Datasets:

---

DanielPFlorian

/

Transformers-Github-Issues

like 0

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

[]

[ "Hello, does this still crash if you replace `TRAIN_FILE` with `$TRAIN_FILE` and `TEST_FILE` with `$TEST_FILE` in your command ?", "@LysandreJik - Yes, it does.\r\n```\r\n10/14/2019 16:55:10 - WARNING - __main__ - Process rank: -1, device: cpu, n_gpu: 0, distributed training: False, 16-bits training: False\r\n10/14/2019 16:55:10 - INFO - transformers.file_utils - https://s3.amazonaws.com/models.huggingface.co/bert/gpt2-config.json not found in cache or force_download set to True, downloading to /tmp/tmpi7g0lm6a\r\n100% 176/176 [00:00<00:00, 131305.14B/s]\r\n10/14/2019 16:55:10 - INFO - transformers.file_utils - copying /tmp/tmpi7g0lm6a to cache at /root/.cache/torch/transformers/4be02c5697d91738003fb1685c9872f284166aa32e061576bbe6aaeb95649fcf.085d5f6a8e7812ea05ff0e6ed0645ab2e75d80387ad55c1ad9806ee70d272f80\r\n10/14/2019 16:55:10 - INFO - transformers.file_utils - creating metadata file for /root/.cache/torch/transformers/4be02c5697d91738003fb1685c9872f284166aa32e061576bbe6aaeb95649fcf.085d5f6a8e7812ea05ff0e6ed0645ab2e75d80387ad55c1ad9806ee70d272f80\r\n10/14/2019 16:55:10 - INFO - transformers.file_utils - removing temp file /tmp/tmpi7g0lm6a\r\n10/14/2019 16:55:10 - INFO - transformers.configuration_utils - loading configuration file https://s3.amazonaws.com/models.huggingface.co/bert/gpt2-config.json from cache at /root/.cache/torch/transformers/4be02c5697d91738003fb1685c9872f284166aa32e061576bbe6aaeb95649fcf.085d5f6a8e7812ea05ff0e6ed0645ab2e75d80387ad55c1ad9806ee70d272f80\r\n10/14/2019 16:55:10 - INFO - transformers.configuration_utils - Model config {\r\n \"attn_pdrop\": 0.1,\r\n \"embd_pdrop\": 0.1,\r\n \"finetuning_task\": null,\r\n \"initializer_range\": 0.02,\r\n \"layer_norm_epsilon\": 1e-05,\r\n \"n_ctx\": 1024,\r\n \"n_embd\": 768,\r\n \"n_head\": 12,\r\n \"n_layer\": 12,\r\n \"n_positions\": 1024,\r\n \"num_labels\": 1,\r\n \"output_attentions\": false,\r\n \"output_hidden_states\": false,\r\n \"output_past\": true,\r\n \"pruned_heads\": {},\r\n \"resid_pdrop\": 0.1,\r\n \"summary_activation\": null,\r\n \"summary_first_dropout\": 0.1,\r\n \"summary_proj_to_labels\": true,\r\n \"summary_type\": \"cls_index\",\r\n \"summary_use_proj\": true,\r\n \"torchscript\": false,\r\n \"use_bfloat16\": false,\r\n \"vocab_size\": 50257\r\n}\r\n\r\n10/14/2019 16:55:11 - INFO - transformers.file_utils - https://s3.amazonaws.com/models.huggingface.co/bert/gpt2-vocab.json not found in cache or force_download set to True, downloading to /tmp/tmp103hb26e\r\n100% 1042301/1042301 [00:00<00:00, 3111277.27B/s]\r\n10/14/2019 16:55:11 - INFO - transformers.file_utils - copying /tmp/tmp103hb26e to cache at /root/.cache/torch/transformers/f2808208f9bec2320371a9f5f891c184ae0b674ef866b79c58177067d15732dd.1512018be4ba4e8726e41b9145129dc30651ea4fec86aa61f4b9f40bf94eac71\r\n10/14/2019 16:55:11 - INFO - transformers.file_utils - creating metadata file for /root/.cache/torch/transformers/f2808208f9bec2320371a9f5f891c184ae0b674ef866b79c58177067d15732dd.1512018be4ba4e8726e41b9145129dc30651ea4fec86aa61f4b9f40bf94eac71\r\n10/14/2019 16:55:11 - INFO - transformers.file_utils - removing temp file /tmp/tmp103hb26e\r\n10/14/2019 16:55:12 - INFO - transformers.file_utils - https://s3.amazonaws.com/models.huggingface.co/bert/gpt2-merges.txt not found in cache or force_download set to True, downloading to /tmp/tmpp3c_047z\r\n100% 456318/456318 [00:00<00:00, 1830222.06B/s]\r\n10/14/2019 16:55:12 - INFO - transformers.file_utils - copying /tmp/tmpp3c_047z to cache at /root/.cache/torch/transformers/d629f792e430b3c76a1291bb2766b0a047e36fae0588f9dbc1ae51decdff691b.70bec105b4158ed9a1747fea67a43f5dee97855c64d62b6ec3742f4cfdb5feda\r\n10/14/2019 16:55:12 - INFO - transformers.file_utils - creating metadata file for /root/.cache/torch/transformers/d629f792e430b3c76a1291bb2766b0a047e36fae0588f9dbc1ae51decdff691b.70bec105b4158ed9a1747fea67a43f5dee97855c64d62b6ec3742f4cfdb5feda\r\n10/14/2019 16:55:12 - INFO - transformers.file_utils - removing temp file /tmp/tmpp3c_047z\r\n10/14/2019 16:55:12 - INFO - transformers.tokenization_utils - loading file https://s3.amazonaws.com/models.huggingface.co/bert/gpt2-vocab.json from cache at /root/.cache/torch/transformers/f2808208f9bec2320371a9f5f891c184ae0b674ef866b79c58177067d15732dd.1512018be4ba4e8726e41b9145129dc30651ea4fec86aa61f4b9f40bf94eac71\r\n10/14/2019 16:55:12 - INFO - transformers.tokenization_utils - loading file https://s3.amazonaws.com/models.huggingface.co/bert/gpt2-merges.txt from cache at /root/.cache/torch/transformers/d629f792e430b3c76a1291bb2766b0a047e36fae0588f9dbc1ae51decdff691b.70bec105b4158ed9a1747fea67a43f5dee97855c64d62b6ec3742f4cfdb5feda\r\n10/14/2019 16:55:13 - INFO - transformers.file_utils - https://s3.amazonaws.com/models.huggingface.co/bert/gpt2-pytorch_model.bin not found in cache or force_download set to True, downloading to /tmp/tmpfgx0mkjn\r\n100% 548118077/548118077 [00:15<00:00, 34424870.34B/s]\r\n10/14/2019 16:55:29 - INFO - transformers.file_utils - copying /tmp/tmpfgx0mkjn to cache at /root/.cache/torch/transformers/4295d67f022061768f4adc386234dbdb781c814c39662dd1662221c309962c55.778cf36f5c4e5d94c8cd9cefcf2a580c8643570eb327f0d4a1f007fab2acbdf1\r\n10/14/2019 16:55:31 - INFO - transformers.file_utils - creating metadata file for /root/.cache/torch/transformers/4295d67f022061768f4adc386234dbdb781c814c39662dd1662221c309962c55.778cf36f5c4e5d94c8cd9cefcf2a580c8643570eb327f0d4a1f007fab2acbdf1\r\n10/14/2019 16:55:31 - INFO - transformers.file_utils - removing temp file /tmp/tmpfgx0mkjn\r\n10/14/2019 16:55:31 - INFO - transformers.modeling_utils - loading weights file https://s3.amazonaws.com/models.huggingface.co/bert/gpt2-pytorch_model.bin from cache at /root/.cache/torch/transformers/4295d67f022061768f4adc386234dbdb781c814c39662dd1662221c309962c55.778cf36f5c4e5d94c8cd9cefcf2a580c8643570eb327f0d4a1f007fab2acbdf1\r\n10/14/2019 16:55:35 - INFO - __main__ - Training/evaluation parameters Namespace(adam_epsilon=1e-08, block_size=1024, cache_dir='', config_name='', device=device(type='cpu'), do_eval=True, do_lower_case=False, do_train=True, eval_all_checkpoints=False, eval_data_file='/content/wikitext-103-raw/wiki.test.raw', evaluate_during_training=False, fp16=False, fp16_opt_level='O1', gradient_accumulation_steps=1, learning_rate=5e-05, local_rank=-1, logging_steps=50, max_grad_norm=1.0, max_steps=-1, mlm=False, mlm_probability=0.15, model_name_or_path='gpt2', model_type='gpt2', n_gpu=0, no_cuda=False, num_train_epochs=1.0, output_dir='output', overwrite_cache=False, overwrite_output_dir=False, per_gpu_eval_batch_size=4, per_gpu_train_batch_size=4, save_steps=50, save_total_limit=None, seed=42, server_ip='', server_port='', tokenizer_name='', train_data_file='/content/wikitext-103-raw/wiki.train.raw', warmup_steps=0, weight_decay=0.0)\r\n10/14/2019 16:55:35 - INFO - __main__ - Creating features from dataset file at /content/wikitext-103-raw\r\ntcmalloc: large alloc 1081139200 bytes == 0x8b1a4000 @ 0x7f2fddd0b1e7 0x50ca4f 0x50440b 0x504bff 0x52d7c2 0x59aa60 0x4f858d 0x4f98c7 0x4f6128 0x4f42e7 0x5a1481 0x57c57c 0x57e6ae 0x583d97 0x627fff 0x4f858d 0x4f98c7 0x4f6128 0x4f426e 0x5a1481 0x512a60 0x53ee21 0x57ec0c 0x4f88ba 0x4fa6c0 0x4f6128 0x4f7d60 0x4f876d 0x4fa6c0 0x4f6128 0x4f7d60\r\ntcmalloc: large alloc 2162278400 bytes == 0xcb8b2000 @ 0x7f2fddd0b1e7 0x50ca4f 0x50440b 0x504bff 0x52d7c2 0x59aa60 0x4f858d 0x4f98c7 0x4f6128 0x4f42e7 0x5a1481 0x57c57c 0x57e6ae 0x583d97 0x627fff 0x4f858d 0x4f98c7 0x4f6128 0x4f426e 0x5a1481 0x512a60 0x53ee21 0x57ec0c 0x4f88ba 0x4fa6c0 0x4f6128 0x4f7d60 0x4f876d 0x4fa6c0 0x4f6128 0x4f7d60\r\ntcmalloc: large alloc 2158272512 bytes == 0x14c6ce000 @ 0x7f2fddd0b1e7 0x5bd1cb 0x583f51 0x627fff 0x4f858d 0x4f98c7 0x4f6128 0x4f426e 0x5a1481 0x512a60 0x53ee21 0x57ec0c 0x4f88ba 0x4fa6c0 0x4f6128 0x4f7d60 0x4f876d 0x4fa6c0 0x4f6128 0x4f7d60 0x4f876d 0x4f98c7 0x4f6128 0x4f9023 0x6415b2 0x64166a 0x643730 0x62b26e 0x4b4cb0 0x7f2fdd908b97 0x5bdf6a\r\ntcmalloc: large alloc 2158272512 bytes == 0x6ae1c000 @ 0x7f2fddd0b1e7 0x50ca4f 0x50de4a 0x58405c 0x627fff 0x4f858d 0x4f98c7 0x4f6128 0x4f426e 0x5a1481 0x512a60 0x53ee21 0x57ec0c 0x4f88ba 0x4fa6c0 0x4f6128 0x4f7d60 0x4f876d 