> {'loss': 0.9655043, 'learning_rate': 1.0992908e-05, 'epoch': 2.3404255319148937, 'step': 440}\r\n> {'loss': 0.9694119, 'learning_rate': 1.0106382e-05, 'epoch': 2.393617021276596, 'step': 450}\r\n> {'loss': 0.95613927, 'learning_rate': 9.219858e-06, 'epoch': 2.4468085106382977, 'step': 460}\r\n> {'loss': 0.9483009, 'learning_rate': 8.333333e-06, 'epoch': 2.5, 'step': 470}\r\n> {'loss': 0.93453395, 'learning_rate': 7.4468076e-06, 'epoch': 2.5531914893617023, 'step': 480}\r\n> {'loss': 0.92573655, 'learning_rate': 6.5602835e-06, 'epoch': 2.6063829787234045, 'step': 490}\r\n> {'loss': 0.9156919, 'learning_rate': 5.67376e-06, 'epoch': 2.6595744680851063, 'step': 500}\r\n> {'loss': 0.9098517, 'learning_rate': 4.787233e-06, 'epoch': 2.7127659574468086, 'step': 510}\r\n> {'loss': 0.9110744, 'learning_rate': 3.9007095e-06, 'epoch': 2.7659574468085104, 'step': 520}\r\n> {'loss': 0.89820355, 'learning_rate': 3.014183e-06, 'epoch': 2.8191489361702127, 'step': 530}\r\n> {'loss': 0.8982555, 'learning_rate': 2.1276592e-06, 'epoch': 2.872340425531915, 'step': 540}\r\n> {'loss': 0.8997822, 'learning_rate': 1.2411356e-06, 'epoch': 2.925531914893617, 'step': 550}\r\n> {'loss': 0.89611685, 'learning_rate': 3.5460886e-07, 'epoch': 2.978723404255319, 'step': 560}\r\n> Training took: 0:17:32.846350\r\n> Saving model in /home/jplu/model\r\n> Configuration saved in /home/jplu/model/config.json\r\n> Model weights saved in /home/jplu/model/tf_model.h5\r\n> ***** Running Evaluation *****\r\n> Num examples = 2200\r\n> Batch size = 32\r\n> {'eval_loss': 1.4825085626132246, 'eval_precision': 0.8298914945747288, 'eval_recall': 0.8713708195516354, 'eval_f1': 0.8501254930082466, 'epoch': 3.0, 'step': 564}\r\n> ```\r\n> \r\n> With gradient accumulation:\r\n> \r\n> ```\r\n> ***** Running training *****\r\n> Num examples = 24000\r\n> Num Epochs = 3\r\n> Instantaneous batch size per device = 32\r\n> Total train batch size (w. parallel, distributed & accumulation) = 256\r\n> Gradient Accumulation steps = 2\r\n> Steps per epoch = 94\r\n> Total optimization steps = 282\r\n> {'loss': 17.029345, 'learning_rate': 4.822695e-05, 'epoch': 0.11702127659574468, 'step': 10}\r\n> {'loss': 13.583568, 'learning_rate': 4.6453897e-05, 'epoch': 0.22340425531914893, 'step': 20}\r\n> {'loss': 11.413903, 'learning_rate': 4.468085e-05, 'epoch': 0.32978723404255317, 'step': 30}\r\n> {'loss': 9.977048, 'learning_rate': 4.29078e-05, 'epoch': 0.43617021276595747, 'step': 40}\r\n> {'loss': 8.904137, 'learning_rate': 4.113475e-05, 'epoch': 0.5425531914893617, 'step': 50}\r\n> {'loss': 8.056796, 'learning_rate': 3.93617e-05, 'epoch': 0.648936170212766, 'step': 60}\r\n> {'loss': 7.339738, 'learning_rate': 3.758865e-05, 'epoch': 0.7553191489361702, 'step': 70}\r\n> {'loss': 6.7678766, 'learning_rate': 3.5815603e-05, 'epoch': 0.8617021276595744, 'step': 80}\r\n> {'loss': 6.2809086, 'learning_rate': 3.4042554e-05, 'epoch': 0.9680851063829787, 'step': 90}\r\n> 2020-09-11 23:36:37.142279: I tensorflow/core/kernels/data/shuffle_dataset_op.cc:172] Filling up shuffle buffer (this may take a while): 23596 of 24000\r\n> 2020-09-11 23:36:37.311979: I tensorflow/core/kernels/data/shuffle_dataset_op.cc:221] Shuffle buffer filled.\r\n> {'loss': 2.2556372, 'learning_rate': 3.22695e-05, 'epoch': 1.0638297872340425, 'step': 100}\r\n> {'loss': 2.0573573, 'learning_rate': 3.0496454e-05, 'epoch': 1.1702127659574468, 'step': 110}\r\n> {'loss': 1.9544038, 'learning_rate': 2.8723403e-05, 'epoch': 1.2765957446808511, 'step': 120}\r\n> {'loss': 1.8848253, 'learning_rate': 2.6950353e-05, 'epoch': 1.3829787234042552, 'step': 130}\r\n> {'loss': 1.837054, 'learning_rate': 2.5177305e-05, 'epoch': 1.4893617021276595, 'step': 140}\r\n> {'loss': 1.7885295, 'learning_rate': 2.3404255e-05, 'epoch': 1.5957446808510638, 'step': 150}\r\n> {'loss': 1.7535, 'learning_rate': 2.1631204e-05, 'epoch': 1.702127659574468, 'step': 160}\r\n> {'loss': 1.7068337, 'learning_rate': 1.9858155e-05, 'epoch': 1.8085106382978724, 'step': 170}\r\n> {'loss': 1.6874169, 'learning_rate': 1.8085108e-05, 'epoch': 1.9148936170212765, 'step': 180}\r\n> 2020-09-11 23:41:05.123380: I tensorflow/core/kernels/data/shuffle_dataset_op.cc:172] Filling up shuffle buffer (this may take a while): 23596 of 24000\r\n> 2020-09-11 23:36:05.332080: I tensorflow/core/kernels/data/shuffle_dataset_op.cc:221] Shuffle buffer filled.\r\n> {'loss': 1.189175, 'learning_rate': 1.6312057e-05, 'epoch': 2.021276595744681, 'step': 190}\r\n> {'loss': 1.2820004, 'learning_rate': 1.4539006e-05, 'epoch': 2.127659574468085, 'step': 200}\r\n> {'loss': 1.2558761, 'learning_rate': 1.27659605e-05, 'epoch': 2.2340425531914896, 'step': 210}\r\n> {'loss': 1.2865627, 'learning_rate': 1.0992908e-05, 'epoch': 2.3404255319148937, 'step': 220}\r\n> {'loss': 1.2853887, 'learning_rate': 9.219858e-06, 'epoch': 2.4468085106382977, 'step': 230}\r\n> {'loss': 1.2650005, 'learning_rate': 7.4468076e-06, 'epoch': 2.5531914893617023, 'step': 240}\r\n> {'loss': 1.2414478, 'learning_rate': 5.67376e-06, 'epoch': 2.6595744680851063, 'step': 250}\r\n> {'loss': 1.243986, 'learning_rate': 3.9007095e-06, 'epoch': 2.7659574468085104, 'step': 260}\r\n> {'loss': 1.2295542, 'learning_rate': 2.1276592e-06, 'epoch': 2.872340425531915, 'step': 270}\r\n> {'loss': 1.2296557, 'learning_rate': 3.5460886e-07, 'epoch': 2.978723404255319, 'step': 280}\r\n> Training took: 0:19:10.070496\r\n> Saving model in /home/jplu/model\r\n> Configuration saved in /home/jplu/model/config.json\r\n> Model weights saved in /home/jplu/model/tf_model.h5\r\n> ***** Running Evaluation *****\r\n> Num examples = 2200\r\n> Batch size = 32\r\n> {'eval_loss': 1.5663623533387114, 'eval_precision': 0.8107354478912513, 'eval_recall': 0.8548327820654171, 'eval_f1': 0.8322003577817532, 'epoch': 3.0, 'step': 282}\r\n> ```\r\n> \r\n> Looks ok to me. I have also tested for text classification and question answering without any error.\r\n\r\nWow! Thank you @jplu, a test and a reply on Friday night :) \r\nI might need to get some gpu though if I continue to contribute - can't let you do all such tests all the time.\r\nGreat to see it works.", "Ahah no worries it is ok, not everybody can have such setup.\n\n@LysandreJik looks ok to merge." ]

[ { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "Indeed, it seems it hasn't been up to date. Did you try running `run_squad_trainer.py`? It should be more up to date.\r\n\r\n@sgugger we should probably deprecate `run_squad.py` now that we have a Trainer-based SQuAD script.", "The missing part was the eval which shouldn't be too hard to add (https://github.com/huggingface/transformers/pull/4829#issuecomment-645994130)\r\n\r\nAnd then we can rename files as in https://github.com/huggingface/transformers/pull/5582", "> Indeed, it seems it hasn't been up to date. Did you try running `run_squad_trainer.py`? It should be more up to date.\r\n> \r\n> @sgugger we should probably deprecate `run_squad.py` now that we have a Trainer-based SQuAD script.\r\n\r\nNo run_squad_trainer.py also gives error in 3.1.0\r\n`!python transformers/examples/question-answering/run_squad_trainer.py --help\r\n`\r\nError is \r\n`python3: can't open file 'transformers/examples/question-answering/run_squad_trainer.py': [Errno 2] No such file or directory`", "@deepanshu650 the file exists, it's [here](https://github.com/huggingface/transformers/blob/master/examples/question-answering/run_squad_trainer.py). Are you sure you cloned the v3.1.0 repo?", "Yes it's working , initially I was working on copy of someone's notebook which was installing v2.3.0 and I had to pip install v3.1.0, so after changing to new notebook it starts running.\r\nBut it doesn't give evaluation results(which `run_squad.py` does) as `run_squad_trainer.py` not calling trainer.evaluate() though it makes eval_dataset .\r\nSo how do I evaluate . Thanks for replying.", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n" ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6996?src=pr&el=h1) Report\n> Merging [#6996](https://codecov.io/gh/huggingface/transformers/pull/6996?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/90ec78b5140251f093f658ebd4d2925e8c03f5e6?el=desc) will **decrease** coverage by `0.55%`.\n> The diff coverage is `100.00%`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/6996?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #6996 +/- ##\n==========================================\n- Coverage 80.58% 80.03% -0.56% \n==========================================\n Files 161 161 \n Lines 30123 30123 \n==========================================\n- Hits 24276 24109 -167 \n- Misses 5847 6014 +167 \n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/6996?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/generation\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6996/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9nZW5lcmF0aW9uX3V0aWxzLnB5) | `97.21% <100.00%> (ø)` | |\n| [src/transformers/modeling\\_tf\\_electra.py](https://codecov.io/gh/huggingface/transformers/pull/6996/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9lbGVjdHJhLnB5) | `25.13% <0.00%> (-73.83%)` | :arrow_down: |\n| [src/transformers/modeling\\_tf\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6996/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl91dGlscy5weQ==) | `86.97% <0.00%> (-0.33%)` | :arrow_down: |\n| [src/transformers/tokenization\\_utils\\_base.py](https://codecov.io/gh/huggingface/transformers/pull/6996/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy90b2tlbml6YXRpb25fdXRpbHNfYmFzZS5weQ==) | `93.63% <0.00%> (-0.14%)` | :arrow_down: |\n| [src/transformers/data/data\\_collator.py](https://codecov.io/gh/huggingface/transformers/pull/6996/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9kYXRhL2RhdGFfY29sbGF0b3IucHk=) | `91.90% <0.00%> (+5.26%)` | :arrow_up: |\n| [src/transformers/modeling\\_tf\\_flaubert.py](https://codecov.io/gh/huggingface/transformers/pull/6996/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9mbGF1YmVydC5weQ==) | `88.34% <0.00%> (+63.80%)` | :arrow_up: |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/6996?src=pr&el=continue).\n> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta)\n> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`\n> Powered by [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6996?src=pr&el=footer). Last update [90ec78b...fd199a5](https://codecov.io/gh/huggingface/transformers/pull/6996?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n", "Shouldn't this be part of #6996? Or would this PR be needed either way?", "> Shouldn't this be part of #6996? Or would this PR be needed either way?\r\n\r\nDid you mean \"part of \" https://github.com/huggingface/transformers/pull/6940?\r\n\r\nIt's there already, and yes it's required to work. I just thought that it's the best to change core functions in separate PRs and not as a part of a much larger PR. Please advise if this is not the right approach.\r\n\r\nIn this particular case `config.decoder` is more specific than `config` and therefore should be checked first. I don't think currently there is any more that actively uses `config.decoder` (grep didn't find any), therefore it must have been untested since it was added in first place perhaps?\r\n\r\n", "Don't really agree with this PR - I think `bos_token_id` should have higher priority than `model.config.decoder.bos_token_id`. The `model.config.decoder.bos_token_id` was mainly added because of the `EncoderDecoderModel` framework", "Yes I meant \"part of \" #6940? \r\nIf two PRs do not exist/make sense without each other I think they should be together. \r\nOtherwise we can merge one without the other and have either broken or dead code", "@stas00 why can't you use `decoder_start_token_id` for FSMT?", "> why can't you use `decoder_start_token_id` for FSMT?\r\n\r\nThat works - thank you for the suggestion, @sshleifer \r\n", "> Don't really agree with this PR - I think `bos_token_id` should have higher priority than `model.config.decoder.bos_token_id`. The `model.config.decoder.bos_token_id` was mainly added because of the `EncoderDecoderModel` framework\r\n\r\nThe way I read the intention is that if it goes:\r\n```\r\n if self.config.is_encoder_decoder:\r\n```\r\nwe are in encoder-decoder zone and as such `.decoder` should take priority\r\n\r\nI guess Bart and friends are semi-encoder-decoder as far as the framework goes.\r\n\r\nThank you for the feedback, @patrickvonplaten \r\n\r\nI suppose in the ideal world we should have tests that validate such scenarios." ]

[]

[]

[ "> Not sure I fully understand the use case, but nothing against the principle of it.\r\n\r\nThe idea is to prevent combinatorial explosion of \"model types\" when only the tokenizer is different (e.g. Flaubert, CamemBERT if we wanted to support them today)\r\n\r\nIn the future we might even want to have a few model-agnostic tokenizer classes like ByteLevelBPETokenizer (basically RobertaTokenizer), as they can be initialized pretty exhaustively from the init args stored in `tokenizer_config.json`\r\n\r\n\r\n\r\n", "Documented by @sgugger in https://github.com/huggingface/transformers/pull/8152" ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6994?src=pr&el=h1) Report\n> Merging [#6994](https://codecov.io/gh/huggingface/transformers/pull/6994?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/90ec78b5140251f093f658ebd4d2925e8c03f5e6?el=desc) will **decrease** coverage by `0.54%`.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/6994?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #6994 +/- ##\n==========================================\n- Coverage 80.58% 80.04% -0.55% \n==========================================\n Files 161 161 \n Lines 30123 30123 \n==========================================\n- Hits 24276 24111 -165 \n- Misses 5847 6012 +165 \n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/6994?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/modeling\\_tf\\_electra.py](https://codecov.io/gh/huggingface/transformers/pull/6994/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9lbGVjdHJhLnB5) | `25.13% <0.00%> (-73.83%)` | :arrow_down: |\n| [src/transformers/modeling\\_tf\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6994/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl91dGlscy5weQ==) | `86.97% <0.00%> (-0.33%)` | :arrow_down: |\n| [src/transformers/file\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6994/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9maWxlX3V0aWxzLnB5) | `82.41% <0.00%> (-0.26%)` | :arrow_down: |\n| [src/transformers/data/data\\_collator.py](https://codecov.io/gh/huggingface/transformers/pull/6994/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9kYXRhL2RhdGFfY29sbGF0b3IucHk=) | `92.71% <0.00%> (+6.07%)` | :arrow_up: |\n| [src/transformers/modeling\\_tf\\_flaubert.py](https://codecov.io/gh/huggingface/transformers/pull/6994/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9mbGF1YmVydC5weQ==) | `88.34% <0.00%> (+63.80%)` | :arrow_up: |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/6994?src=pr&el=continue).\n> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta)\n> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`\n> Powered by [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6994?src=pr&el=footer). Last update [90ec78b...09dda6e](https://codecov.io/gh/huggingface/transformers/pull/6994?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n" ]

[ { "id": 1845609017, "node_id": "MDU6TGFiZWwxODQ1NjA5MDE3", "url": "https://api.github.com/repos/huggingface/transformers/labels/seq2seq", "name": "seq2seq", "color": "fef2c0", "default": false, "description": "" } ]

[ { "login": "sshleifer", "id": 6045025, "node_id": "MDQ6VXNlcjYwNDUwMjU=", "avatar_url": "https://avatars.githubusercontent.com/u/6045025?v=4", "gravatar_id": "", "url": "https://api.github.com/users/sshleifer", "html_url": "https://github.com/sshleifer", "followers_url": "https://api.github.com/users/sshleifer/followers", "following_url": "https://api.github.com/users/sshleifer/following{/other_user}", "gists_url": "https://api.github.com/users/sshleifer/gists{/gist_id}", "starred_url": "https://api.github.com/users/sshleifer/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/sshleifer/subscriptions", "organizations_url": "https://api.github.com/users/sshleifer/orgs", "repos_url": "https://api.github.com/users/sshleifer/repos", "events_url": "https://api.github.com/users/sshleifer/events{/privacy}", "received_events_url": "https://api.github.com/users/sshleifer/received_events", "type": "User", "site_admin": false } ]

[ "The easiest way I can think of is to avoid generating the unk token altogether.\r\n\r\nadd the following method to `PegasusForConditionalGeneration`\r\n\r\n```python\r\n def adjust_logits_during_generation(self, logits, cur_len, max_length):\r\n # Note, this will break with a tokenizer that is not PegasusTokenizer\r\n logits[:, list(range(2, 105))] = float(\"-inf\") # never predict unk tokens\r\n if cur_len == max_length - 1 and self.config.eos_token_id is not None:\r\n self._force_token_ids_generation(logits, self.config.eos_token_id)\r\n return logits\r\n```\r\n\r\nLet me know if that helps!\r\n\r\n", "> The easiest way I can think of is to avoid generating the unk token altogether.\r\n> \r\n> add the following method to `PegasusForConditionalGeneration`\r\n> \r\n> ```python\r\n> def adjust_logits_during_generation(self, logits, cur_len, max_length):\r\n> # Note, this will break with a tokenizer that is not PegasusTokenizer\r\n> logits[:, list(range(2, 105))] = float(\"-inf\") # never predict unk tokens\r\n> if cur_len == max_length - 1 and self.config.eos_token_id is not None:\r\n> self._force_token_ids_generation(logits, self.config.eos_token_id)\r\n> return logits\r\n> ```\r\n> \r\n> Let me know if that helps!\r\n\r\nThank you for the reply!\r\nI've added the code and checked it's being run, but unfortunately the output still stops at an unknown token.", "I can't replicate.\r\ngot \r\n> As a child, the documentary maker, who was born in the Indian state of Kerala but now lives in Toronto, saw ceremonial elephants being paraded and thought they were beautiful.\r\n\r\non the branch of #7014 \r\n", "Setting the min_length parameter to 100 yields the problem (as I should have mentioned). Might this be an issue with the minimum length being too long relative to the size of the input?", "Yeah. the `xsum` model especially is trained to generate very short summaries.\r\n`pegasus-arxiv` for example, can generate up to 256 tokens.\r\n\r\nyou can see each available checkpoint and it's maximum input and output sizes [here](https://github.com/huggingface/transformers/blob/0f58903bb62870342eae52f5a02c9105ec6f9b1e/src/transformers/configuration_pegasus.py#L50)\r\n\r\n+ `max_length`: max length to generate\r\n+ `max_position_embeddings`: max input size." ]

[]

[]

[ "MobileBERT is supported, see [here](https://github.com/huggingface/transformers/blob/master/src/transformers/modeling_mobilebert.py) for the code, and [here](https://huggingface.co/transformers/model_doc/mobilebert.html) for the docs. [Here](https://huggingface.co/models?search=mobilebert) are all the available mobilebert models on the hub." ]

[]

[]

[ "https://github.com/huggingface/transformers/blob/0203ad43bcd0b29423dec6ca1a58ed58300f0d61/src/transformers/convert_mbart_original_checkpoint_to_pytorch.py#L7\r\n\r\nHello! Does this line also need to be changed?" ]

[ { "id": 1838412367, "node_id": "MDU6TGFiZWwxODM4NDEyMzY3", "url": "https://api.github.com/repos/huggingface/transformers/labels/model%20card", "name": "model card", "color": "92d5f4", "default": false, "description": "Related to pretrained model cards" } ]

[]

[]

[]

[]

[ "Solved with https://github.com/huggingface/transformers/issues/319" ]

[]

[]

[]

[]

[]

[ "This happens because you don't have sufficient memory in your machine, indeed. You can try reducing the batch size.", "> This happens because you don't have sufficient memory in your machine, indeed. You can try reducing the batch size.\r\n\r\nIn my opinion, the problem occurs may due to my dataset is too large(Relative to the memory of my machine). But the loading data part may be optimized referring to the following issue. \r\nhttps://stackoverflow.com/questions/51444059/how-to-iterate-over-two-dataloaders-simultaneously-using-pytorch/57890309#57890309 ", "There are two questions:\r\n(1) I've changed a machine to run the code. It started running normal, but will quit midway while training. Is this also related to my machine's memory? \r\nThis is the data of my machine during training.\r\n\r\nThis is the exit interface. I don't know what's the matter. \r\n\r\n\r\n(2) In addition , how can i use gpu to run `run_language_modeling.py` please. Thanks a lot.\r\n", "I have a similar problem. Was there any solution in your case?\r\n", "same problem", "same problem", "Were there any solutions to this? I am encountering the exact same issue trying to train Dolly on the original dataset.", "same question when i train bloom", "same issue, please let me know if any solution on this has been done\r\n" ]

[]

[]

[]

[ { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "Have you tried the finetune.sh script shown [here](https://github.com/huggingface/transformers/blob/master/examples/seq2seq/finetune.sh)? In addition to the short list of CLI flags listed there, you could try adding:\r\n\r\n```\r\n--src_lang \"en\" \\\r\n--tgt_lang \"de\" \\\r\n--num_train_epochs 400 \\\r\n--warmup_steps 20 \\\r\n--train_batch_size 32 \\\r\n--eval_batch_size 32 \\\r\n--data_dir \"/data/dir\" \\\r\n--output_dir \"/path/to/store/model/etc\" \\\r\n--cache_dir \"/path/for/misc/files\" \\\r\n--max_source_length 128 \\\r\n--max_target_length 128 \\\r\n--val_max_target_length 128 \\\r\n--test_max_target_length 128 \\\r\n--model_name_or_path \"</path/to/pretrained>\"\r\n```\r\n\r\nwhere the \"/path/to/pretrained\" could be either a local path on your machine or MarianMT model (Opus-en-de or equivalent). The \"data/dir\" has a \"train.source\" and \"train.target\" for the source & target languages, such that line number x of the target is a translation of line x in the source (and same with \"val.source\" and \"val.target\"). I have changed the finetune.py script [here](https://github.com/huggingface/transformers/blob/77cd0e13d2d09f60d2f6d8fb8b08f493d7ca51fe/examples/seq2seq/finetune.py#L415) to \r\n```\r\nparser = TranslationModule.add_model_specific_args(parser, os.getcwd())\r\n\r\n```\r\nand then ran the finetune.sh script.\r\n\r\n\r\nNote: The gradients blew up when I used the \"fp16\" flag (with Pytorch 1.6), so I had removed it. Also, you might want to check on the \"val_check_interval\", \"check_val_every_n_epoch\", and probably check [this issue](https://github.com/huggingface/transformers/issues/3447) on how to save multiple checkpoints.", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n" ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6984?src=pr&el=h1) Report\n> Merging [#6984](https://codecov.io/gh/huggingface/transformers/pull/6984?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/995a958dd18d4326e608efc3bfc4005acfef8e56?el=desc) will **increase** coverage by `0.26%`.\n> The diff coverage is `100.00%`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/6984?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #6984 +/- ##\n==========================================\n+ Coverage 80.03% 80.30% +0.26% \n==========================================\n Files 161 161 \n Lines 30122 30123 +1 \n==========================================\n+ Hits 24108 24190 +82 \n+ Misses 6014 5933 -81 \n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/6984?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/modeling\\_bert.py](https://codecov.io/gh/huggingface/transformers/pull/6984/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19iZXJ0LnB5) | `88.44% <100.00%> (+0.01%)` | :arrow_up: |\n| [src/transformers/modeling\\_tf\\_albert.py](https://codecov.io/gh/huggingface/transformers/pull/6984/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9hbGJlcnQucHk=) | `21.47% <0.00%> (-69.44%)` | :arrow_down: |\n| [src/transformers/modeling\\_tf\\_flaubert.py](https://codecov.io/gh/huggingface/transformers/pull/6984/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9mbGF1YmVydC5weQ==) | `24.53% <0.00%> (-63.81%)` | :arrow_down: |\n| [src/transformers/tokenization\\_albert.py](https://codecov.io/gh/huggingface/transformers/pull/6984/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy90b2tlbml6YXRpb25fYWxiZXJ0LnB5) | `28.84% <0.00%> (-58.66%)` | :arrow_down: |\n| [src/transformers/modeling\\_lxmert.py](https://codecov.io/gh/huggingface/transformers/pull/6984/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19seG1lcnQucHk=) | `70.01% <0.00%> (-20.75%)` | :arrow_down: |\n| [src/transformers/generation\\_tf\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6984/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9nZW5lcmF0aW9uX3RmX3V0aWxzLnB5) | `85.46% <0.00%> (-1.26%)` | :arrow_down: |\n| [src/transformers/modeling\\_tf\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6984/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl91dGlscy5weQ==) | `87.29% <0.00%> (+0.32%)` | :arrow_up: |\n| [src/transformers/tokenization\\_utils\\_base.py](https://codecov.io/gh/huggingface/transformers/pull/6984/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy90b2tlbml6YXRpb25fdXRpbHNfYmFzZS5weQ==) | `94.03% <0.00%> (+0.40%)` | :arrow_up: |\n| [src/transformers/modeling\\_bart.py](https://codecov.io/gh/huggingface/transformers/pull/6984/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19iYXJ0LnB5) | `94.41% <0.00%> (+0.50%)` | :arrow_up: |\n| [src/transformers/activations.py](https://codecov.io/gh/huggingface/transformers/pull/6984/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9hY3RpdmF0aW9ucy5weQ==) | `90.00% <0.00%> (+5.00%)` | :arrow_up: |\n| ... and [3 more](https://codecov.io/gh/huggingface/transformers/pull/6984/diff?src=pr&el=tree-more) | |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/6984?src=pr&el=continue).\n> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta)\n> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`\n> Powered by [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6984?src=pr&el=footer). Last update [995a958...8ecbd15](https://codecov.io/gh/huggingface/transformers/pull/6984?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n" ]