0x4fa6c0 0x4f6128 0x4f7d60 0x4f876d 0x4f98c7 0x4f6128 0x4f9023 0x6415b2 0x64166a 0x643730 0x62b26e 0x4b4cb0 0x7f2fdd908b97\r\ntcmalloc: large alloc 2158272512 bytes == 0xeb866000 @ 0x7f2fddd0b1e7 0x50ca4f 0x50de4a 0x5aebf9 0x4f858d 0x4f98c7 0x4f6128 0x4f7d60 0x4f876d 0x4f98c7 0x4f6128 0x4f7d60 0x4f876d 0x4f98c7 0x4f6128 0x4f7d60 0x4f876d 0x4f98c7 0x4f6128 0x4f426e 0x5a1481 0x512a60 0x53ee21 0x57ec0c 0x4f88ba 0x4fa6c0 0x4f6128 0x4f7d60 0x4f876d 0x4fa6c0 0x4f6128\r\n---------------------------------------------------------------------------\r\nCalledProcessError Traceback (most recent call last)\r\n<ipython-input-5-cbb21af32de2> in <module>()\r\n----> 1 get_ipython().run_cell_magic('shell', '', 'cd /content/transformers\\nexport TRAIN_FILE=/content/wikitext-103-raw/wiki.train.raw\\nexport TEST_FILE=/content/wikitext-103-raw/wiki.test.raw\\n \\npython /content/transformers/examples/run_lm_finetuning.py \\\\\\n --output_dir=output \\\\\\n --model_type=gpt2 \\\\\\n --model_name_or_path=gpt2 \\\\\\n --do_train \\\\\\n --train_data_file=$TRAIN_FILE \\\\\\n --do_eval \\\\\\n --eval_data_file=$TEST_FILE')\r\n\r\n2 frames\r\n/usr/local/lib/python3.6/dist-packages/google/colab/_system_commands.py in check_returncode(self)\r\n 136 if self.returncode:\r\n 137 raise subprocess.CalledProcessError(\r\n--> 138 returncode=self.returncode, cmd=self.args, output=self.output)\r\n 139 \r\n 140 def _repr_pretty_(self, p, cycle): # pylint:disable=unused-argument\r\n\r\nCalledProcessError: Command 'cd /content/transformers\r\nexport TRAIN_FILE=/content/wikitext-103-raw/wiki.train.raw\r\nexport TEST_FILE=/content/wikitext-103-raw/wiki.test.raw\r\n \r\npython /content/transformers/examples/run_lm_finetuning.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' died with <Signals.SIGKILL: 9>.\r\n```\r\nI've tried: with GPU, without GPU, TPU. All have the same error.", "I currently face the same issue", "I have the same problem:\r\nIm running on GoogleColab:\r\n!python 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=\"../../drive/My Drive/HuggingFace/train.txt\" \\\r\n --per_gpu_train_batch_size=1 \r\n\r\nAnd I get this:\r\n\r\n`03/03/2020 06:46:34 - INFO - __main__ - Creating features from dataset file at ../../drive/My Drive/HuggingFace\r\ntcmalloc: large alloc 1684217856 bytes == 0x14e26c000 @ 0x7f80cc4a21e7 0x5450df 0x52e319 0x52f3cf 0x53e701 0x4f2b30 0x50a8af 0x50c5b9 0x508245 0x5096b7 0x595311 0x5a522c 0x5a670a 0x4bb19c 0x5bd993 0x50a8af 0x50c5b9 0x508245 0x509642 0x595311 0x54a6ff 0x551b81 0x5aa6ec 0x50abb3 0x50d390 0x508245 0x50a080 0x50aa7d 0x50d390 0x508245 0x50a080\r\ntcmalloc: large alloc 3368435712 bytes == 0x7f7f0339c000 @ 0x7f80cc4a21e7 0x5450df 0x52e319 0x52f3cf 0x53e701 0x4f2b30 0x50a8af 0x50c5b9 0x508245 0x5096b7 0x595311 0x5a522c 0x5a670a 0x4bb19c 0x5bd993 0x50a8af 0x50c5b9 0x508245 0x509642 0x595311 0x54a6ff 0x551b81 0x5aa6ec 0x50abb3 0x50d390 0x508245 0x50a080 0x50aa7d 0x50d390 0x508245 0x50a080\r\ntcmalloc: large alloc 3344367616 bytes == 0x7f7e3be2c000 @ 0x7f80cc4a21e7 0x5ad4cb 0x4bb356 0x5bd993 0x50a8af 0x50c5b9 0x508245 0x509642 0x595311 0x54a6ff 0x551b81 0x5aa6ec 0x50abb3 0x50d390 0x508245 0x50a080 0x50aa7d 0x50d390 0x508245 0x50a080 0x50aa7d 0x50c5b9 0x508245 0x50b403 0x635222 0x6352d7 0x638a8f 0x639631 0x4b0f40 0x7f80cc09fb97 0x5b2fda\r\ntcmalloc: large alloc 3343319040 bytes == 0x7f7f0339c000 @ 0x7f80cc4a21e7 0x5450df 0x5464ca 0x4bb455 0x5bd993 0x50a8af 0x50c5b9 0x508245 0x509642 0x595311 0x54a6ff 0x551b81 0x5aa6ec 0x50abb3 0x50d390 0x508245 0x50a080 0x50aa7d 0x50d390 0x508245 0x50a080 0x50aa7d 0x50c5b9 0x508245 0x50b403 0x635222 0x6352d7 0x638a8f 0x639631 0x4b0f40 0x7f80cc09fb97\r\ntcmalloc: large alloc 3343319040 bytes == 0x7f7e3be2c000 @ 0x7f80cc4a21e7 0x5450df 0x5464ca 0x536c89 0x50a8af 0x50c5b9 0x508245 0x50a080 0x50aa7d 0x50c5b9 0x508245 0x50a080 0x50aa7d 0x50c5b9 0x508245 0x509642 0x595311 0x54a6ff 0x551b81 0x5aa6ec 0x50abb3 0x50d390 0x508245 0x50a080 0x50aa7d 0x50d390 0x508245 0x50a080 0x50aa7d 0x50c5b9 0x508245\r\ntcmalloc: large alloc 3343319040 bytes == 0x7f7e3be2c000 @ 0x7f80cc4a21e7 0x5450df 0x5464ca 0x536808 0x50a8af 0x50c5b9 0x508245 0x50a080 0x50aa7d 0x50c5b9 0x508245 0x50a080 0x50aa7d 0x50c5b9 0x508245 0x50a080 0x50aa7d 0x50c5b9 0x508245 0x509642 0x595311 0x54a6ff 0x551b81 0x5aa6ec 0x50abb3 0x50d390 0x508245 0x50a080 0x50aa7d 0x50d390 0x508245\r\n^C`", "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", "I still get this problem. Are there any updates?", "I also got this issue! Are there any updates?\r\n" ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/1509?src=pr&el=h1) Report\n> Merging [#1509](https://codecov.io/gh/huggingface/transformers/pull/1509?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/a701c9b32126f1e6974d9fcb3a5c3700527d8559?src=pr&el=desc) will **decrease** coverage by `1.24%`.\n> The diff coverage is `6.25%`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/1509?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #1509 +/- ##\n==========================================\n- Coverage 85.98% 84.74% -1.25% \n==========================================\n Files 91 91 \n Lines 13579 13594 +15 \n==========================================\n- Hits 11676 11520 -156 \n- Misses 1903 2074 +171\n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/1509?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [transformers/modeling\\_tf\\_pytorch\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/1509/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL21vZGVsaW5nX3RmX3B5dG9yY2hfdXRpbHMucHk=) | `9.85% <0%> (-66.91%)` | :arrow_down: |\n| [transformers/tests/modeling\\_tf\\_common\\_test.py](https://codecov.io/gh/huggingface/transformers/pull/1509/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL3Rlc3RzL21vZGVsaW5nX3RmX2NvbW1vbl90ZXN0LnB5) | `79.78% <6.66%> (-16.75%)` | :arrow_down: |\n| [transformers/modeling\\_roberta.py](https://codecov.io/gh/huggingface/transformers/pull/1509/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL21vZGVsaW5nX3JvYmVydGEucHk=) | `65.46% <0%> (-15.11%)` | :arrow_down: |\n| [transformers/modeling\\_xlnet.py](https://codecov.io/gh/huggingface/transformers/pull/1509/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL21vZGVsaW5nX3hsbmV0LnB5) | `70.87% <0%> (-2.46%)` | :arrow_down: |\n| [transformers/modeling\\_ctrl.py](https://codecov.io/gh/huggingface/transformers/pull/1509/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL21vZGVsaW5nX2N0cmwucHk=) | `93.18% <0%> (-2.28%)` | :arrow_down: |\n| [transformers/modeling\\_openai.py](https://codecov.io/gh/huggingface/transformers/pull/1509/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL21vZGVsaW5nX29wZW5haS5weQ==) | `80.4% <0%> (-1.36%)` | :arrow_down: |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/1509?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/1509?src=pr&el=footer). Last update [a701c9b...898ce06](https://codecov.io/gh/huggingface/transformers/pull/1509?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n", "Oh nice catch. Let's add a test on `load_tf2_checkpoint_in_pytorch_model` to catch such errors going forward.", "Ok, merging, thanks", "I was wondering where I should add that test and saw you just did it yourself :) thanks\r\n" ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/1508?src=pr&el=h1) Report\n> Merging [#1508](https://codecov.io/gh/huggingface/transformers/pull/1508?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/a701c9b32126f1e6974d9fcb3a5c3700527d8559?src=pr&el=desc) will **not change** coverage.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/1508?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #1508 +/- ##\n=======================================\n Coverage 85.98% 85.98% \n=======================================\n Files 91 91 \n Lines 13579 13579 \n=======================================\n Hits 11676 11676 \n Misses 1903 1903\n```\n\n\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/1508?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/1508?src=pr&el=footer). Last update [a701c9b...2c1d556](https://codecov.io/gh/huggingface/transformers/pull/1508?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n", "Great, looks good to me cc @LysandreJik.\r\n\r\nAdding the info of the PR in the example README.", "Ok, merging, thanks @tlkh!" ]