[]

[ { "login": "sshleifer", "id": 6045025, "node_id": "MDQ6VXNlcjYwNDUwMjU=", "avatar_url": "https://avatars.githubusercontent.com/u/6045025?v=4", "gravatar_id": "", "url": "https://api.github.com/users/sshleifer", "html_url": "https://github.com/sshleifer", "followers_url": "https://api.github.com/users/sshleifer/followers", "following_url": "https://api.github.com/users/sshleifer/following{/other_user}", "gists_url": "https://api.github.com/users/sshleifer/gists{/gist_id}", "starred_url": "https://api.github.com/users/sshleifer/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/sshleifer/subscriptions", "organizations_url": "https://api.github.com/users/sshleifer/orgs", "repos_url": "https://api.github.com/users/sshleifer/repos", "events_url": "https://api.github.com/users/sshleifer/events{/privacy}", "received_events_url": "https://api.github.com/users/sshleifer/received_events", "type": "User", "site_admin": false }, { "login": "patrickvonplaten", "id": 23423619, "node_id": "MDQ6VXNlcjIzNDIzNjE5", "avatar_url": "https://avatars.githubusercontent.com/u/23423619?v=4", "gravatar_id": "", "url": "https://api.github.com/users/patrickvonplaten", "html_url": "https://github.com/patrickvonplaten", "followers_url": "https://api.github.com/users/patrickvonplaten/followers", "following_url": "https://api.github.com/users/patrickvonplaten/following{/other_user}", "gists_url": "https://api.github.com/users/patrickvonplaten/gists{/gist_id}", "starred_url": "https://api.github.com/users/patrickvonplaten/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/patrickvonplaten/subscriptions", "organizations_url": "https://api.github.com/users/patrickvonplaten/orgs", "repos_url": "https://api.github.com/users/patrickvonplaten/repos", "events_url": "https://api.github.com/users/patrickvonplaten/events{/privacy}", "received_events_url": "https://api.github.com/users/patrickvonplaten/received_events", "type": "User", "site_admin": false } ]

[ "I think eos is always defined, but I think this (or just checking `pad_token_id` was one of Patrick's first PRs. He would know more.", "Thank you for that feedback, @sshleifer.\r\n\r\nIf it makes things simpler, I could re-work both functions wrt these 2 tokens' definition checks and you can review a PR instead. \r\n\r\nI just wanted to validate that the issue is real and I'm not missing something obvious before I invest time into doing that.", "Hey @stas00,\r\n\r\nThis is definitely part of the code that should be refactored :D Super hard to follow the logic there :-/\r\n\r\nAs a start, this PR is probably quite useful for context: https://github.com/huggingface/transformers/pull/2885. So there are a couple of models where EOS token is not defined and I'm quite sure that the code you linked does not always get hit. It can very well be that we apply beam search to `OpenAIGPT` - with a given `max_length`. `OpenAIGPT` does not have an EOS token, but beam search should work nevertheless. \r\n\r\nIt's quite a tricky pad token / eos token / ... logic that is implemented there. I think we have to be super careful to not break anything here - even if all the slow tests pass, it might not be enough (`OpenAIGPT` beam search is not integration tested...)\r\n\r\n Also, I'm currently working on refactoring the generate function, will ping you guys in a couple of days with a first design proposition. My idea is to pull apart beam search + greedy / beam search + sampling / no beam search + greedy / no beam searh + greedy to make everything more readable. I'm not sure whether it's worth diving deep into the generate() logic before we have a more readable code", "That sounds like a fantastic plan, @patrickvonplaten!\r\n\r\n> So there are a couple of models where EOS token is not defined and I'm quite sure that the code you linked does not always get hit. \r\n\r\nI stand corrected, that's good to know, thank you!. \r\n\r\nThat means that the code is very tricky, since a reader will expect that at some point the generation should be complete and `done` set to True, which currently absolutely requires eos. I haven't considered the case where it'll go through that loop and not hit done. If I follow it carefully it only happens if `max_length` is reached and there is no `done` yet, and moreover it has to be that the hypos are exactly of the same length. if they aren't the same, eos is almost always required.\r\n\r\nAs you are saying there isn't really a test that covers that (odd?) case. Actually, PR https://github.com/huggingface/transformers/pull/6982 is very likely to break it then, since now it requires eos for both situations where hypos are of the same length and are not. But if it breaks that very special case, then the issue lies elsewhere and it just happened to work. (As I suggested I changed \"is\" for \"was\" in an input and suddenly eos was gone from all of the hypos.) \r\n\r\nNote: I have only run the code in my head and haven't validated that in fact it'd break something. It's possible that you're talking about a completely different case.\r\n", "I think your PR is fine because if no `eos_token_id` is defined, this condition can never happen: `sent_lengths[i] < max_length:`. \r\nWhat I mean is that if no `eos_token_id` is defined no matter what `generate()` method is used, all sent_length will always be == `max_length` and the condition will not be hit.", "ah, yes, you're absolutely correct, Patrick - you definitely have been holding that generation code in your head for much longer than I - I don't have the full coverage yet :)", "Reopen if this was a mistake!" ]

[]

[]

[ "IMO, `eos` should not always be there. The reason is that if the user defines `max_length=30` and the EOS token was not generated by the model, then no EOS token should be added. I think EOS token should only be added if it is produced by the model. *E.g.* a generated sentence like \"I will go to the office and\" should not have an added EOS token at the end.", "> [...] _E.g._ a generated sentence like \"I will go to the office and\" should not have an added EOS token at the end.\r\n\r\nThank you for explaining that, @patrickvonplaten.\r\n\r\nCould you also please review this PR, as it's unrelated to max_length or undefined EOS, that was a related question that I wasn't sure about.\r\n", "@stas00\r\nHave you run the slow tests that might be effected? (will take 10-30 mins)\r\n\r\n```\r\nrun_generation_integration_tests () {\r\n\t# assumes USE_CUDA is exported, rather than passed\r\n\tRUN_SLOW=1 pytest tests/test_modeling_pegasus.py\r\n\tRUN_SLOW=1 pytest tests/test_modeling_bart.py\r\n\tRUN_SLOW=1 pytest tests/test_modeling_t5.py\r\n\tRUN_SLOW=1 pytest tests/test_modeling_marian.py\r\n\tRUN_SLOW=1 pytest tests/test_modeling_mbart.py\r\n\tRUN_SLOW=1 pytest tests/test_modeling_encoder_decoder.py\r\n\tRUN_SLOW=1 pytest tests/test_pipelines.py\r\n\tRUN_SLOW=1 pytest tests/test_modeling_gpt2.py\r\n}\r\n```", "Good call, @sshleifer! (I edited the last one to `tests/test_modeling_gpt2.py`)\r\n\r\n```\r\nRUN_SLOW=1 pytest --disable-warnings tests/test_modeling_pegasus.py tests/test_modeling_bart.py tests/test_modeling_t5.py tests/test_modeling_marian.py tests/test_modeling_mbart.py tests/test_modeling_encoder_decoder.py tests/test_pipelines.py tests/test_modeling_gpt2.py\r\n====================================================================== test session starts =======================================================================\r\nplatform linux -- Python 3.7.5, pytest-5.4.1, py-1.8.1, pluggy-0.13.1\r\nrootdir: /mnt/nvme1/code/huggingface/transformers\r\nplugins: hypothesis-5.5.4, timeout-1.4.2, filter-subpackage-0.1.1, arraydiff-0.3, flaky-3.6.1, ipynb-1.1.1.dev0, cov-2.10.0, astropy-header-0.1.2, forked-1.2.0, doctestplus-0.5.0, openfiles-0.4.0, remotedata-0.3.2, xdist-1.32.0, repeat-0.8.0, flakefinder-1.0.0\r\ncollected 211 items\r\n\r\ntests/test_modeling_pegasus.py .. [ 0%]\r\ntests/test_modeling_bart.py ..............s....s............................ [ 23%]\r\ntests/test_modeling_t5.py ......................s..s........... [ 41%]\r\ntests/test_modeling_marian.py ................ [ 48%]\r\ntests/test_modeling_mbart.py s...s. [ 51%]\r\ntests/test_modeling_encoder_decoder.py ............................... [ 66%]\r\ntests/test_pipelines.py ....................................... [ 84%]\r\ntests/test_modeling_gpt2.py .......................s........ [100%]\r\n\r\n==================================================== 204 passed, 7 skipped, 45 warnings in 980.34s (0:16:20) =====================================================\r\n```", "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6982?src=pr&el=h1) Report\n> Merging [#6982](https://codecov.io/gh/huggingface/transformers/pull/6982?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/ce37be9d94da57897cce9c49b3421e6a8a927d4a?el=desc) will **increase** coverage by `2.39%`.\n> The diff coverage is `100.00%`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/6982?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #6982 +/- ##\n==========================================\n+ Coverage 77.60% 80.00% +2.39% \n==========================================\n Files 161 161 \n Lines 30120 30119 -1 \n==========================================\n+ Hits 23374 24096 +722 \n+ Misses 6746 6023 -723 \n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/6982?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/generation\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6982/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9nZW5lcmF0aW9uX3V0aWxzLnB5) | `97.20% <100.00%> (-0.01%)` | :arrow_down: |\n| [src/transformers/modeling\\_tf\\_electra.py](https://codecov.io/gh/huggingface/transformers/pull/6982/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9lbGVjdHJhLnB5) | `25.13% <0.00%> (-73.83%)` | :arrow_down: |\n| [src/transformers/modeling\\_marian.py](https://codecov.io/gh/huggingface/transformers/pull/6982/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19tYXJpYW4ucHk=) | `60.00% <0.00%> (-30.00%)` | :arrow_down: |\n| [src/transformers/activations.py](https://codecov.io/gh/huggingface/transformers/pull/6982/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9hY3RpdmF0aW9ucy5weQ==) | `85.00% <0.00%> (-5.00%)` | :arrow_down: |\n| [src/transformers/modeling\\_bart.py](https://codecov.io/gh/huggingface/transformers/pull/6982/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19iYXJ0LnB5) | `93.90% <0.00%> (-0.51%)` | :arrow_down: |\n| [src/transformers/modeling\\_tf\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6982/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl91dGlscy5weQ==) | `86.97% <0.00%> (-0.33%)` | :arrow_down: |\n| [src/transformers/tokenization\\_utils\\_base.py](https://codecov.io/gh/huggingface/transformers/pull/6982/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy90b2tlbml6YXRpb25fdXRpbHNfYmFzZS5weQ==) | `93.76% <0.00%> (-0.28%)` | :arrow_down: |\n| [src/transformers/generation\\_tf\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6982/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9nZW5lcmF0aW9uX3RmX3V0aWxzLnB5) | `85.21% <0.00%> (+1.00%)` | :arrow_up: |\n| [src/transformers/modeling\\_lxmert.py](https://codecov.io/gh/huggingface/transformers/pull/6982/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19seG1lcnQucHk=) | `90.76% <0.00%> (+20.74%)` | :arrow_up: |\n| [src/transformers/modeling\\_tf\\_flaubert.py](https://codecov.io/gh/huggingface/transformers/pull/6982/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9mbGF1YmVydC5weQ==) | `88.34% <0.00%> (+63.80%)` | :arrow_up: |\n| ... and [2 more](https://codecov.io/gh/huggingface/transformers/pull/6982/diff?src=pr&el=tree-more) | |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/6982?src=pr&el=continue).\n> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta)\n> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`\n> Powered by [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6982?src=pr&el=footer). Last update [ce37be9...85dd09d](https://codecov.io/gh/huggingface/transformers/pull/6982?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n" ]

[]

[]