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

[]

[ "the same here.\r\n\r\ninstalling last apex code from repository: git clone https://github.com/NVIDIA/apex\r\n\r\nit says it's apex-0.1 version, but i think it should say apex-1.0", "I run into this problem while trying to create virtual GPU devices:\r\n\r\n```python\r\nimport tensorflow as tf\r\nimport transformers\r\n\r\ndevices = tf.config.experimental.list_physical_devices('GPU')\r\ntf.config.experimental.set_virtual_device_configuration(devices[0], [tf.config.experimental.VirtualDeviceConfiguration(memory_limit=1024)]) \r\n```\r\nthat ends up with error:\r\n```\r\nRuntimeError Traceback (most recent call last)\r\n<ipython-input-4-d51e3d242817> in <module>\r\n----> 1 tf.config.experimental.set_virtual_device_configuration(devices[0], [tf.config.experimental.VirtualDeviceConfiguration(memory_limit=1024)])\r\n\r\n~/miniconda3/envs/transformers/lib/python3.7/site-packages/tensorflow_core/python/framework/config.py in set_virtual_device_configuration(device, virtual_devices)\r\n 554 virtual_devices: (optional) Need to update\r\n 555 \"\"\"\r\n--> 556 context.context().set_virtual_device_configuration(device, virtual_devices)\r\n\r\n~/miniconda3/envs/transformers/lib/python3.7/site-packages/tensorflow_core/python/eager/context.py in set_virtual_device_configuration(self, dev, virtual_devices)\r\n 1269 if self._context_handle is not None:\r\n 1270 raise RuntimeError(\r\n-> 1271 \"Virtual devices cannot be modified after being initialized\")\r\n 1272 \r\n 1273 self._virtual_device_map[dev] = virtual_devices\r\n\r\nRuntimeError: Virtual devices cannot be modified after being initialized\r\n```\r\nversions:\r\n* Platform Linux-5.0.9-050009-generic-x86_64-with-debian-buster-sid\r\n* Python 3.7.5 (default, Oct 25 2019, 15:51:11) \r\n* [GCC 7.3.0]\r\n* PyTorch 1.2.0\r\n* Tensorflow 2.0.0\r\n\r\nThe reason is the use of class variable `dummy_inputs` in `transformers/modeling_tf_utils.py:54` where tensorflow is initialized (and starts using GPU) at import time. I created a [PR](https://github.com/huggingface/transformers/pull/1735) that should fix this.", "Thanks for the PR and figuring it where the issue lied.", "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" ]

[]

[]