[ "`tokenizer` is not a callable in v.2.11.0, which is why you are getting this error. Also the example that you posted is from stable docs not from 2.11.0, upgrading to latest stable version should resolve the issue.", "@patil-suraj\r\nSorry for asking this naive ques but how to upgrade to latest stable version?\r\nand\r\nWhat about the first error?\r\nTypeError: init() got an unexpected keyword argument 'return_dict'\r\n\r\nCan you please point to some reference or any link?", "Upgrading to latest version should also resolve the first issue. To upgrade\r\n\r\n`pip install -U transformers`", "Thank You. It worked." ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6980?src=pr&el=h1) Report\n> Merging [#6980](https://codecov.io/gh/huggingface/transformers/pull/6980?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/ce37be9d94da57897cce9c49b3421e6a8a927d4a?el=desc) will **increase** coverage by `0.39%`.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/6980?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #6980 +/- ##\n==========================================\n+ Coverage 77.60% 77.99% +0.39% \n==========================================\n Files 161 161 \n Lines 30120 30120 \n==========================================\n+ Hits 23374 23492 +118 \n+ Misses 6746 6628 -118 \n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/6980?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/generation\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6980/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9nZW5lcmF0aW9uX3V0aWxzLnB5) | `97.21% <ø> (ø)` | |\n| [src/transformers/modeling\\_tf\\_longformer.py](https://codecov.io/gh/huggingface/transformers/pull/6980/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9sb25nZm9ybWVyLnB5) | `16.37% <0.00%> (-82.31%)` | :arrow_down: |\n| [src/transformers/configuration\\_lxmert.py](https://codecov.io/gh/huggingface/transformers/pull/6980/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9jb25maWd1cmF0aW9uX2x4bWVydC5weQ==) | `20.00% <0.00%> (-80.00%)` | :arrow_down: |\n| [src/transformers/modeling\\_tf\\_lxmert.py](https://codecov.io/gh/huggingface/transformers/pull/6980/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9seG1lcnQucHk=) | `22.49% <0.00%> (-71.63%)` | :arrow_down: |\n| [src/transformers/tokenization\\_albert.py](https://codecov.io/gh/huggingface/transformers/pull/6980/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy90b2tlbml6YXRpb25fYWxiZXJ0LnB5) | `28.84% <0.00%> (-58.66%)` | :arrow_down: |\n| [src/transformers/modeling\\_lxmert.py](https://codecov.io/gh/huggingface/transformers/pull/6980/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19seG1lcnQucHk=) | `23.50% <0.00%> (-46.52%)` | :arrow_down: |\n| [src/transformers/modeling\\_roberta.py](https://codecov.io/gh/huggingface/transformers/pull/6980/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19yb2JlcnRhLnB5) | `77.73% <0.00%> (-19.35%)` | :arrow_down: |\n| [src/transformers/generation\\_tf\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6980/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9nZW5lcmF0aW9uX3RmX3V0aWxzLnB5) | `86.46% <0.00%> (+2.25%)` | :arrow_up: |\n| [src/transformers/modeling\\_tf\\_flaubert.py](https://codecov.io/gh/huggingface/transformers/pull/6980/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9mbGF1YmVydC5weQ==) | `88.34% <0.00%> (+63.80%)` | :arrow_up: |\n| [src/transformers/modeling\\_tf\\_mobilebert.py](https://codecov.io/gh/huggingface/transformers/pull/6980/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9tb2JpbGViZXJ0LnB5) | `96.91% <0.00%> (+72.35%)` | :arrow_up: |\n| ... and [2 more](https://codecov.io/gh/huggingface/transformers/pull/6980/diff?src=pr&el=tree-more) | |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/6980?src=pr&el=continue).\n> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta)\n> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`\n> Powered by [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6980?src=pr&el=footer). Last update [ce37be9...2b9171e](https://codecov.io/gh/huggingface/transformers/pull/6980?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n" ]

[]

[]

[ "Hi @seyonechithrananda , I'm facing the same problem, how did you solve this issue?", "Ping here about that, I'm having the same problem", "I am suffering from the same problem, would like to know the solution if there is any", "Same problem here. When I run this code on my machine:\r\nhttps://github.com/huggingface/notebooks/blob/master/examples/token_classification.ipynb\r\nhttps://colab.research.google.com/drive/1-JIJlao4dI-Ilww_NnTc0rxtp-ymgDgM?usp=sharing\r\n\r\nI get:\r\nRuntimeError: CUDA out of memory. Tried to allocate 192.00 MiB (GPU 0; 1.96 GiB total capacity; 785.01 MiB already allocated; 111.25 MiB free; 832.00 MiB reserved in total by PyTorch)\r\n\r\n```\r\n+-----------------------------------------------------------------------------+\r\n| NVIDIA-SMI 460.32.03 Driver Version: 460.32.03 CUDA Version: 11.2 |\r\n|-------------------------------+----------------------+----------------------+\r\n| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |\r\n| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |\r\n| | | MIG M. |\r\n|===============================+======================+======================|\r\n| 0 GeForce MX250 Off | 00000000:02:00.0 Off | N/A |\r\n| N/A 56C P3 N/A / N/A | 1897MiB / 2002MiB | 0% Default |\r\n| | | N/A |\r\n+-------------------------------+----------------------+----------------------+\r\n```\r\n\r\n", "Hi ,I'm too facing this issue, any solution found on this ?", "I am having the same issue and it is recurring. Any solution?" ]

[]

[]

[ "sent PR https://github.com/huggingface/transformers/pull/6980" ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6977?src=pr&el=h1) Report\n> Merging [#6977](https://codecov.io/gh/huggingface/transformers/pull/6977?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/f72fe1f31aca235c7f675680832cc364efe4088e?el=desc) will **increase** coverage by `0.59%`.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/6977?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #6977 +/- ##\n==========================================\n+ Coverage 79.45% 80.04% +0.59% \n==========================================\n Files 161 161 \n Lines 30120 30120 \n==========================================\n+ Hits 23931 24109 +178 \n+ Misses 6189 6011 -178 \n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/6977?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/modeling\\_tf\\_flaubert.py](https://codecov.io/gh/huggingface/transformers/pull/6977/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9mbGF1YmVydC5weQ==) | `24.53% <0.00%> (-63.81%)` | :arrow_down: |\n| [src/transformers/modeling\\_tf\\_t5.py](https://codecov.io/gh/huggingface/transformers/pull/6977/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl90NS5weQ==) | `26.05% <0.00%> (-63.52%)` | :arrow_down: |\n| [src/transformers/modeling\\_lxmert.py](https://codecov.io/gh/huggingface/transformers/pull/6977/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19seG1lcnQucHk=) | `70.01% <0.00%> (-20.75%)` | :arrow_down: |\n| [src/transformers/configuration\\_t5.py](https://codecov.io/gh/huggingface/transformers/pull/6977/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9jb25maWd1cmF0aW9uX3Q1LnB5) | `85.71% <0.00%> (-10.72%)` | :arrow_down: |\n| [src/transformers/modeling\\_t5.py](https://codecov.io/gh/huggingface/transformers/pull/6977/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190NS5weQ==) | `71.02% <0.00%> (-5.69%)` | :arrow_down: |\n| [src/transformers/generation\\_tf\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6977/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9nZW5lcmF0aW9uX3RmX3V0aWxzLnB5) | `82.95% <0.00%> (-3.76%)` | :arrow_down: |\n| [src/transformers/modeling\\_tf\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6977/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl91dGlscy5weQ==) | `85.01% <0.00%> (-2.29%)` | :arrow_down: |\n| [src/transformers/data/data\\_collator.py](https://codecov.io/gh/huggingface/transformers/pull/6977/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9kYXRhL2RhdGFfY29sbGF0b3IucHk=) | `91.90% <0.00%> (-0.41%)` | :arrow_down: |\n| [src/transformers/file\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6977/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9maWxlX3V0aWxzLnB5) | `82.41% <0.00%> (-0.26%)` | :arrow_down: |\n| [src/transformers/modeling\\_bart.py](https://codecov.io/gh/huggingface/transformers/pull/6977/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19iYXJ0LnB5) | `94.41% <0.00%> (+0.16%)` | :arrow_up: |\n| ... and [11 more](https://codecov.io/gh/huggingface/transformers/pull/6977/diff?src=pr&el=tree-more) | |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/6977?src=pr&el=continue).\n> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta)\n> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`\n> Powered by [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6977?src=pr&el=footer). Last update [f72fe1f...6b88230](https://codecov.io/gh/huggingface/transformers/pull/6977?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n" ]

[]

[]

[ "Hey @ecekt, \r\n\r\nThanks for your issue!\r\nYes, you are correct ```\"unc-nlp/lxmert-base-uncased\"``` should only be imported with the `AutoModel` or `LxmertModel` class.\r\n\r\n@julien-c - the config looks correct I'm not sure why it states `AutoModelWithLMHeadModel` in the examlpe here: https://huggingface.co/unc-nlp/lxmert-base-uncased", "Hi @patrickvonplaten, thank you for the reply! I will explore its use with the LxmertModel import then.\r\n\r\nBest.", "This is now fixed on https://huggingface.co/unc-nlp/lxmert-base-uncased, thanks for the heads up" ]

[ { "id": 1838412367, "node_id": "MDU6TGFiZWwxODM4NDEyMzY3", "url": "https://api.github.com/repos/huggingface/transformers/labels/model%20card", "name": "model card", "color": "92d5f4", "default": false, "description": "Related to pretrained model cards" } ]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6975?src=pr&el=h1) Report\n> Merging [#6975](https://codecov.io/gh/huggingface/transformers/pull/6975?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/f72fe1f31aca235c7f675680832cc364efe4088e?el=desc) will **increase** coverage by `0.56%`.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/6975?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #6975 +/- ##\n==========================================\n+ Coverage 79.45% 80.01% +0.56% \n==========================================\n Files 161 161 \n Lines 30120 30120 \n==========================================\n+ Hits 23931 24102 +171 \n+ Misses 6189 6018 -171 \n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/6975?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/modeling\\_tf\\_electra.py](https://codecov.io/gh/huggingface/transformers/pull/6975/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9lbGVjdHJhLnB5) | `25.13% <0.00%> (-73.83%)` | :arrow_down: |\n| [src/transformers/modeling\\_tf\\_mobilebert.py](https://codecov.io/gh/huggingface/transformers/pull/6975/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9tb2JpbGViZXJ0LnB5) | `24.55% <0.00%> (-72.36%)` | :arrow_down: |\n| [src/transformers/modeling\\_marian.py](https://codecov.io/gh/huggingface/transformers/pull/6975/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19tYXJpYW4ucHk=) | `60.00% <0.00%> (-30.00%)` | :arrow_down: |\n| [src/transformers/activations.py](https://codecov.io/gh/huggingface/transformers/pull/6975/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9hY3RpdmF0aW9ucy5weQ==) | `85.00% <0.00%> (-5.00%)` | :arrow_down: |\n| [src/transformers/data/data\\_collator.py](https://codecov.io/gh/huggingface/transformers/pull/6975/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9kYXRhL2RhdGFfY29sbGF0b3IucHk=) | `91.90% <0.00%> (-0.41%)` | :arrow_down: |\n| [src/transformers/modeling\\_bart.py](https://codecov.io/gh/huggingface/transformers/pull/6975/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19iYXJ0LnB5) | `93.90% <0.00%> (-0.34%)` | :arrow_down: |\n| [src/transformers/modeling\\_tf\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6975/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl91dGlscy5weQ==) | `86.97% <0.00%> (-0.33%)` | :arrow_down: |\n| [src/transformers/file\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6975/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9maWxlX3V0aWxzLnB5) | `82.41% <0.00%> (-0.26%)` | :arrow_down: |\n| [src/transformers/tokenization\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6975/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy90b2tlbml6YXRpb25fdXRpbHMucHk=) | `89.84% <0.00%> (+0.39%)` | :arrow_up: |\n| [src/transformers/tokenization\\_bert.py](https://codecov.io/gh/huggingface/transformers/pull/6975/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy90b2tlbml6YXRpb25fYmVydC5weQ==) | `92.41% <0.00%> (+0.44%)` | :arrow_up: |\n| ... and [10 more](https://codecov.io/gh/huggingface/transformers/pull/6975/diff?src=pr&el=tree-more) | |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/6975?src=pr&el=continue).\n> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta)\n> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`\n> Powered by [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6975?src=pr&el=footer). Last update [f72fe1f...b65c486](https://codecov.io/gh/huggingface/transformers/pull/6975?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n", "Thanks for sharing this is great!" ]