[ "Hey @juliahane, glad you’re asking: I am currently working on this (See PR #1455) Stay tuned! Closing this as it is not an issue per se.", "Hi Remi\nthanks a lot for the great work, since I need it for a deadline approaching\nvery soon, I would really appreciate\nif you may know approximately when could be possible to use?\nthanks a lot again for your efforts.\nBest regards\nJulia\n\nOn Sun, Oct 13, 2019 at 7:29 PM Rémi Louf <[email protected]> wrote:\n\n> Closed #1506 <https://github.com/huggingface/transformers/issues/1506>.\n>\n> —\n> You are receiving this because you were mentioned.\n> Reply to this email directly, view it on GitHub\n> <https://github.com/huggingface/transformers/issues/1506?email_source=notifications&email_token=AM3GZM3GKOHDP6CEUFCVFY3QONLGJA5CNFSM4JACXTZKYY3PNVWWK3TUL52HS4DFWZEXG43VMVCXMZLOORHG65DJMZUWGYLUNFXW5KTDN5WW2ZLOORPWSZGOUF2MLRY#event-2708784583>,\n> or unsubscribe\n> <https://github.com/notifications/unsubscribe-auth/AM3GZM5SYZMX4IAXP4YJNUDQONLGJANCNFSM4JACXTZA>\n> .\n>\n", "Hi, probably not in time for your deadline. We are expecting a first working version in a few weeks.", "Hi Thomas,\r\nI really need to make this code working for a deadline, I really appreciate to point me to the current existing implementations you may be aware of which I could use for now, thank you so much for your help", "@thomwolf , I see you have run_lm_finetuning.py script, can I use this script for seq2seq generation task? Does it work for this purpose? thanks", "Hi @juliahane, no you cannot use `run_lm_finetuning` for seq2seq generation.\r\n\r\nIf you cannot wait, I think this repo is a good place to start. It's based on our library and specifically target seq2seq for summarization: https://github.com/nlpyang/PreSumm", "Let's keep this issue open to gather all threads asking about seq2seq in the repo.", "> Hi Thomas,\r\n> I really need to make this code working for a deadline, I really appreciate to point me to the current existing implementations you may be aware of which I could use for now, thank you so much for your help\r\n\r\nYou can have a look at PR #1455 . What you're looking for is in the `modeling_seq2seq.py` and `run_seq2seq_finetuning.py` scripts. This only works for Bert at the moment.", "Hi\nthanks a lot for the response, I cannot see the files, I really\nappreciate sharing the files with me, thanks\n\nOn Tue, Oct 22, 2019 at 9:21 PM Rémi Louf <[email protected]> wrote:\n\n> Hi Thomas,\n> I really need to make this code working for a deadline, I really\n> appreciate to point me to the current existing implementations you may be\n> aware of which I could use for now, thank you so much for your help\n>\n> You can have a look at PR #1455\n> <https://github.com/huggingface/transformers/pull/1455> . What you're\n> looking for is in the modeling_seq2seq.py and run_seq2seq_finetuning.py\n> scripts. This only works for Bert at the moment.\n>\n> —\n> You are receiving this because you were mentioned.\n> Reply to this email directly, view it on GitHub\n> <https://github.com/huggingface/transformers/issues/1506?email_source=notifications&email_token=AM3GZM6WIWPHCKAUXPHXTM3QP5HCDA5CNFSM4JACXTZKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEB64OFI#issuecomment-545113877>,\n> or unsubscribe\n> <https://github.com/notifications/unsubscribe-auth/AM3GZMZ4LJJNLRIT6LOL5GDQP5HCDANCNFSM4JACXTZA>\n> .\n>\n", "BERT is sufficient for me, I really appreciate sharing the files and\ntelling me the commands to run them, thanks\n\nOn Wed, Oct 23, 2019 at 11:25 PM julia hane <[email protected]> wrote:\n\n> Hi\n> thanks a lot for the response, I cannot see the files, I really\n> appreciate sharing the files with me, thanks\n>\n> On Tue, Oct 22, 2019 at 9:21 PM Rémi Louf <[email protected]>\n> wrote:\n>\n>> Hi Thomas,\n>> I really need to make this code working for a deadline, I really\n>> appreciate to point me to the current existing implementations you may be\n>> aware of which I could use for now, thank you so much for your help\n>>\n>> You can have a look at PR #1455\n>> <https://github.com/huggingface/transformers/pull/1455> . What you're\n>> looking for is in the modeling_seq2seq.py and run_seq2seq_finetuning.py\n>> scripts. This only works for Bert at the moment.\n>>\n>> —\n>> You are receiving this because you were mentioned.\n>> Reply to this email directly, view it on GitHub\n>> <https://github.com/huggingface/transformers/issues/1506?email_source=notifications&email_token=AM3GZM6WIWPHCKAUXPHXTM3QP5HCDA5CNFSM4JACXTZKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEB64OFI#issuecomment-545113877>,\n>> or unsubscribe\n>> <https://github.com/notifications/unsubscribe-auth/AM3GZMZ4LJJNLRIT6LOL5GDQP5HCDANCNFSM4JACXTZA>\n>> .\n>>\n>\n", "Hi Remi\nI really appreciate providing me with the command that I could get this\npull request in my installed huggingface library, thanks\nBest\nJulia\n\nOn Tue, Oct 22, 2019 at 9:21 PM Rémi Louf <[email protected]> wrote:\n\n> Hi Thomas,\n> I really need to make this code working for a deadline, I really\n> appreciate to point me to the current existing implementations you may be\n> aware of which I could use for now, thank you so much for your help\n>\n> You can have a look at PR #1455\n> <https://github.com/huggingface/transformers/pull/1455> . What you're\n> looking for is in the modeling_seq2seq.py and run_seq2seq_finetuning.py\n> scripts. This only works for Bert at the moment.\n>\n> —\n> You are receiving this because you were mentioned.\n> Reply to this email directly, view it on GitHub\n> <https://github.com/huggingface/transformers/issues/1506?email_source=notifications&email_token=AM3GZM6WIWPHCKAUXPHXTM3QP5HCDA5CNFSM4JACXTZKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEB64OFI#issuecomment-545113877>,\n> or unsubscribe\n> <https://github.com/notifications/unsubscribe-auth/AM3GZMZ4LJJNLRIT6LOL5GDQP5HCDANCNFSM4JACXTZA>\n> .\n>\n", "```\r\ngit checkout --track origin/conditional-generation\r\n```\r\n\r\nShould work if you cloned the original repository.\r\n\r\nHowever I am afraid we cannot provide support for work that has not made its way into the library yet as the interface is very likely to change.", "Hi Remi\nI was trying to run the bert seq2seq based codes, It gots a lot of errors,\nI really appreciate if you could run it, and\nmaking sure BERT one works, thanks a lot\n\n\nOn Tue, Oct 22, 2019 at 9:21 PM Rémi Louf <[email protected]> wrote:\n\n> Hi Thomas,\n> I really need to make this code working for a deadline, I really\n> appreciate to point me to the current existing implementations you may be\n> aware of which I could use for now, thank you so much for your help\n>\n> You can have a look at PR #1455\n> <https://github.com/huggingface/transformers/pull/1455> . What you're\n> looking for is in the modeling_seq2seq.py and run_seq2seq_finetuning.py\n> scripts. This only works for Bert at the moment.\n>\n> —\n> You are receiving this because you were mentioned.\n> Reply to this email directly, view it on GitHub\n> <https://github.com/huggingface/transformers/issues/1506?email_source=notifications&email_token=AM3GZM6WIWPHCKAUXPHXTM3QP5HCDA5CNFSM4JACXTZKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEB64OFI#issuecomment-545113877>,\n> or unsubscribe\n> <https://github.com/notifications/unsubscribe-auth/AM3GZMZ4LJJNLRIT6LOL5GDQP5HCDANCNFSM4JACXTZA>\n> .\n>\n", "Hi Remi\nSure, I understand you cannot provide support for ongoing work, I anyway\nhave a deadline and will need to use it,\ncould you tell me please just how much this code is tested? Does it work\nfor BERT? what I saw the code had\nseveral bugs in the optimizer part and does not run, I really appreciate if\nyou could just tell me how much this\ncode is tested\nthanks\n\nOn Fri, Oct 25, 2019 at 12:15 PM Rémi Louf <[email protected]> wrote:\n\n> https://stackoverflow.com/questions/9537392/git-fetch-remote-branch\n>\n> The name of the branch is conditional-generation. However I am afraid we\n> cannot provide support for ongoing work that has not made its way into the\n> library yet.\n>\n> —\n> You are receiving this because you were mentioned.\n> Reply to this email directly, view it on GitHub\n> <https://github.com/huggingface/transformers/issues/1506?email_source=notifications&email_token=AM3GZM77Y7LXZSDFIIOE26TQQLBL7A5CNFSM4JACXTZKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOECH4WBQ#issuecomment-546294534>,\n> or unsubscribe\n> <https://github.com/notifications/unsubscribe-auth/AM3GZM5CMVSXO56YSYEYPILQQLBL7ANCNFSM4JACXTZA>\n> .\n>\n", "Hi Remi\nI made this work, could you please tell me how can I get the generated\nsequence from decoder please?\nthanks\n", "Hi Thomas\r\nRemi was saying in PR:#1455 it has the bert seq2seq ready, could you move in a gradual way please? So merging the codes for BERT already so people can use the BERT one, this is already great, then after a while when this is ready for also other encoders, add them later, I really appreciate adding the BERT ones thanks ", "#1455 was merged and it is now possible to define and train encoder-decoder models. Only Bert is supported at the moment.", "Hi Remi and thomas\nThank you so much for the great help, this is awesome, and I would like to\nreally appreciate your hard work,\nBest regards\nJulia\n\nOn Wed, Oct 30, 2019 at 5:47 PM Rémi Louf <[email protected]> wrote:\n\n> #1455 <https://github.com/huggingface/transformers/pull/1455> was merged\n> and it is now possible to define and train encoder-decoder models. Only\n> Bert is supported at the moment.\n>\n> —\n> You are receiving this because you were mentioned.\n> Reply to this email directly, view it on GitHub\n> <https://github.com/huggingface/transformers/issues/1506?email_source=notifications&email_token=AM3GZM6IFGCLINEDGL6ELWLQRG3CLA5CNFSM4JACXTZKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOECU5W6A#issuecomment-548002680>,\n> or unsubscribe\n> <https://github.com/notifications/unsubscribe-auth/AM3GZMY5O7QGB55RCHUSCP3QRG3CLANCNFSM4JACXTZA>\n> .\n>\n", "Hi\nI was wondering if you could give some explanations how the decoder part\nwork, I see this is a masked language model head BERT,\nused as decoder, I think masked language model head bert mask some part and\npredict specific masked tokens,\nI am not sure how this work as a generation module, thanks for clarifying.\n\nOn Wed, Oct 30, 2019 at 8:47 PM julia hane <[email protected]> wrote:\n\n> Hi Remi and thomas\n> Thank you so much for the great help, this is awesome, and I would like to\n> really appreciate your hard work,\n> Best regards\n> Julia\n>\n> On Wed, Oct 30, 2019 at 5:47 PM Rémi Louf <[email protected]>\n> wrote:\n>\n>> #1455 <https://github.com/huggingface/transformers/pull/1455> was merged\n>> and it is now possible to define and train encoder-decoder models. Only\n>> Bert is supported at the moment.\n>>\n>> —\n>> You are receiving this because you were mentioned.\n>> Reply to this email directly, view it on GitHub\n>> <https://github.com/huggingface/transformers/issues/1506?email_source=notifications&email_token=AM3GZM6IFGCLINEDGL6ELWLQRG3CLA5CNFSM4JACXTZKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOECU5W6A#issuecomment-548002680>,\n>> or unsubscribe\n>> <https://github.com/notifications/unsubscribe-auth/AM3GZMY5O7QGB55RCHUSCP3QRG3CLANCNFSM4JACXTZA>\n>> .\n>>\n>\n", "> #1455 was merged and it is now possible to define and train encoder-decoder models. Only Bert is supported at the moment.\r\n\r\nHi Remi, I posted some bugs/suggestions about this code at #1674, thanks ", "Hi\nwhen I run this code I got this erorr, thanks for help.\n\n File \"/user/julia/dev/temp/transformers/examples/utils_summarization.py\",\nline 143, in encode_for_summarization\n for line in story_lines\n File \"/user/julia/dev/temp/transformers/examples/utils_summarization.py\",\nline 143, in <listcomp>\n for line in story_lines\nAttributeError: 'BertTokenizer' object has no attribute\n'add_special_tokens_single_sequence'\n\nOn Fri, Nov 1, 2019 at 12:08 PM Rabeeh Karimi Mahabadi <\[email protected]> wrote:\n\n> #1455 <https://github.com/huggingface/transformers/pull/1455> was merged\n> and it is now possible to define and train encoder-decoder models. Only\n> Bert is supported at the moment.\n>\n> Hi Remi, I posted some bugs/suggestions about this code at #1674\n> <https://github.com/huggingface/transformers/issues/1674>, thanks\n>\n> —\n> You are receiving this because you were mentioned.\n> Reply to this email directly, view it on GitHub\n> <https://github.com/huggingface/transformers/issues/1506?email_source=notifications&email_token=AM3GZM6BVCAVV3DJCFEBAJ3QRQE3PA5CNFSM4JACXTZKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEC2UPNQ#issuecomment-548751286>,\n> or unsubscribe\n> <https://github.com/notifications/unsubscribe-auth/AM3GZMYWFHZTBG7CP2BGOR3QRQE3PANCNFSM4JACXTZA>\n> .\n>\n", "Hi\ncan you please also add a way to see the generated sequences? thanks\n\nOn Fri, Nov 1, 2019 at 3:19 PM julia hane <[email protected]> wrote:\n\n> Hi\n> when I run this code I got this erorr, thanks for help.\n>\n> File \"/user/julia/dev/temp/transformers/examples/utils_summarization.py\",\n> line 143, in encode_for_summarization\n> for line in story_lines\n> File\n> \"/user/julia/dev/temp/transformers/examples/utils_summarization.py\", line\n> 143, in <listcomp>\n> for line in story_lines\n> AttributeError: 'BertTokenizer' object has no attribute\n> 'add_special_tokens_single_sequence'\n>\n> On Fri, Nov 1, 2019 at 12:08 PM Rabeeh Karimi Mahabadi <\n> [email protected]> wrote:\n>\n>> #1455 <https://github.com/huggingface/transformers/pull/1455> was merged\n>> and it is now possible to define and train encoder-decoder models. Only\n>> Bert is supported at the moment.\n>>\n>> Hi Remi, I posted some bugs/suggestions about this code at #1674\n>> <https://github.com/huggingface/transformers/issues/1674>, thanks\n>>\n>> —\n>> You are receiving this because you were mentioned.\n>> Reply to this email directly, view it on GitHub\n>> <https://github.com/huggingface/transformers/issues/1506?email_source=notifications&email_token=AM3GZM6BVCAVV3DJCFEBAJ3QRQE3PA5CNFSM4JACXTZKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEC2UPNQ#issuecomment-548751286>,\n>> or unsubscribe\n>> <https://github.com/notifications/unsubscribe-auth/AM3GZMYWFHZTBG7CP2BGOR3QRQE3PANCNFSM4JACXTZA>\n>> .\n>>\n>\n", "If both your source are target belong to same language (summarization etc.):\r\n\r\nWell, with a next word prediction language model like GPT2, you can just create a dataset like \"source [SEP] target\" and the run the LM (```run_lm_finetuning.py```) on it. During test time, you can provide \"source [SEP]\" as your prompt and you will get \"target\" as your prediction.\r\n\r\nOne small thing that you can do is mask your source tokens in the loss computation because you don't want to predict the source tokens as well! This will give you better performance and results.\r\n\r\nThis is not much different that Seq2Seq I believe. You are sharing the same parameters for source and target.", "> #1455 was merged and it is now possible to define and train encoder-decoder models. Only Bert is supported at the moment.\r\n\r\n could you tell me how to get the two file modeling_seq2seq.py and run_seq2seq_finetuning.py,\r\nso l could fine tune seq2seq model with pretrained encode model like bert?", "any news about seq2seq training script using transformers?" ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/1505?src=pr&el=h1) Report\n> Merging [#1505](https://codecov.io/gh/huggingface/transformers/pull/1505?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/a701c9b32126f1e6974d9fcb3a5c3700527d8559?src=pr&el=desc) will **not change** coverage.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/1505?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #1505 +/- ##\n=======================================\n Coverage 85.98% 85.98% \n=======================================\n Files 91 91 \n Lines 13579 13579 \n=======================================\n Hits 11676 11676 \n Misses 1903 1903\n```\n\n\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/1505?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/1505?src=pr&el=footer). Last update [a701c9b...5a8c6e7](https://codecov.io/gh/huggingface/transformers/pull/1505?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n", "Thanks!" ]

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

[]

[ "Runs are in a dedicated environment with only the following packages:\r\n\r\npython 3.7.4\r\npytorch 1.3.0, install includes cudatoolkit 10.1\r\ntensorflow_gpu 2.0 and dependencies\r\napex 0.1\r\ntransformers 2.1.1\r\n\r\nComplete terminal output: \r\n\r\n[output_term_ERROR.TXT](https://github.com/huggingface/transformers/files/3720906/output_term_ERROR.TXT)\r\n\r\n\r\n", "Change the line in run_**.py \r\ndevice = torch.device(\"cuda\" if torch.cuda.is_available() and not args.no_cuda else \"cpu\")\r\nto \r\ndevice = torch.device(\"cuda:0\" if torch.cuda.is_available() and not args.no_cuda else \"cpu\").\r\n\r\nIn my environment, it works.", "> Change the line in run_**.py\r\n> device = torch.device(\"cuda\" if torch.cuda.is_available() and not args.no_cuda else \"cpu\")\r\n> to\r\n> device = torch.device(\"cuda:0\" if torch.cuda.is_available() and not args.no_cuda else \"cpu\").\r\n> \r\n> In my environment, it works.\r\n\r\nIt seems that all GPUs will still be used even if we specify \"cuda:0\" here. But I am not sure how much the other GPUs contribute to the computation. In my case, I have 8-way 1080ti but the other 7 are hardly fully loaded.\r\n\r\nDoes anyone compare the training speed with/without this error?", "In my case, the solution is changing \r\n```python\r\nif args.n_gpu > 1:\r\n model = torch.nn.DataParallel(model)\r\n```\r\nto\r\n```python\r\nif args.n_gpu > 1 and not isinstance(model, torch.nn.DataParallel):\r\n model = torch.nn.DataParallel(model)\r\n```\r\n", "> In my case, the solution is changing\r\n> \r\n> ```python\r\n> if args.n_gpu > 1:\r\n> model = torch.nn.DataParallel(model)\r\n> ```\r\n> \r\n> to\r\n> \r\n> ```python\r\n> if args.n_gpu > 1 and not isinstance(model, torch.nn.DataParallel):\r\n> model = torch.nn.DataParallel(model)\r\n> ```\r\n\r\nchanging this in evaluate function fixes the error, when i run with ```--evaluate_during_training```", "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", "> In my case, the solution is changing\r\n> \r\n> ```python\r\n> if args.n_gpu > 1:\r\n> model = torch.nn.DataParallel(model)\r\n> ```\r\n> \r\n> to\r\n> \r\n> ```python\r\n> if args.n_gpu > 1 and not isinstance(model, torch.nn.DataParallel):\r\n> model = torch.nn.DataParallel(model)\r\n> ```\r\n\r\nAgree, also notice that \r\n`args.eval_batch_size = args.per_gpu_eval_batch_size * max(1, args.n_gpu)` is now multiplied by n_gpu again which is undesired ", "> In my case, the solution is changing\r\n> \r\n> ```python\r\n> if args.n_gpu > 1:\r\n> model = torch.nn.DataParallel(model)\r\n> ```\r\n> \r\n> to\r\n> \r\n> ```python\r\n> if args.n_gpu > 1 and not isinstance(model, torch.nn.DataParallel):\r\n> model = torch.nn.DataParallel(model)\r\n> ```\r\n\r\nThanks! I have met the same error in evaluation function. It works for me.", "> > In my case, the solution is changing\r\n> > ```python\r\n> > if args.n_gpu > 1:\r\n> > model = torch.nn.DataParallel(model)\r\n> > ```\r\n> > \r\n> > \r\n> > to\r\n> > ```python\r\n> > if args.n_gpu > 1 and not isinstance(model, torch.nn.DataParallel):\r\n> > model = torch.nn.DataParallel(model)\r\n> > ```\r\n> \r\n> changing this in evaluate function fixes the error, when i run with `--evaluate_during_training`\r\n\r\nThis solution fixed the issue for me. I am observing this while training a new LM using transformers 2.5.1. The issue happened during evaluation.", "One more comment about this fixing. If you use a validation set with odd number of instances, it will raise an error on line`outputs = model(inputs, masked_lm_labels=labels) if args.mlm else model(inputs, labels=labels)`, if using run_language_modeling.py. This happens because the parall gpu needs two instances to be fed into. \r\n\r\nI dont know how to fix properly. All I do is add a copy of instance of the last one to meet the number requirement. \r\n\r\n\r\n> In my case, the solution is changing\r\n> \r\n> ```python\r\n> if args.n_gpu > 1:\r\n> model = torch.nn.DataParallel(model)\r\n> ```\r\n> \r\n> to\r\n> \r\n> ```python\r\n> if args.n_gpu > 1 and not isinstance(model, torch.nn.DataParallel):\r\n> model = torch.nn.DataParallel(model)\r\n> ```\r\n\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" ]

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

[]