[ { "id": 1838412367, "node_id": "MDU6TGFiZWwxODM4NDEyMzY3", "url": "https://api.github.com/repos/huggingface/transformers/labels/model%20card", "name": "model card", "color": "92d5f4", "default": false, "description": "Related to pretrained model cards" } ]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6974?src=pr&el=h1) Report\n> Merging [#6974](https://codecov.io/gh/huggingface/transformers/pull/6974?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/f72fe1f31aca235c7f675680832cc364efe4088e?el=desc) will **increase** coverage by `0.57%`.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/6974?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #6974 +/- ##\n==========================================\n+ Coverage 79.45% 80.02% +0.57% \n==========================================\n Files 161 161 \n Lines 30120 30120 \n==========================================\n+ Hits 23931 24105 +174 \n+ Misses 6189 6015 -174 \n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/6974?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/modeling\\_tf\\_electra.py](https://codecov.io/gh/huggingface/transformers/pull/6974/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9lbGVjdHJhLnB5) | `25.13% <0.00%> (-73.83%)` | :arrow_down: |\n| [src/transformers/modeling\\_tf\\_mobilebert.py](https://codecov.io/gh/huggingface/transformers/pull/6974/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9tb2JpbGViZXJ0LnB5) | `24.55% <0.00%> (-72.36%)` | :arrow_down: |\n| [src/transformers/modeling\\_marian.py](https://codecov.io/gh/huggingface/transformers/pull/6974/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19tYXJpYW4ucHk=) | `60.00% <0.00%> (-30.00%)` | :arrow_down: |\n| [src/transformers/activations.py](https://codecov.io/gh/huggingface/transformers/pull/6974/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9hY3RpdmF0aW9ucy5weQ==) | `85.00% <0.00%> (-5.00%)` | :arrow_down: |\n| [src/transformers/modeling\\_bart.py](https://codecov.io/gh/huggingface/transformers/pull/6974/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19iYXJ0LnB5) | `93.90% <0.00%> (-0.34%)` | :arrow_down: |\n| [src/transformers/modeling\\_tf\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6974/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl91dGlscy5weQ==) | `86.97% <0.00%> (-0.33%)` | :arrow_down: |\n| [src/transformers/generation\\_tf\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6974/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9nZW5lcmF0aW9uX3RmX3V0aWxzLnB5) | `86.46% <0.00%> (-0.26%)` | :arrow_down: |\n| [src/transformers/tokenization\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6974/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy90b2tlbml6YXRpb25fdXRpbHMucHk=) | `89.84% <0.00%> (+0.39%)` | :arrow_up: |\n| [src/transformers/data/data\\_collator.py](https://codecov.io/gh/huggingface/transformers/pull/6974/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9kYXRhL2RhdGFfY29sbGF0b3IucHk=) | `92.71% <0.00%> (+0.40%)` | :arrow_up: |\n| [src/transformers/tokenization\\_bert.py](https://codecov.io/gh/huggingface/transformers/pull/6974/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy90b2tlbml6YXRpb25fYmVydC5weQ==) | `92.41% <0.00%> (+0.44%)` | :arrow_up: |\n| ... and [10 more](https://codecov.io/gh/huggingface/transformers/pull/6974/diff?src=pr&el=tree-more) | |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/6974?src=pr&el=continue).\n> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta)\n> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`\n> Powered by [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6974?src=pr&el=footer). Last update [f72fe1f...efa8495](https://codecov.io/gh/huggingface/transformers/pull/6974?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n" ]

[]

[]

[ "I'm a bit indifferent to this change, but I'm ok with setting it to `12`" ]

[]

[ { "login": "patrickvonplaten", "id": 23423619, "node_id": "MDQ6VXNlcjIzNDIzNjE5", "avatar_url": "https://avatars.githubusercontent.com/u/23423619?v=4", "gravatar_id": "", "url": "https://api.github.com/users/patrickvonplaten", "html_url": "https://github.com/patrickvonplaten", "followers_url": "https://api.github.com/users/patrickvonplaten/followers", "following_url": "https://api.github.com/users/patrickvonplaten/following{/other_user}", "gists_url": "https://api.github.com/users/patrickvonplaten/gists{/gist_id}", "starred_url": "https://api.github.com/users/patrickvonplaten/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/patrickvonplaten/subscriptions", "organizations_url": "https://api.github.com/users/patrickvonplaten/orgs", "repos_url": "https://api.github.com/users/patrickvonplaten/repos", "events_url": "https://api.github.com/users/patrickvonplaten/events{/privacy}", "received_events_url": "https://api.github.com/users/patrickvonplaten/received_events", "type": "User", "site_admin": false } ]

[ "issue #6950 is also this problem.", "I see what you mean! From `3.1.0` onwards every configuration has a `config.chunk_size_feed_forward` parameter. So as far as I can see whenever a config is loaded (whether via `.from_pretrained()` or with `BertConfig(....)`, this parameter is part of the config...). Can you give me an example where this would not be the case? ", "@patrickvonplaten Thanks for your replay.\r\n\r\nYes, I load the pretrained model with my own code, which cause this problem.\r\nFor some users who only use part of structures/classes/others in Transfomers, is it necessary to have certain compatibility?", "> I see what you mean! From `3.1.0` onwards every configuration has a `config.chunk_size_feed_forward` parameter. So as far as I can see whenever a config is loaded (whether via `.from_pretrained()` or with `BertConfig(....)`, this parameter is part of the config...). Can you give me an example where this would not be the case?\r\n\r\nI saw the `chunk_size_feed_forward` in BertConfig() after opened this issue.🤪", "@patrickvonplaten \r\nIn the future, I will make more normative to call modules.\r\nThanks for your replay.\r\nI close this issue." ]

[ { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null }, { "id": 1843244711, "node_id": "MDU6TGFiZWwxODQzMjQ0NzEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/New%20model", "name": "New model", "color": "fbca04", "default": false, "description": "" } ]

[]

[ "Hi @kkissmart \r\nYes, distilbart models are available [here](https://huggingface.co/sshleifer/distilbart-cnn-6-6)", "@patil-suraj I meant a distill Bart ( a pretrain model) not a summarization model. Do I misunderstand the name?", "No, we don't have that.", "Ohh, AFAIK there is no pre-trained distilbart like distilbert. \r\nThere are two types of distillation \r\n1. No teacher distillation: which copies alternate layers from the pre-trained model and creates a small student model.\r\n2. With teacher distillation: enforce that the student and teacher produce similar encoder_outputs, logits, and hidden_states\r\n\r\nYou can easily create a student (no teacher) model using the scripts [here](https://github.com/huggingface/transformers/tree/master/examples/seq2seq#no-teacher-distillation) , you'll just need to use `bart-large` instead of `bart-large-cnn`.\r\n\r\npre-training distilbart is still not included, however you can train the large model on the down-stream task and then do with teacher distillation for a smaller distilled model.", "Also, consider asking such non-bug questions on the forum https://discuss.huggingface.co/ -:)", "Is there a distilBART base model that does not have pretaining weights?", "There is no distilbart-base model.\r\nThere are only distilled models fine-tuned on summarization tasks.", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n", "@sshleifer @patil-suraj \r\n```\r\nYou can easily create a student (no teacher) model using the scripts here , you'll just need to use bart-large instead of bart-large-cnn.\r\nhttps://github.com/huggingface/transformers/tree/master/examples/seq2seq#no-teacher-distillation or\r\nhttps://github.com/huggingface/transformers/tree/master/examples/legacy/seq2seq#no-teacher-distillation ?\r\n```\r\nI want to make a DistilBART model from my japanese BART-large mode. but, No script in this link. Has the script been kept private? I want to see the script.\r\n", "Hi @hisashi-ito \r\n\r\nThe seq2seq distillation scripts are now moved under `examples/research_projects/seq2seq-distillation` directory. You can find them here.\r\nhttps://github.com/huggingface/transformers/tree/master/examples/research_projects/seq2seq-distillation", "Hi @patil-suraj\r\nThank you for teaching !! " ]

[ { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "Yeah the latest release of tokenizers is 0.8.1 but the current code still references the rc2:\r\nhttps://github.com/huggingface/transformers/blob/master/setup.py#L113", "I was able to get this working by installing tokenizers 0.8.1. I then\ninstalled transformers 3.1.0 without dependencies using --no-dependencies\nflag (had to install a few other dependencies manually).\n\nYour mileage may vary.\n\nOn Fri, Sep 11, 2020 at 23:21 Fabrizio Milo <[email protected]>\nwrote:\n\n>\n>\n> Yeah the latest release of tokenizers is 0.8.1 but the current code still\n> references the rc2:\n>\n>\n> https://github.com/huggingface/transformers/blob/master/setup.py#L113\n>\n>\n>\n>\n> —\n> You are receiving this because you are subscribed to this thread.\n> Reply to this email directly, view it on GitHub\n> <https://github.com/huggingface/transformers/issues/6970#issuecomment-691394983>,\n> or unsubscribe\n> <https://github.com/notifications/unsubscribe-auth/ACMFA4WL57SIWJT5U2LIBHDSFLSLZANCNFSM4Q3OYZ6Q>\n> .\n>\n>\n>\n", "I simply installed the transformer 3.0.0 version until they fix this problem. \r\n`python3 -m pip install transformers==3.0.0`", "> I simply installed the transformer 3.0.0 version until they fix this problem.\r\n> `python3 -m pip install transformers==3.0.0`\r\n\r\nI need version 3.1.0 for the latest 0-shot pipeline. But the following fixed the problem that @alexuadler mentioned:\r\n\r\npip3 install tokenizers==\"0.8.1\"\r\npip3 install transformers==\"3.1.0\" --no-dependencies", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n" ]

[ { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[ { "login": "jplu", "id": 959590, "node_id": "MDQ6VXNlcjk1OTU5MA==", "avatar_url": "https://avatars.githubusercontent.com/u/959590?v=4", "gravatar_id": "", "url": "https://api.github.com/users/jplu", "html_url": "https://github.com/jplu", "followers_url": "https://api.github.com/users/jplu/followers", "following_url": "https://api.github.com/users/jplu/following{/other_user}", "gists_url": "https://api.github.com/users/jplu/gists{/gist_id}", "starred_url": "https://api.github.com/users/jplu/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/jplu/subscriptions", "organizations_url": "https://api.github.com/users/jplu/orgs", "repos_url": "https://api.github.com/users/jplu/repos", "events_url": "https://api.github.com/users/jplu/events{/privacy}", "received_events_url": "https://api.github.com/users/jplu/received_events", "type": "User", "site_admin": false } ]

[ "Thanks @chiapas! Indeed if the batch size becomes lower for the last step we divide by a wrong number. I will investigate this to better handle this edge case.", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n" ]

[]

[ { "login": "jplu", "id": 959590, "node_id": "MDQ6VXNlcjk1OTU5MA==", "avatar_url": "https://avatars.githubusercontent.com/u/959590?v=4", "gravatar_id": "", "url": "https://api.github.com/users/jplu", "html_url": "https://github.com/jplu", "followers_url": "https://api.github.com/users/jplu/followers", "following_url": "https://api.github.com/users/jplu/following{/other_user}", "gists_url": "https://api.github.com/users/jplu/gists{/gist_id}", "starred_url": "https://api.github.com/users/jplu/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/jplu/subscriptions", "organizations_url": "https://api.github.com/users/jplu/orgs", "repos_url": "https://api.github.com/users/jplu/repos", "events_url": "https://api.github.com/users/jplu/events{/privacy}", "received_events_url": "https://api.github.com/users/jplu/received_events", "type": "User", "site_admin": false } ]