[ "Most people truncate the document at 512 tokens.\r\n\r\nMost of the time it is enough. For example on CNNDM dataset, the lead-3 baseline give a pretty strong score, for simply using the first 3 sentences of the article as summary. \r\n\r\nIt indicates that most salient information are located at the beginning of the document (in this particular case).\r\n\r\n---\r\n\r\nBut I'm also curious of the possible solutions to **really** handle longer sequences (truncating is not really handling it...)", "Good information ... thanks.\n\nAre any of the Transformer models available capable of summarization\ntasks?\n\nFrom what I can tell they all seem geared for classification, Language\nmodeling, question/answering type tasks.\n\nOn Sun, Oct 13, 2019 at 7:42 PM Cola <[email protected]> wrote:\n\n> Most people truncate the document at 512 tokens.\n>\n> Most of the time it is enough. For example on CNNDM dataset, the lead-3\n> baseline give a pretty strong score, for simply using the first 3 sentences\n> of the article as summary.\n>\n> It indicates that most salient information are located at the beginning of\n> the document (in this particular case).\n>\n> —\n> You are receiving this because you authored the thread.\n> Reply to this email directly, view it on GitHub\n> <https://github.com/huggingface/transformers/issues/1503?email_source=notifications&email_token=AAADNMH4PSVJXDZXNZVS5STQOPL75A5CNFSM4JAAO2K2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEBDHAEQ#issuecomment-541487122>,\n> or unsubscribe\n> <https://github.com/notifications/unsubscribe-auth/AAADNMHXAPKQLZKPX2DOALTQOPL75ANCNFSM4JAAO2KQ>\n> .\n>\n", "You can take a look at this repo :\r\nhttps://github.com/nlpyang/PreSumm", "Nice paper/code ... thanks much for your time and the link!\n\n-wg\n\nOn Mon, Oct 14, 2019 at 4:29 PM Cola <[email protected]> wrote:\n\n> You can take a look at this repo :\n> https://github.com/nlpyang/PreSumm\n>\n> —\n> You are receiving this because you authored the thread.\n> Reply to this email directly, view it on GitHub\n> <https://github.com/huggingface/transformers/issues/1503?email_source=notifications&email_token=AAADNMAWWMBWPF4K5EKTMMDQOT6HLA5CNFSM4JAAO2K2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEBG5S6A#issuecomment-541972856>,\n> or unsubscribe\n> <https://github.com/notifications/unsubscribe-auth/AAADNMAKZ4LKHQJOUZKCOU3QOT6HLANCNFSM4JAAO2KQ>\n> .\n>\n", "@Colanim Indeed, for newspaper articles most of the information is contained in the first sentences. This is how journalists are taught to write! The dataset does not really push the models to their limits. If only longer pieces like New Yorker articles were available in a big dataset...\r\n\r\n@ohmeow I am currently working on the implementation of several seq2seq models that use transformers, and our first example will be abstractive summarization (PR #1455 )\r\n\r\nI am also curious about solutions to the finite number of tokens limit :)", "> Good information ... thanks. Are any of the Transformer models available capable of summarization tasks? From what I can tell they all seem geared for classification, Language modeling, question/answering type tasks.\r\n> […](#)\r\n> On Sun, Oct 13, 2019 at 7:42 PM Cola ***@***.***> wrote: Most people truncate the document at 512 tokens. Most of the time it is enough. For example on CNNDM dataset, the lead-3 baseline give a pretty strong score, for simply using the first 3 sentences of the article as summary. It indicates that most salient information are located at the beginning of the document (in this particular case). — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#1503?email_source=notifications&email_token=AAADNMH4PSVJXDZXNZVS5STQOPL75A5CNFSM4JAAO2K2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEBDHAEQ#issuecomment-541487122>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAADNMHXAPKQLZKPX2DOALTQOPL75ANCNFSM4JAAO2KQ> .\r\n\r\nmaybe this repo also helps\r\nhttps://github.com/caitian521/RCTransformer", "Thanks Remi!\n\nYah I'm playing with your summarization code in huggingface as we speak.\nLooking great! Would be nice to have fine-tuning scripts included for\nreference as well.\n\nAre you all working on implementing the extractive summarization and the\ndouble-fine-tuning example for abstractive in the paper?\n\nThanks - wg\n\nOn Tue, Oct 15, 2019 at 12:32 PM Rémi Louf <[email protected]> wrote:\n\n> @Colanim <https://github.com/Colanim> Indeed, for newspaper articles most\n> of the information is contained in the first sentences. This is how\n> journalists are taught to write! The dataset does not really push the\n> models to their limits. If only longer pieces like New Yorker articles were\n> available in a big dataset...\n>\n> @ohmeow <https://github.com/ohmeow> I am currently working on the\n> implementation of several seq2seq models that use transformers, and our\n> first example will be abstractive summarization (PR #1455\n> <https://github.com/huggingface/transformers/pull/1455> )\n>\n> I am also curious about solutions to the finite number of tokens limit :)\n>\n> —\n> You are receiving this because you were mentioned.\n> Reply to this email directly, view it on GitHub\n> <https://github.com/huggingface/transformers/issues/1503?email_source=notifications&email_token=AAADNMEZGOZPKRB4RN43H73QOYLEDA5CNFSM4JAAO2K2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEBJ6M6A#issuecomment-542369400>,\n> or unsubscribe\n> <https://github.com/notifications/unsubscribe-auth/AAADNMAIQQQMAPB3F7BK5FLQOYLEDANCNFSM4JAAO2KQ>\n> .\n>\n", "Glad it works! This is not on the roadmap at the moment, but we may come back to it later.", "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/1502?src=pr&el=h1) Report\n> Merging [#1502](https://codecov.io/gh/huggingface/transformers/pull/1502?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/a701c9b32126f1e6974d9fcb3a5c3700527d8559?src=pr&el=desc) will **increase** coverage by `<.01%`.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/1502?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #1502 +/- ##\n==========================================\n+ Coverage 85.98% 85.98% +<.01% \n==========================================\n Files 91 91 \n Lines 13579 13574 -5 \n==========================================\n- Hits 11676 11672 -4 \n+ Misses 1903 1902 -1\n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/1502?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [transformers/modeling\\_bert.py](https://codecov.io/gh/huggingface/transformers/pull/1502/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL21vZGVsaW5nX2JlcnQucHk=) | `88.17% <ø> (ø)` | :arrow_up: |\n| [transformers/tests/modeling\\_tf\\_common\\_test.py](https://codecov.io/gh/huggingface/transformers/pull/1502/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL3Rlc3RzL21vZGVsaW5nX3RmX2NvbW1vbl90ZXN0LnB5) | `94.79% <0%> (-1.74%)` | :arrow_down: |\n| [transformers/modeling\\_tf\\_auto.py](https://codecov.io/gh/huggingface/transformers/pull/1502/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL21vZGVsaW5nX3RmX2F1dG8ucHk=) | `53.33% <0%> (+2.08%)` | :arrow_up: |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/1502?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/1502?src=pr&el=footer). Last update [a701c9b...e76d715](https://codecov.io/gh/huggingface/transformers/pull/1502?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n", "Nice, thanks!" ]

[]

[]

[ "I think you have the wrong Base-Class imported? [Ref](https://github.com/huggingface/transformers/blob/a701c9b32126f1e6974d9fcb3a5c3700527d8559/transformers/modeling_xlnet.py#L959)\r\n```\r\nfrom transformers.modeling_xlnet import XLNetPreTrainedModel\r\nclass XLNetForSequenceClassification(XLNetPreTrainedModel):\r\n def __init__(self, config):\r\n super(XLNetForSequenceClassification, self).__init__(config)\r\n ....\r\n ....\r\n```", "@AdityaSoni19031997 Sorry for the delay.\r\nI fixed the issue but I don't remember what was the issue though, anyways Thanks.\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 } ]

[]

[ "You can use torch to load it, convert the weights using the helper files;\r\nNot sure about your task, but for mine, i was using a BertModel with different pre-trained weights,\r\n```\r\nmodel = BertForSequenceClassification(MODEL_PATH, num_labels=len(np.unique(y_train_torch)))\r\n```\r\n(iirc from_tf is also a param to the function)\r\n\r\nwhere `MODEL_PATH` is a directory that has \r\n- config.json.\r\n- your model [checkpoint/bin file].\r\n- a vocab file as well.", "Thank you for your reply. \r\n\r\nThe issue is a little different. All the 3 files: config.json, checkpoint, and vocab.txt are linked by a symbolic link in their repo. I am not sure how to get the actual files. Any suggestions for such a case?", "Well if you are running the experiments yourself, you will be downloading them either ways, just make changes where ever needed?\r\n\r\n(i haven't tried passing a symbolic link to this func so not sure myself but it should work imo as well)", "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" ]

[]

[]

[ "This seems weird, I'm looking into this.", "By running the run_glue.py script as it is right now with your exact parameters, I timed to model.to and it took 6.4 seconds", "Ok, thanks for looking into that! I'm using my own dataset so I made adjustments to the processor, but I don't think that should be causing the issue when transferring the model to the GPU. I'll run a few more tests and see if I can pinpoint what is going on. It's super helpful to know that you are seeing it take only 6.4 seconds. Thank you!", "I just tested again using the SST-2 data keeping the run_glue.py code as is and I'm still having the same issue. My guess is that there is something with my VM set up that's causing the hanging issue. I'm having a hard time identifying what might be the exact cause of the issue.", "Hmm do you think you can reproduce it on another VM? Are you running into the same issue if you simply put the model on the device in a standalone script?", "Ok, it's definitely an issue with my setup. I have the same issue when running the following: \r\n`from torchvision import models\r\n\r\nmodel = models.densenet121(pretrained=True)\r\nmodel.to('cuda')`\r\n\r\nI'll close the issue and keep troublehsooting on my end. Thanks!", "Reopening because I found the issue and hopefully it can help someone else. I was comparing model loading times to what I was seeing on the hosted runtimes in Google Colab notebooks. \r\n\r\nEven through they have cuda toolkit 10.1 installed as you can see when running the command !nvidia-smi, when you run torch.version.cuda they have 10.0.130 installed instead of the 10.1 version. They are also running pytorch 1.2.0. \r\n\r\nI downgraded my environment to match and the model from models.densenet121(pretrained=True) loaded in 4.9 seconds.\r\n\r\nThanks for the help!" ]

[]

[]

[ "Can you check your workflow to stop opening/closing these PRs?", "@thomwolf Yeah. I have checked it. It's really embarrased to opening/closing these PRs." ]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/1494?src=pr&el=h1) Report\n> Merging [#1494](https://codecov.io/gh/huggingface/transformers/pull/1494?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/700331b5ece63381ad1b775fc8661cf3ae4493fd?src=pr&el=desc) will **decrease** coverage by `5.94%`.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/1494?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #1494 +/- ##\n==========================================\n- Coverage 85.56% 79.61% -5.95% \n==========================================\n Files 91 42 -49 \n Lines 13534 6898 -6636 \n==========================================\n- Hits 11580 5492 -6088 \n+ Misses 1954 1406 -548\n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/1494?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [transformers/modeling\\_tf\\_xlm.py](https://codecov.io/gh/huggingface/transformers/pull/1494/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL21vZGVsaW5nX3RmX3hsbS5weQ==) | | |\n| [transformers/configuration\\_distilbert.py](https://codecov.io/gh/huggingface/transformers/pull/1494/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL2NvbmZpZ3VyYXRpb25fZGlzdGlsYmVydC5weQ==) | | |\n| [transformers/configuration\\_bert.py](https://codecov.io/gh/huggingface/transformers/pull/1494/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL2NvbmZpZ3VyYXRpb25fYmVydC5weQ==) | | |\n| [transformers/tests/tokenization\\_transfo\\_xl\\_test.py](https://codecov.io/gh/huggingface/transformers/pull/1494/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL3Rlc3RzL3Rva2VuaXphdGlvbl90cmFuc2ZvX3hsX3Rlc3QucHk=) | | |\n| [transformers/tests/modeling\\_bert\\_test.py](https://codecov.io/gh/huggingface/transformers/pull/1494/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL3Rlc3RzL21vZGVsaW5nX2JlcnRfdGVzdC5weQ==) | | |\n| [transformers/tests/tokenization\\_utils\\_test.py](https://codecov.io/gh/huggingface/transformers/pull/1494/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL3Rlc3RzL3Rva2VuaXphdGlvbl91dGlsc190ZXN0LnB5) | | |\n| [transformers/tests/modeling\\_tf\\_ctrl\\_test.py](https://codecov.io/gh/huggingface/transformers/pull/1494/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL3Rlc3RzL21vZGVsaW5nX3RmX2N0cmxfdGVzdC5weQ==) | | |\n| [transformers/modeling\\_xlnet.py](https://codecov.io/gh/huggingface/transformers/pull/1494/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL21vZGVsaW5nX3hsbmV0LnB5) | | |\n| [transformers/tokenization\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/1494/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL3Rva2VuaXphdGlvbl91dGlscy5weQ==) | | |\n| [transformers/tests/modeling\\_tf\\_transfo\\_xl\\_test.py](https://codecov.io/gh/huggingface/transformers/pull/1494/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL3Rlc3RzL21vZGVsaW5nX3RmX3RyYW5zZm9feGxfdGVzdC5weQ==) | | |\n| ... and [123 more](https://codecov.io/gh/huggingface/transformers/pull/1494/diff?src=pr&el=tree-more) | |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/1494?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/1494?src=pr&el=footer). Last update [700331b...a2cfe98](https://codecov.io/gh/huggingface/transformers/pull/1494?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 } ]

[]

[ "Looks like a pretty simple and natural extension, what do you think @LysandreJik?", "It would be easy to implement indeed, do you use this for dialog (because of the alternating token type ids)?", "Not particularly, but I can certainly imagine that being a useful use-case.\r\n\r\nI am more interested in adding extra features directly to the language model when classifying a piece of text. For example:\r\n`[Location] + [Occupation] + [Social media post]`", "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" ]

[]

[]

[ "Great, ok all the models should be public. Merging this now.\r\nAwesome work @stefan-it!" ]

[]

[]

[ "Hi! It seems to me that the file that was downloaded was corrupted, probably because of lacking space or a network error. Could you try using the `from_pretrained` with the `force_download` option ?", "That worked. Thanks!", "If you are using Window 10 machine, deleting `vgg16-something` in folder `C:\\Users\\UserName\\.cache\\torch\\checkpoints` would solve probelm.", "Using `force_download` option also works for me.", "> Hi! It seems to me that the file that was downloaded was corrupted, probably because of lacking space or a network error. Could you try using the `from_pretrained` with the `force_download` option ?\r\n\r\nwhere to use this in the code?\r\n\r\n\r\n> Using `force_download` option also works for me.\r\n\r\n\r\n\r\n> Using `force_download` option also works for me.\r\n\r\n\r\n\r\n> Hi! It seems to me that the file that was downloaded was corrupted, probably because of lacking space or a network error. Could you try using the `from_pretrained` with the `force_download` option ?\r\n\r\nhow or where to use this in my code\r\n", "Well, what's your code? `from_pretrained` should be the method you use to load models/configurations/tokenizers.\r\n\r\n```py\r\nmodel = model_class.from_pretrained(pretrained_weights, force_download=True)\r\n```", "I want to run mmdetection demo image_demo.py but has this problems\r\nI use google colab pytorch 1.3.1 .\r\nTraceback (most recent call last):\r\n File \"demo/image_demo.py\", line 26, in <module>\r\n main()\r\n File \"demo/image_demo.py\", line 18, in main\r\n model = init_detector(args.config, args.checkpoint, device=args.device)\r\n File \"/content/mmdetection/mmdet/apis/inference.py\", line 35, in init_detector\r\n checkpoint = load_checkpoint(model, checkpoint)\r\n File \"/root/mmcv/mmcv/runner/checkpoint.py\", line 224, in load_checkpoint\r\n checkpoint = _load_checkpoint(filename, map_location)\r\n File \"/root/mmcv/mmcv/runner/checkpoint.py\", line 200, in _load_checkpoint\r\n checkpoint = torch.load(filename, map_location=map_location)\r\n File \"/content/anaconda3/lib/python3.7/site-packages/torch/serialization.py\", line 426, in load\r\n return _load(f, map_location, pickle_module, **pickle_load_args)\r\n File \"/content/anaconda3/lib/python3.7/site-packages/torch/serialization.py\", line 620, in _load\r\n deserialized_objects[key]._set_from_file(f, offset, f_should_read_directly)\r\nRuntimeError: storage has wrong size: expected -4934180888905747925 got 64", "if you are loading any weights in code, there might be problem with that just redownload the weights.. worked for me.", "> Using `force_download` option also works for me.\r\n\r\nWhere to add this argument ?", "See this comment https://github.com/huggingface/transformers/issues/1491#issuecomment-618626059", "\r\nhere is my code:\r\n` model = torchvision.models.detection.fasterrcnn_resnet50_fpn(pretrained=True)`\r\nand I encountered the same problem, i delete the relevant files in \"C:\\Users\\UserName\\.cache\\torch\\checkpoints\" then solve the problem.", "I am experiencing the same issue, I am using Ubuntu 18 WSL. When adding the `force_download=True` I am getting the following error: \r\n`/tape/models/modeling_utils.py\", line 506, in from_pretrained\r\n model = cls(config, *model_args, **model_kwargs)\r\nTypeError: __init__() got an unexpected keyword argument 'force_download'`\r\n\r\nAny solutions will be highly appreciated. ", "> If you are using Window 10 machine, deleting `vgg16-something` in folder `C:\\Users\\UserName\\.cache\\torch\\checkpoints` would solve probelm.\r\n\r\nThis worked for me", "so how to solve this problem? @Geraldene ", "> \r\n\r\nsorry, what does it mean? could you be more pricesly?", "> Using `force_download` option also works for me.\r\n\r\nhow did you use it/?" ]

[]

[]