[ "Hello @chiapas!\r\n\r\nThanks for having investigating this! Have you checked the way we compute the loss for Token classification? Right here https://github.com/huggingface/transformers/blob/master/src/transformers/modeling_tf_utils.py#L161\r\n\r\nWe are already ignoring all the tokens that have -100 as label. So if there is an issue it might come from somewhere else.\r\n", "Hi, @jplu,\r\n\r\nYes, I checked that. But the issue in this bug report is not about ignoring -100 or not. The problem is that, the loss is calculated from the per example losses, then divided by `total_train_batch_size`. However, for token level tasks, it should be divided by `the number of actual tokens (i.e. tokens not ignored) in the global batch (i.e. the batch that having the size total_train_batch_size)`.", "I don't get what you mean by the number of actual token?\r\n\r\nYou mean the number of batches that contain actual tokens, no? In your example, just 1? ", "If you want a support for the argument about the denominator I claimed, we can look at the pytorch implementation in [DistilBert](https://github.com/huggingface/transformers/blob/master/src/transformers/modeling_distilbert.py#L820) about the loss for token classification, you will see\r\n\r\n loss_fct = CrossEntropyLoss() \r\n\r\nand \r\n\r\n loss = loss_fct(active_logits, active_labels)\r\n\r\nAnd from [torch's doc](https://pytorch.org/docs/stable/generated/torch.nn.CrossEntropyLoss.html#crossentropyloss), the default reduction is `mean`. So this corresponds to the per example losses (-100 ignored) divided by the number of actual tokens (i.e. -100 ignored again).\r\n\r\n\r\n\r\n", "> \r\n> \r\n> I don't get what you mean by the number of actual token?\r\n> \r\n> You mean the number of batches that contain actual tokens, no? In your example, just 1?\r\n\r\nFor any global batch (that has size `total_train_batch_size`), after distributed to replicas, we compute the per example losses on the smaller batches (where the tokens with label -100 are ignored). Then in the current implementation, this per example losses are divided by `total_train_batch_size`.\r\n\r\nBy the actual tokens, it means `the tokens in a global batch with labels != -100`. And what I said is that, the per example losses should be divided by the number of tokens with label != -100 in that global batch.\r\n\r\nThis number of `actual` tokens will be varied for different global batch however.\r\n\r\nIn the example in the code snippet, they are only `4` actual tokens. By when `n_empty_string = 9`, the current implementation divided the per example losses by ` 10 (1 + 9 dummy sentences)`. \r\n", "So basically what you propose is to return a mean reduction in the `call` function instead of the effective per example loss? Something like to replace:\r\n\r\n```\r\nloss = None if labels is None else self.compute_loss(labels, logits)\r\n```\r\nBy\r\n```\r\nloss = None if labels is None else tf.reduce_mean(self.compute_loss(labels, logits))\r\n```\r\n\r\nAnd then dividing the result per the number of replica:\r\n```\r\nper_example_loss, _ = self.run_model(features, labels, True)\r\nscaled_loss = per_example_loss / self.args.n_replicas\r\n```\r\n\r\nOtherwise, please show me you notebook because I still don't get it.", "@jplu ,\r\n\r\nI will explain a bit more and also show my notebook later. But no, I am not suggesting using\r\n\r\n tf.reduce_mean(self.compute_loss(labels, logits)) # If we do so, the average occurs on the small batch in each replica.\r\n\r\nI mentioned the pytorch version just to show that the loss should be averaged over the tokens, not over the sentences in the batch. However, the average shouldn't be over the small batches received by each replica, it should be over the global batches.", "@jplu \r\n\r\nIf you want to look the code directly, here is my (kaggle) notebook\r\n\r\n[Masked, My Dear Watson - MLM with TPU](https://www.kaggle.com/yihdarshieh/masked-my-dear-watson-mlm-with-tpu#MLM-loss-calculation). Please check `def mlm_fine_tune_step(batch):` just below that markdown cell, which has\r\n\r\n loss_mlm = loss_fn(\r\n labels_at_masked_tokens,\r\n logits_at_masked_tokens\r\n )\r\n\r\n # divide the number of masked tokens in the global batch, i.e. the whole batch that is distributed to different replicas.\r\n loss_mlm = loss_mlm / tf.cast(nb_tokens_masked[0], dtype=tf.float32)\r\n\r\nIf you prefer, I can work on this and make a PR. But I think it is better for us to agree that the loss calculation should be corrected. So I will try to explain:\r\n\r\n1. For token level tasks, the loss value is the per example (and here, example = tokens) losses in a batch, divided by the number of tokens in that batch.\r\n2. If we have tokens being ignored for loss calculation, the denominator above becomes the number of tokens not ignored in that batch.\r\n3. By `batch`, it is the whole set used for 1 parameter update by gradients - which is called a `global batch`.\r\n4. Since we use distributed strategy, and optionally gradient accumulation, while `training_step()` processes a batch, it is a small batch (i.e. a batch for `only 1 gradient accumulation step` on a `single replica`). However, the denominator in step `1.` or `2.` should be the `the number of token, not being ignored, in a global batch`, despite the per example losses are still based on the small batch received by a replica.\r\n5. Since gradient accumulation will add the gradients, and distributed strategy will sync across replicas by summing the gradients before applying gradients, the above steps will give us the `averaged losses over the tokens (not ignored) in a global batch`.\r\n\r\nHope it make things a bit clear.", "OK now with an example and the explanation I got it. Thank you very much!\n\nI prefer you do a PR and then you get the credit of this fix :) And if you can tag me as reviewer I will be able to help you if needed, as there is certainly a nicer way to do. Maybe with a class field?\n\nThanks again, waiting your PR ^^", "I am not able to assign reviewer, since I am not a collaborator yet on transformer repository.", "Nice! I will review that carefuly tomorrow. I have assigned 2 other persons and myself as reviewer.", "Hi, sorry, I created the pull request as draft. It is not ready to review at all. I will let you know when it is ready.", "No problem! Take the time you need and let me know." ]

[]

[]

[]

[]

[]

[ "Hi @hichiaty ,\r\n\r\nin that case I would try the CamemBERT Tokenizer or the one from XLM-RoBERTa (last one contains some hacky workaround to align fairseq vocab with SPM...) :)", "hey @stefan-it thanks for the advice! I am still slightly confused though, I loaded my spm model with CamemBERT but for some reason it still doesn't match the tokens from fairseq's roberta.encode.\r\n\r\nUsing Fairseq's encode I get:\r\n\r\n```\r\nroberta.encode('HELLO')\r\n>tensor([0, 7, 4, 6, 2]) \r\n```\r\nWith CamemBERT I get:\r\n\r\n```\r\ntokenizer('HELLO')\r\n>{'input_ids': [5, 45, 36, 3863, 3595, 3595, 19, 5], 'attention_mask': [1, 1, 1, 1, 1, 1, 1, 1]}\r\n```\r\nI'm just trying to figure out how fairseq works at this point because I don't even pass the spm model to it but a dict.txt file", "Solved, created a custom tokenizer (based on camembert) that uses spm to encode as pieces, then fairseq's dict.txt to get ids." ]

[]

[]

[ "Well, I wrote a little script to generate the long commands that are required now - perhaps it'd be useful to someone:\r\n```\r\nperl -le 'for $f (@ARGV) { print qq[yes Y | transformers-cli upload $_/$f --filename $_/$f] for map { \"fsmt-wmt19-$_\" } (\"en-ru\", \"ru-en\", \"de-en\", \"en-de\")}' vocab-src.json vocab-tgt.json tokenizer_config.json\r\n```\r\n\r\ngenerated:\r\n```\r\nyes Y | transformers-cli upload fsmt-wmt19-en-ru/vocab-src.json --filename fsmt-wmt19-en-ru/vocab-src.json\r\nyes Y | transformers-cli upload fsmt-wmt19-ru-en/vocab-src.json --filename fsmt-wmt19-ru-en/vocab-src.json\r\nyes Y | transformers-cli upload fsmt-wmt19-de-en/vocab-src.json --filename fsmt-wmt19-de-en/vocab-src.json\r\nyes Y | transformers-cli upload fsmt-wmt19-en-de/vocab-src.json --filename fsmt-wmt19-en-de/vocab-src.json\r\nyes Y | transformers-cli upload fsmt-wmt19-en-ru/vocab-tgt.json --filename fsmt-wmt19-en-ru/vocab-tgt.json\r\nyes Y | transformers-cli upload fsmt-wmt19-ru-en/vocab-tgt.json --filename fsmt-wmt19-ru-en/vocab-tgt.json\r\nyes Y | transformers-cli upload fsmt-wmt19-de-en/vocab-tgt.json --filename fsmt-wmt19-de-en/vocab-tgt.json\r\nyes Y | transformers-cli upload fsmt-wmt19-en-de/vocab-tgt.json --filename fsmt-wmt19-en-de/vocab-tgt.json\r\nyes Y | transformers-cli upload fsmt-wmt19-en-ru/tokenizer_config.json --filename fsmt-wmt19-en-ru/tokenizer_config.json\r\nyes Y | transformers-cli upload fsmt-wmt19-ru-en/tokenizer_config.json --filename fsmt-wmt19-ru-en/tokenizer_config.json\r\nyes Y | transformers-cli upload fsmt-wmt19-de-en/tokenizer_config.json --filename fsmt-wmt19-de-en/tokenizer_config.json\r\nyes Y | transformers-cli upload fsmt-wmt19-en-de/tokenizer_config.json --filename fsmt-wmt19-en-de/tokenizer_config.json\r\n```\r\n\r\nAs I have an easy workaround that works well, unless others feel the suggested improvements in the OP would be useful, I'd be happy to close this ticket.", "pinging @julien-c ", "@julien-c, Should this be closed or fixed? Thanks.", "Closing as we are migrating to a new system anyways (more info soon)" ]

[ { "id": 1838412367, "node_id": "MDU6TGFiZWwxODM4NDEyMzY3", "url": "https://api.github.com/repos/huggingface/transformers/labels/model%20card", "name": "model card", "color": "92d5f4", "default": false, "description": "Related to pretrained model cards" } ]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6964?src=pr&el=h1) Report\n> Merging [#6964](https://codecov.io/gh/huggingface/transformers/pull/6964?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/d31031f603043281d4fbac6cbdcfb6497fd500ab?el=desc) will **decrease** coverage by `4.23%`.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/6964?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #6964 +/- ##\n==========================================\n- Coverage 80.03% 75.80% -4.24% \n==========================================\n Files 161 161 \n Lines 30120 30120 \n==========================================\n- Hits 24108 22833 -1275 \n- Misses 6012 7287 +1275 \n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/6964?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/configuration\\_lxmert.py](https://codecov.io/gh/huggingface/transformers/pull/6964/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9jb25maWd1cmF0aW9uX2x4bWVydC5weQ==) | `20.00% <0.00%> (-80.00%)` | :arrow_down: |\n| [src/transformers/modeling\\_tf\\_lxmert.py](https://codecov.io/gh/huggingface/transformers/pull/6964/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9seG1lcnQucHk=) | `22.49% <0.00%> (-71.63%)` | :arrow_down: |\n| [src/transformers/pipelines.py](https://codecov.io/gh/huggingface/transformers/pull/6964/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9waXBlbGluZXMucHk=) | `25.30% <0.00%> (-55.16%)` | :arrow_down: |\n| [src/transformers/optimization.py](https://codecov.io/gh/huggingface/transformers/pull/6964/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9vcHRpbWl6YXRpb24ucHk=) | `34.28% <0.00%> (-48.00%)` | :arrow_down: |\n| [src/transformers/modeling\\_lxmert.py](https://codecov.io/gh/huggingface/transformers/pull/6964/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19seG1lcnQucHk=) | `23.50% <0.00%> (-46.52%)` | :arrow_down: |\n| [src/transformers/optimization\\_tf.py](https://codecov.io/gh/huggingface/transformers/pull/6964/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9vcHRpbWl6YXRpb25fdGYucHk=) | `33.33% <0.00%> (-24.33%)` | :arrow_down: |\n| [src/transformers/modeling\\_roberta.py](https://codecov.io/gh/huggingface/transformers/pull/6964/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19yb2JlcnRhLnB5) | `77.00% <0.00%> (-20.08%)` | :arrow_down: |\n| [src/transformers/modeling\\_tf\\_auto.py](https://codecov.io/gh/huggingface/transformers/pull/6964/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9hdXRvLnB5) | `48.79% <0.00%> (-18.08%)` | :arrow_down: |\n| [src/transformers/data/processors/squad.py](https://codecov.io/gh/huggingface/transformers/pull/6964/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9kYXRhL3Byb2Nlc3NvcnMvc3F1YWQucHk=) | `13.76% <0.00%> (-14.38%)` | :arrow_down: |\n| [src/transformers/modeling\\_auto.py](https://codecov.io/gh/huggingface/transformers/pull/6964/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19hdXRvLnB5) | `64.57% <0.00%> (-14.29%)` | :arrow_down: |\n| ... and [15 more](https://codecov.io/gh/huggingface/transformers/pull/6964/diff?src=pr&el=tree-more) | |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/6964?src=pr&el=continue).\n> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta)\n> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`\n> Powered by [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6964?src=pr&el=footer). Last update [d31031f...4ff71ec](https://codecov.io/gh/huggingface/transformers/pull/6964?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n", "This is great! Added some custom prompts for the inference widget. Thanks for sharing!" ]

[]

[]

[ "Hey @blawok, \r\n\r\nwhich model are you referring to exactly? This longformer model: https://s3.amazonaws.com/models.huggingface.co/bert/allenai/longformer-base-4096/config.json has vocab_size set to 50265 ", "Thanks for the answer @patrickvonplaten :) \r\n\r\nUsing the code from the docs (https://huggingface.co/transformers/model_doc/longformer.html#longformerconfig):\r\n\r\n```python\r\nfrom transformers import LongformerConfig, LongformerModel\r\n# Initializing a Longformer configuration\r\nconfiguration = LongformerConfig()\r\n# Initializing a model from the configuration\r\nmodel = LongformerModel(configuration)\r\n# Accessing the model configuration\r\nconfiguration = model.config\r\nprint(configuration)\r\n```\r\n\r\nI am getting this output:\r\n\r\n```\r\nLongformerConfig {\r\n \"attention_probs_dropout_prob\": 0.1,\r\n \"attention_window\": [\r\n 512,\r\n 512,\r\n 512,\r\n 512,\r\n 512,\r\n 512,\r\n 512,\r\n 512,\r\n 512,\r\n 512,\r\n 512,\r\n 512\r\n ],\r\n \"bos_token_id\": 0,\r\n \"eos_token_id\": 2,\r\n \"gradient_checkpointing\": false,\r\n \"hidden_act\": \"gelu\",\r\n \"hidden_dropout_prob\": 0.1,\r\n \"hidden_size\": 768,\r\n \"initializer_range\": 0.02,\r\n \"intermediate_size\": 3072,\r\n \"layer_norm_eps\": 1e-12,\r\n \"max_position_embeddings\": 512,\r\n \"model_type\": \"longformer\",\r\n \"num_attention_heads\": 12,\r\n \"num_hidden_layers\": 12,\r\n \"pad_token_id\": 1,\r\n \"sep_token_id\": 2,\r\n \"type_vocab_size\": 2,\r\n \"vocab_size\": 30522\r\n}\r\n```\r\n\r\nIt resulted in error while trying to train:\r\n```python\r\nconfig = LongformerConfig()\r\nmodel = TFLongformerModel.from_pretrained('allenai/longformer-base-4096', config=config) \r\n```\r\nLongformerTokenizer correctly used the 50265 vocab, but the model expected it to be 30522. However, I am not getting this error when I am not specifying the configuration.\r\n\r\nExample to reproduce:\r\n\r\n```python\r\nfrom transformers import LongformerTokenizer, TFLongformerModel, LongformerConfig\r\n\r\ntokenizer = LongformerTokenizer.from_pretrained('allenai/longformer-base-4096')\r\ninput = tokenizer('Hello world')\r\n\r\nconfig = LongformerConfig()\r\nmodel = TFLongformerModel.from_pretrained('allenai/longformer-base-4096',\r\n config=config) \r\n```\r\n\r\nError I am getting with the code above:\r\n\r\n```python\r\n---------------------------------------------------------------------------\r\n\r\nValueError Traceback (most recent call last)\r\n\r\n<ipython-input-4-80be6efb2343> in <module>()\r\n 6 config = LongformerConfig()\r\n 7 model = TFLongformerModel.from_pretrained('allenai/longformer-base-4096',\r\n----> 8 config=config) \r\n 9 print(model(input))\r\n\r\n2 frames\r\n\r\n/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/saving/hdf5_format.py in load_weights_from_hdf5_group_by_name(f, layers, skip_mismatch)\r\n 784 symbolic_weights[i])) +\r\n 785 ', but the saved weight has shape ' +\r\n--> 786 str(weight_values[i].shape) + '.')\r\n 787 \r\n 788 else:\r\n\r\nValueError: Layer #0 (named \"longformer\"), weight <tf.Variable 'tf_longformer_model/longformer/embeddings/word_embeddings/weight:0' shape=(30522, 768) dtype=float32, numpy=\r\narray([[-0.01930627, -0.01715518, -0.00557071, ..., -0.01202598,\r\n 0.01007012, -0.00184635],\r\n [ 0.00633493, 0.00123013, -0.0134872 , ..., -0.01304915,\r\n -0.00157391, 0.00082429],\r\n [-0.01581489, 0.01005882, -0.01242067, ..., 0.00555116,\r\n 0.02116241, 0.03123646],\r\n ...,\r\n [-0.01625618, 0.01438301, 0.03368756, ..., -0.02742909,\r\n 0.00300512, 0.00728624],\r\n [-0.0078434 , -0.01735217, 0.00178284, ..., -0.01191203,\r\n -0.01451435, 0.03031485],\r\n [-0.00814894, 0.01228636, 0.00573935, ..., 0.01143655,\r\n -0.00131886, -0.03910364]], dtype=float32)> has shape (30522, 768), but the saved weight has shape (50265, 768).\r\n```", "Hi @blawok, you're initializing a configuration using the default parameters, which may not be the same as the checkpoint's parameters you're initializing (it's not the case here).\r\n\r\nYou should initialize the configuration from the checkpoint here too:\r\n\r\n```py\r\nconfig = LongformerConfig.from_pretrained('allenai/longformer-base-4096')\r\nmodel = TFLongformerModel.from_pretrained('allenai/longformer-base-4096', config=config)\r\n```", "Great, thank you for the explanation @LysandreJik :) \r\n\r\nI am closing this issue." ]

[ { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "Working with special tokens attributes also became slower:\r\n```python\r\nimport timeit\r\nimport numpy as np\r\nfrom transformers import __version__ as trans_version\r\nfrom transformers import (\r\n CTRLTokenizer,\r\n GPT2Tokenizer,\r\n RobertaTokenizer,\r\n XLMTokenizer,\r\n XLNetTokenizer\r\n)\r\n\r\nTok_CLASSES = {\r\n \"gpt2\": GPT2Tokenizer,\r\n \"ctrl\": CTRLTokenizer,\r\n \"roberta-large\": RobertaTokenizer,\r\n \"xlnet-large-cased\": XLNetTokenizer,\r\n \"xlm-mlm-100-1280\": XLMTokenizer,\r\n}\r\n\r\nprint(trans_version)\r\nrounding = 4\r\nfor k, v in Tok_CLASSES.items():\r\n tokenizer_class = v\r\n tokenizer = tokenizer_class.from_pretrained(k, verbose=False)\r\n\r\n print(tokenizer.__class__)\r\n # print(tokenizer.encode(text))\r\n if tokenizer.eos_token is not None:\r\n r_eos_token = timeit.repeat(stmt=\"tokenizer.eos_token\", repeat=100, number=100000, globals=globals())\r\n r_eos_token_id = timeit.repeat(stmt=\"tokenizer.eos_token_id\", repeat=100, number=100000, globals=globals())\r\n print('Get eos_token, time taken (mean ± 3std):',\r\n str(np.round(np.mean(r_eos_token), rounding)) + '±' + str(np.round(3 * np.std(r_eos_token), rounding)))\r\n print('Get eos_token_id, time taken (mean ± 3std):',\r\n str(np.round(np.mean(r_eos_token_id), rounding)) + '±' + str(\r\n np.round(3 * np.std(r_eos_token_id), rounding)))\r\n if tokenizer.bos_token is not None:\r\n r_bos_token = timeit.repeat(stmt=\"tokenizer.bos_token\", repeat=100, number=100000, globals=globals())\r\n r_bos_token_id = timeit.repeat(stmt=\"tokenizer.bos_token_id\", repeat=100, number=100000, globals=globals())\r\n print('Get bos_token, time taken (mean ± 3std):',\r\n str(np.round(np.mean(r_bos_token), rounding)) + '±' + str(np.round(3 * np.std(r_bos_token), rounding)))\r\n print('Get bos_token_id, time taken (mean ± 3std):',\r\n str(np.round(np.mean(r_bos_token_id), rounding)) + '±' + str(\r\n np.round(3 * np.std(r_bos_token_id), rounding)))\r\n if tokenizer.unk_token is not None:\r\n r_unk_token = timeit.repeat(stmt=\"tokenizer.unk_token\", repeat=100, number=100000, globals=globals())\r\n r_unk_token_id = timeit.repeat(stmt=\"tokenizer.unk_token_id\", repeat=100, number=100000, globals=globals())\r\n print('Get unk_token, time taken (mean ± 3std):',\r\n str(np.round(np.mean(r_unk_token), rounding)) + '±' + str(np.round(3 * np.std(r_unk_token), rounding)))\r\n print('Get unk_token_id, time taken (mean ± 3std):',\r\n str(np.round(np.mean(r_unk_token_id), rounding)) + '±' + str(\r\n np.round(3 * np.std(r_unk_token_id), rounding)))\r\n```\r\ngives for 2.8.0\r\n```\r\n2.8.0\r\n<class 'transformers.tokenization_gpt2.GPT2Tokenizer'>\r\nGet eos_token, time taken (mean ± 3std): 0.0104±0.0003\r\nGet eos_token_id, time taken (mean ± 3std): 0.0715±0.0059\r\nGet bos_token, time taken (mean ± 3std): 0.0101±0.0021\r\nGet bos_token_id, time taken (mean ± 3std): 0.0687±0.0155\r\nGet unk_token, time taken (mean ± 3std): 0.0101±0.0001\r\nGet unk_token_id, time taken (mean ± 3std): 0.0632±0.0004\r\n<class 'transformers.tokenization_ctrl.CTRLTokenizer'>\r\nGet unk_token, time taken (mean ± 3std): 0.0098±0.0002\r\nGet unk_token_id, time taken (mean ± 3std): 0.0639±0.0008\r\n<class 'transformers.tokenization_roberta.RobertaTokenizer'>\r\nGet eos_token, time taken (mean ± 3std): 0.0099±0.0003\r\nGet eos_token_id, time taken (mean ± 3std): 0.0644±0.0017\r\nGet bos_token, time taken (mean ± 3std): 0.0093±0.0001\r\nGet bos_token_id, time taken (mean ± 3std): 0.064±0.0003\r\nGet unk_token, time taken (mean ± 3std): 0.0094±0.0002\r\nGet unk_token_id, time taken (mean ± 3std): 0.0727±0.0039\r\n<class 'transformers.tokenization_xlnet.XLNetTokenizer'>\r\nGet eos_token, time taken (mean ± 3std): 0.01±0.0002\r\nGet eos_token_id, time taken (mean ± 3std): 0.0848±0.0021\r\nGet bos_token, time taken (mean ± 3std): 0.0104±0.0003\r\nGet bos_token_id, time taken (mean ± 3std): 0.0847±0.0072\r\nGet unk_token, time taken (mean ± 3std): 0.0097±0.0001\r\nGet unk_token_id, time taken (mean ± 3std): 0.084±0.0007\r\n<class 'transformers.tokenization_xlm.XLMTokenizer'>\r\nGet bos_token, time taken (mean ± 3std): 0.01±0.0001\r\nGet bos_token_id, time taken (mean ± 3std): 0.0646±0.001\r\nGet unk_token, time taken (mean ± 3std): 0.0098±0.0001\r\nGet unk_token_id, time taken (mean ± 3std): 0.0639±0.0004\r\n```\r\nand for 3.1.0 (2x...4x slower):\r\n```\r\n3.1.0\r\n<class 'transformers.tokenization_gpt2.GPT2Tokenizer'>\r\nGet eos_token, time taken (mean ± 3std): 0.0422±0.0004\r\nGet eos_token_id, time taken (mean ± 3std): 0.1465±0.0015\r\nGet bos_token, time taken (mean ± 3std): 0.0418±0.0005\r\nGet bos_token_id, time taken (mean ± 3std): 0.1453±0.0009\r\nGet unk_token, time taken (mean ± 3std): 0.0417±0.0002\r\nGet unk_token_id, time taken (mean ± 3std): 0.1519±0.0186\r\n<class 'transformers.tokenization_ctrl.CTRLTokenizer'>\r\nGet unk_token, time taken (mean ± 3std): 0.0163±0.0003\r\nGet unk_token_id, time taken (mean ± 3std): 0.0821±0.0006\r\n<class 'transformers.tokenization_roberta.RobertaTokenizer'>\r\nGet eos_token, time taken (mean ± 3std): 0.0419±0.0029\r\nGet eos_token_id, time taken (mean ± 3std): 0.1462±0.004\r\nGet bos_token, time taken (mean ± 3std): 0.042±0.0004\r\nGet bos_token_id, time taken (mean ± 3std): 0.1544±0.0311\r\nGet unk_token, time taken (mean ± 3std): 0.0449±0.0016\r\nGet unk_token_id, time taken (mean ± 3std): 0.1511±0.006\r\n<class 'transformers.tokenization_xlnet.XLNetTokenizer'>\r\nGet eos_token, time taken (mean ± 3std): 0.0165±0.0004\r\nGet eos_token_id, time taken (mean ± 3std): 0.0918±0.0043\r\nGet bos_token, time taken (mean ± 3std): 0.0164±0.0003\r\nGet bos_token_id, time taken (mean ± 3std): 0.0931±0.0034\r\nGet unk_token, time taken (mean ± 3std): 0.0166±0.0002\r\nGet unk_token_id, time taken (mean ± 3std): 0.0933±0.0004\r\n<class 'transformers.tokenization_xlm.XLMTokenizer'>\r\nGet bos_token, time taken (mean ± 3std): 0.0162±0.0003\r\nGet bos_token_id, time taken (mean ± 3std): 0.0801±0.0008\r\nGet unk_token, time taken (mean ± 3std): 0.016±0.0003\r\nGet unk_token_id, time taken (mean ± 3std): 0.0827±0.0002\r\n```", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n" ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6961?src=pr&el=h1) Report\n> Merging [#6961](https://codecov.io/gh/huggingface/transformers/pull/6961?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/56742e9f610231d7b28fe2387770dc56014b79de?el=desc) will **increase** coverage by `0.65%`.\n> The diff coverage is `96.42%`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/6961?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #6961 +/- ##\n==========================================\n+ Coverage 80.00% 80.65% +0.65% \n==========================================\n Files 161 161 \n Lines 30120 30147 +27 \n==========================================\n+ Hits 24097 24315 +218 \n+ Misses 6023 5832 -191 \n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/6961?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/testing\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6961/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy90ZXN0aW5nX3V0aWxzLnB5) | `69.17% <94.11%> (+3.28%)` | :arrow_up: |\n| [src/transformers/utils/logging.py](https://codecov.io/gh/huggingface/transformers/pull/6961/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy91dGlscy9sb2dnaW5nLnB5) | `85.89% <100.00%> (+10.89%)` | :arrow_up: |\n| [src/transformers/modeling\\_tf\\_flaubert.py](https://codecov.io/gh/huggingface/transformers/pull/6961/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9mbGF1YmVydC5weQ==) | `24.53% <0.00%> (-63.81%)` | :arrow_down: |\n| [src/transformers/modeling\\_tf\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6961/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl91dGlscy5weQ==) | `87.29% <0.00%> (+0.32%)` | :arrow_up: |\n| [src/transformers/data/data\\_collator.py](https://codecov.io/gh/huggingface/transformers/pull/6961/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9kYXRhL2RhdGFfY29sbGF0b3IucHk=) | `92.71% <0.00%> (+0.80%)` | :arrow_up: |\n| [src/transformers/generation\\_tf\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6961/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9nZW5lcmF0aW9uX3RmX3V0aWxzLnB5) | `86.46% <0.00%> (+1.25%)` | :arrow_up: |\n| [src/transformers/modeling\\_tf\\_electra.py](https://codecov.io/gh/huggingface/transformers/pull/6961/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9lbGVjdHJhLnB5) | `98.95% <0.00%> (+73.82%)` | :arrow_up: |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/6961?src=pr&el=continue).\n> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta)\n> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`\n> Powered by [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6961?src=pr&el=footer). Last update [56742e9...3c194b7](https://codecov.io/gh/huggingface/transformers/pull/6961?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n", "All the requested changes have been done.", "Great, thanks @stas00 " ]

[ { "id": 1838412367, "node_id": "MDU6TGFiZWwxODM4NDEyMzY3", "url": "https://api.github.com/repos/huggingface/transformers/labels/model%20card", "name": "model card", "color": "92d5f4", "default": false, "description": "Related to pretrained model cards" } ]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6960?src=pr&el=h1) Report\n> Merging [#6960](https://codecov.io/gh/huggingface/transformers/pull/6960?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/4ebb52afdb4dc4bcd599e7cb503763e5d4afc962?el=desc) will **increase** coverage by `1.50%`.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/6960?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #6960 +/- ##\n==========================================\n+ Coverage 77.81% 79.31% +1.50% \n==========================================\n Files 157 157 \n Lines 28853 28853 \n==========================================\n+ Hits 22452 22885 +433 \n+ Misses 6401 5968 -433 \n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/6960?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/modeling\\_tf\\_flaubert.py](https://codecov.io/gh/huggingface/transformers/pull/6960/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9mbGF1YmVydC5weQ==) | `24.53% <0.00%> (-63.81%)` | :arrow_down: |\n| [src/transformers/trainer.py](https://codecov.io/gh/huggingface/transformers/pull/6960/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy90cmFpbmVyLnB5) | `13.54% <0.00%> (-41.13%)` | :arrow_down: |\n| [src/transformers/training\\_args.py](https://codecov.io/gh/huggingface/transformers/pull/6960/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy90cmFpbmluZ19hcmdzLnB5) | `66.66% <0.00%> (-25.00%)` | :arrow_down: |\n| [src/transformers/trainer\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6960/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy90cmFpbmVyX3V0aWxzLnB5) | `59.57% <0.00%> (-19.15%)` | :arrow_down: |\n| [src/transformers/modeling\\_t5.py](https://codecov.io/gh/huggingface/transformers/pull/6960/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190NS5weQ==) | `76.70% <0.00%> (-6.07%)` | :arrow_down: |\n| [src/transformers/data/datasets/glue.py](https://codecov.io/gh/huggingface/transformers/pull/6960/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9kYXRhL2RhdGFzZXRzL2dsdWUucHk=) | `83.58% <0.00%> (-2.99%)` | :arrow_down: |\n| [src/transformers/data/data\\_collator.py](https://codecov.io/gh/huggingface/transformers/pull/6960/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9kYXRhL2RhdGFfY29sbGF0b3IucHk=) | `91.90% <0.00%> (-0.81%)` | :arrow_down: |\n| [src/transformers/modeling\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6960/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ191dGlscy5weQ==) | `87.50% <0.00%> (-0.37%)` | :arrow_down: |\n| [src/transformers/tokenization\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6960/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy90b2tlbml6YXRpb25fdXRpbHMucHk=) | `89.84% <0.00%> (+0.39%)` | :arrow_up: |\n| [src/transformers/generation\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6960/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9nZW5lcmF0aW9uX3V0aWxzLnB5) | `97.21% <0.00%> (+0.83%)` | :arrow_up: |\n| ... and [14 more](https://codecov.io/gh/huggingface/transformers/pull/6960/diff?src=pr&el=tree-more) | |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/6960?src=pr&el=continue).\n> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta)\n> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`\n> Powered by [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6960?src=pr&el=footer). Last update [4ebb52a...402e26a](https://codecov.io/gh/huggingface/transformers/pull/6960?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n", "really cool model, thanks for sharing:\r\n\r\n<img width=\"736\" alt=\"Screenshot 2020-09-05 at 18 49 27\" src=\"https://user-images.githubusercontent.com/326577/92309755-4883d480-ef76-11ea-90e5-27a97e7b2746.png\">\r\n", "ヘ( ^o^)ノ＼(^_^ ) thanks @julien-c " ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6959?src=pr&el=h1) Report\n> Merging [#6959](https://codecov.io/gh/huggingface/transformers/pull/6959?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/56742e9f610231d7b28fe2387770dc56014b79de?el=desc) will **decrease** coverage by `0.01%`.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/6959?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #6959 +/- ##\n==========================================\n- Coverage 80.00% 79.98% -0.02% \n==========================================\n Files 161 161 \n Lines 30120 30120 \n==========================================\n- Hits 24097 24092 -5 \n- Misses 6023 6028 +5 \n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/6959?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/generation\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6959/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9nZW5lcmF0aW9uX3V0aWxzLnB5) | `97.21% <ø> (ø)` | |\n| [src/transformers/generation\\_tf\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6959/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9nZW5lcmF0aW9uX3RmX3V0aWxzLnB5) | `83.45% <0.00%> (-1.76%)` | :arrow_down: |\n| [src/transformers/file\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6959/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9maWxlX3V0aWxzLnB5) | `82.66% <0.00%> (+0.25%)` | :arrow_up: |\n| [src/transformers/data/data\\_collator.py](https://codecov.io/gh/huggingface/transformers/pull/6959/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9kYXRhL2RhdGFfY29sbGF0b3IucHk=) | `92.30% <0.00%> (+0.40%)` | :arrow_up: |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/6959?src=pr&el=continue).\n> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta)\n> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`\n> Powered by [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6959?src=pr&el=footer). Last update [56742e9...12a1792](https://codecov.io/gh/huggingface/transformers/pull/6959?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n" ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6958?src=pr&el=h1) Report\n> Merging [#6958](https://codecov.io/gh/huggingface/transformers/pull/6958?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/56742e9f610231d7b28fe2387770dc56014b79de?el=desc) will **increase** coverage by `0.30%`.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/6958?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #6958 +/- ##\n==========================================\n+ Coverage 80.00% 80.30% +0.30% \n==========================================\n Files 161 161 \n Lines 30120 30120 \n==========================================\n+ Hits 24097 24189 +92 \n+ Misses 6023 5931 -92 \n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/6958?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/modeling\\_tf\\_albert.py](https://codecov.io/gh/huggingface/transformers/pull/6958/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9hbGJlcnQucHk=) | `21.47% <0.00%> (-69.44%)` | :arrow_down: |\n| [src/transformers/modeling\\_tf\\_flaubert.py](https://codecov.io/gh/huggingface/transformers/pull/6958/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9mbGF1YmVydC5weQ==) | `24.53% <0.00%> (-63.81%)` | :arrow_down: |\n| [src/transformers/tokenization\\_albert.py](https://codecov.io/gh/huggingface/transformers/pull/6958/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy90b2tlbml6YXRpb25fYWxiZXJ0LnB5) | `28.84% <0.00%> (-58.66%)` | :arrow_down: |\n| [src/transformers/modeling\\_lxmert.py](https://codecov.io/gh/huggingface/transformers/pull/6958/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19seG1lcnQucHk=) | `70.01% <0.00%> (-20.75%)` | :arrow_down: |\n| [src/transformers/tokenization\\_utils\\_base.py](https://codecov.io/gh/huggingface/transformers/pull/6958/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy90b2tlbml6YXRpb25fdXRpbHNfYmFzZS5weQ==) | `94.03% <0.00%> (+0.27%)` | :arrow_up: |\n| [src/transformers/modeling\\_tf\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6958/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl91dGlscy5weQ==) | `87.29% <0.00%> (+0.32%)` | :arrow_up: |\n| [src/transformers/data/data\\_collator.py](https://codecov.io/gh/huggingface/transformers/pull/6958/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9kYXRhL2RhdGFfY29sbGF0b3IucHk=) | `92.30% <0.00%> (+0.40%)` | :arrow_up: |\n| [src/transformers/modeling\\_bart.py](https://codecov.io/gh/huggingface/transformers/pull/6958/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19iYXJ0LnB5) | `94.41% <0.00%> (+0.50%)` | :arrow_up: |\n| [src/transformers/generation\\_tf\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/6958/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9nZW5lcmF0aW9uX3RmX3V0aWxzLnB5) | `86.71% <0.00%> (+1.50%)` | :arrow_up: |\n| [src/transformers/activations.py](https://codecov.io/gh/huggingface/transformers/pull/6958/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9hY3RpdmF0aW9ucy5weQ==) | `90.00% <0.00%> (+5.00%)` | :arrow_up: |\n| ... and [3 more](https://codecov.io/gh/huggingface/transformers/pull/6958/diff?src=pr&el=tree-more) | |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/6958?src=pr&el=continue).\n> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta)\n> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`\n> Powered by [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6958?src=pr&el=footer). Last update [56742e9...6a043c8](https://codecov.io/gh/huggingface/transformers/pull/6958?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n" ]

[]

[]

[ "I found a sort of band-aid, I added this code in the model's tokenization code, right after init of `PRETRAINED_INIT_CONFIGURATION, PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES`:\r\n\r\n```\r\nLOCALIZE=1\r\nif LOCALIZE:\r\n old, new = (\"stas/\", \"/mnt/nvme1/code/huggingface/transformers-fair-wmt/data/\")\r\n\r\n def localize(buf): return buf.replace(old, new)\r\n\r\n for d in [PRETRAINED_INIT_CONFIGURATION, PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES]:\r\n for k, v in d.copy().items():\r\n d[localize(k)] = v\r\n\r\n for d in [PRETRAINED_VOCAB_FILES_MAP]:\r\n for tk, tv in d.items():\r\n for k, v in tv.copy().items():\r\n tv[localize(k)] = v\r\n```\r\n\r\nIt's still not great, since now I can't commit this file to repo as this leads to a problem of committing other changes in this file. But at least I can move forward.", "I dug dipper and found how to solve it - needed to create `model_dir/tokenizer_config.json` and put the special init params there. And do several more tweaks to make the vocab files to not include the language names in the filename, but use the generic 'vocab-src.txt', 'vocab-tgt.txt'." ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6956?src=pr&el=h1) Report\n> Merging [#6956](https://codecov.io/gh/huggingface/transformers/pull/6956?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/eff274d629c95fca459969b530b4ad0da5563918?el=desc) will **decrease** coverage by `6.41%`.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/6956?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #6956 +/- ##\n==========================================\n- Coverage 80.02% 73.61% -6.42% \n==========================================\n Files 161 161 \n Lines 30120 30120 \n==========================================\n- Hits 24105 22172 -1933 \n- Misses 6015 7948 +1933 \n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/6956?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/configuration\\_bart.py](https://codecov.io/gh/huggingface/transformers/pull/6956/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9jb25maWd1cmF0aW9uX2JhcnQucHk=) | `94.00% <ø> (ø)` | |\n| [src/transformers/configuration\\_bert.py](https://codecov.io/gh/huggingface/transformers/pull/6956/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9jb25maWd1cmF0aW9uX2JlcnQucHk=) | `100.00% <ø> (ø)` | |\n| [src/transformers/configuration\\_electra.py](https://codecov.io/gh/huggingface/transformers/pull/6956/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9jb25maWd1cmF0aW9uX2VsZWN0cmEucHk=) | `100.00% <ø> (ø)` | |\n| [src/transformers/configuration\\_flaubert.py](https://codecov.io/gh/huggingface/transformers/pull/6956/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9jb25maWd1cmF0aW9uX2ZsYXViZXJ0LnB5) | `100.00% <ø> (ø)` | |\n| [src/transformers/configuration\\_gpt2.py](https://codecov.io/gh/huggingface/transformers/pull/6956/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9jb25maWd1cmF0aW9uX2dwdDIucHk=) | `97.29% <ø> (ø)` | |\n| [src/transformers/configuration\\_lxmert.py](https://codecov.io/gh/huggingface/transformers/pull/6956/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9jb25maWd1cmF0aW9uX2x4bWVydC5weQ==) | `20.00% <ø> (-80.00%)` | :arrow_down: |\n| [src/transformers/configuration\\_mobilebert.py](https://codecov.io/gh/huggingface/transformers/pull/6956/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9jb25maWd1cmF0aW9uX21vYmlsZWJlcnQucHk=) | `97.05% <ø> (ø)` | |\n| [src/transformers/configuration\\_openai.py](https://codecov.io/gh/huggingface/transformers/pull/6956/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9jb25maWd1cmF0aW9uX29wZW5haS5weQ==) | `97.14% <ø> (ø)` | |\n| [src/transformers/configuration\\_reformer.py](https://codecov.io/gh/huggingface/transformers/pull/6956/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9jb25maWd1cmF0aW9uX3JlZm9ybWVyLnB5) | `21.62% <ø> (-78.38%)` | :arrow_down: |\n| [src/transformers/configuration\\_retribert.py](https://codecov.io/gh/huggingface/transformers/pull/6956/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9jb25maWd1cmF0aW9uX3JldHJpYmVydC5weQ==) | `34.78% <ø> (ø)` | |\n| ... and [89 more](https://codecov.io/gh/huggingface/transformers/pull/6956/diff?src=pr&el=tree-more) | |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/6956?src=pr&el=continue).\n> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta)\n> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`\n> Powered by [Codecov](https://codecov.io/gh/huggingface/transformers/pull/6956?src=pr&el=footer). Last update [eff274d...e93aa69](https://codecov.io/gh/huggingface/transformers/pull/6956?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n", "Great, thanks a lot!" ]