[ "From what I remember (can't check now), padding up to max model seq length is not done and not necessary. The tokenizer will limit longer sequences to the max seq length, but otherwise you can just make sure the batch sizes are equal (so pad up to max _batch_ length, so you can actually create m-dimensional tensors (all rows in a matrix have to have the same length).", "@BramVanroy Not sure I understand. The above code does the encoding for one row of a text column. The encodings are appended to a list to capture all the encoding for the text column. If I convert that list to a tensor, without any padding it errors out due to different vector lengths. \r\n", "Exactly. So the tokenizer limits the length (to a max seq length) but doesn't pad it. You'll have to do that manually. You can pad up to the largest sequence _in the batch_ (rather than the max seq length) so that all items in the batch are the same size, which you can then convert to a tensor. A general usage could look like this. The padding happens at the end. Here I pad up to the MAX_SEQ_LEN if available, and otherwise up to the largest sequence in the batch.\r\n\r\n```python\r\ndef tokenize(text):\r\n all_input_ids = []\r\n all_input_mask = []\r\n\r\n for sentence in text:\r\n tokens = tokenizer.tokenize(sentence)\r\n\r\n # limit size to make room for special tokens\r\n if MAX_SEQ_LEN:\r\n tokens = tokens[0:(MAX_SEQ_LEN - 2)]\r\n\r\n # add special tokens\r\n tokens = [tokenizer.cls_token, *tokens, tokenizer.sep_token]\r\n \r\n # convert tokens to IDs\r\n input_ids = tokenizer.convert_tokens_to_ids(tokens)\r\n # create mask same size of input\r\n input_mask = [1] * len(input_ids)\r\n\r\n all_input_ids.append(input_ids)\r\n all_input_mask.append(input_mask)\r\n\r\n # pad up to max length\r\n # up to max_seq_len if provided, otherwise the max of current batch\r\n max_length = MAX_SEQ_LEN if MAX_SEQ_LEN else max([len(ids) for ids in all_input_ids])\r\n\r\n all_input_ids = torch.LongTensor([i + [tokenizer.pad_token_id] * (max_length - len(i))\r\n for i in all_input_ids])\r\n all_input_mask = torch.FloatTensor([m + [0] * (max_length - len(m)) for m in all_input_mask])\r\n\r\n return all_input_ids, all_input_mask\r\n```", "Thanks, that clarifies things. I will close this issue. ", "> From what I remember (can't check now), padding up to max model seq length is not done and not necessary. The tokenizer will limit longer sequences to the max seq length, but otherwise you can just make sure the batch sizes are equal (so pad up to max _batch_ length, so you can actually create m-dimensional tensors (all rows in a matrix have to have the same length).\r\n\r\nI am wondering if there are any disadvantages to just padding all inputs to 512. It would certainly cut down on batch processing time. But standard practice seems to be dynamically padding to the largest sequence length. ", "I have wondered about this myself but I have no answer. Perhaps someone else can help. ", "I think it comes down to loading CPU computation vs LA GPU computation. It might be worth doing some statistical analysis on the inputIDs length and checking if 1 standard limit is possible. For example if at least sample of 512 occurs in 90% of your batches, it would be worth just setting to padding to 512 for all batches. " ]

[ { "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" ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/1488?src=pr&el=h1) Report\n> Merging [#1488](https://codecov.io/gh/huggingface/transformers/pull/1488?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/f382a8decda82062bb6911f05b646f404eacfdd4?src=pr&el=desc) will **not change** coverage.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/1488?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #1488 +/- ##\n=======================================\n Coverage 85.59% 85.59% \n=======================================\n Files 91 91 \n Lines 13526 13526 \n=======================================\n Hits 11578 11578 \n Misses 1948 1948\n```\n\n\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/1488?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/1488?src=pr&el=footer). Last update [f382a8d...639f4b7](https://codecov.io/gh/huggingface/transformers/pull/1488?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n", "LGTM!", "It is possible to save/load when on TPU, just move them to CPU and save them:\r\n\r\nhttps://github.com/pytorch/xla/blob/master/API_GUIDE.md#saving-and-loading-xla-tensors" ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/1487?src=pr&el=h1) Report\n> Merging [#1487](https://codecov.io/gh/huggingface/transformers/pull/1487?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/6596e3d56626c921b3920e313866b7412633b91a?src=pr&el=desc) will **not change** coverage.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/1487?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #1487 +/- ##\n=======================================\n Coverage 85.59% 85.59% \n=======================================\n Files 91 91 \n Lines 13526 13526 \n=======================================\n Hits 11578 11578 \n Misses 1948 1948\n```\n\n\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/1487?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/1487?src=pr&el=footer). Last update [6596e3d...dd904e2](https://codecov.io/gh/huggingface/transformers/pull/1487?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n", "Great, thanks!" ]

[]

[]

[ "Hello @kamalravi \r\n\r\nFor the English Wikipedia data, I followed the scripts in XLM [here](https://github.com/facebookresearch/XLM#train-your-own-monolingual-bert-model). It downloads the latest dump and does the necessary pre-processing.\r\nFor BookCorpus, as you probably know, TBC is not distributed anymore and it's not clear to me whether I can distribute it here (I prefer not to). However, there is open-source options to collect a similar dataset (like [this one](https://github.com/soskek/bookcorpus)).\r\nIf you are ever interested in Reddit-based dataset, I used [OpenWebTextCorpus](https://skylion007.github.io/OpenWebTextCorpus/) following RoBERTa to distill DistilGPT2.\r\n\r\nHaving the raw text dumps, I simply use `scripts/binarized_data.py` to pre-process the data.\r\n\r\nVictor" ]

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

[]

[ "Any update?", "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" ]

[]

[]

[ "Hi! It seems like you have taken the example from the latest version but that your library is not up to date. Could you tell me the version of your `transformers` library ?", "Hi! \r\nI am using the version 2.0.0.", "If you´re using the version 2.0.0 you should use the [script that was used in this version](https://github.com/huggingface/transformers/blob/v2.0.0/examples/run_lm_finetuning.py). The current script works on version 2.1!", "Thanks!" ]

[]

[]

[ "Hi @saksham7778. In order to keep the repository clean we would prefer that people open pull requests once a substantial amount of work has been done. Closing for now." ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/1482?src=pr&el=h1) Report\n> Merging [#1482](https://codecov.io/gh/huggingface/transformers/pull/1482?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/6596e3d56626c921b3920e313866b7412633b91a?src=pr&el=desc) will **increase** coverage by `0.44%`.\n> The diff coverage is `97.67%`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/1482?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #1482 +/- ##\n==========================================\n+ Coverage 85.59% 86.04% +0.44% \n==========================================\n Files 91 91 \n Lines 13526 13566 +40 \n==========================================\n+ Hits 11578 11673 +95 \n+ Misses 1948 1893 -55\n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/1482?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [transformers/modeling\\_tf\\_distilbert.py](https://codecov.io/gh/huggingface/transformers/pull/1482/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL21vZGVsaW5nX3RmX2Rpc3RpbGJlcnQucHk=) | `98.59% <ø> (+2%)` | :arrow_up: |\n| [transformers/modeling\\_tf\\_gpt2.py](https://codecov.io/gh/huggingface/transformers/pull/1482/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL21vZGVsaW5nX3RmX2dwdDIucHk=) | `94.79% <ø> (+1.31%)` | :arrow_up: |\n| [transformers/modeling\\_tf\\_openai.py](https://codecov.io/gh/huggingface/transformers/pull/1482/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL21vZGVsaW5nX3RmX29wZW5haS5weQ==) | `96.04% <ø> (+1.43%)` | :arrow_up: |\n| [transformers/modeling\\_tf\\_transfo\\_xl.py](https://codecov.io/gh/huggingface/transformers/pull/1482/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL21vZGVsaW5nX3RmX3RyYW5zZm9feGwucHk=) | `92.21% <ø> (+0.97%)` | :arrow_up: |\n| [transformers/modeling\\_tf\\_bert.py](https://codecov.io/gh/huggingface/transformers/pull/1482/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL21vZGVsaW5nX3RmX2JlcnQucHk=) | `96.6% <ø> (+0.89%)` | :arrow_up: |\n| [transformers/modeling\\_tf\\_pytorch\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/1482/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL21vZGVsaW5nX3RmX3B5dG9yY2hfdXRpbHMucHk=) | `76.76% <0%> (-0.17%)` | :arrow_down: |\n| [transformers/modeling\\_gpt2.py](https://codecov.io/gh/huggingface/transformers/pull/1482/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL21vZGVsaW5nX2dwdDIucHk=) | `84.19% <100%> (+0.2%)` | :arrow_up: |\n| [transformers/modeling\\_tf\\_ctrl.py](https://codecov.io/gh/huggingface/transformers/pull/1482/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL21vZGVsaW5nX3RmX2N0cmwucHk=) | `97.75% <100%> (+1.74%)` | :arrow_up: |\n| [transformers/tests/modeling\\_tf\\_xlnet\\_test.py](https://codecov.io/gh/huggingface/transformers/pull/1482/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL3Rlc3RzL21vZGVsaW5nX3RmX3hsbmV0X3Rlc3QucHk=) | `95.74% <100%> (+0.12%)` | :arrow_up: |\n| [transformers/tests/modeling\\_tf\\_common\\_test.py](https://codecov.io/gh/huggingface/transformers/pull/1482/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL3Rlc3RzL21vZGVsaW5nX3RmX2NvbW1vbl90ZXN0LnB5) | `96.53% <100%> (+1.14%)` | :arrow_up: |\n| ... and [12 more](https://codecov.io/gh/huggingface/transformers/pull/1482/diff?src=pr&el=tree-more) | |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/1482?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/1482?src=pr&el=footer). Last update [6596e3d...4b8f3e8](https://codecov.io/gh/huggingface/transformers/pull/1482?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 } ]

[]

[ "1) Yes, the entire model is fine-tuned.\r\n\r\n2) We follow the fine-tuning that takes place in the BERT paper:\r\n > All of the parameters of BERT and W are fine-tuned jointly to maximize the log-probability of the correct label.", "Thanks.\n", "Why would _The correct way to do this is to first finetune the top n layers for some epochs and then finetune all the layers._ be the \"one and only correct way to do this\"? I've seen this being done to finetune VGG19 (IR), for different tasks, but I haven't see papers showing that one technique will result in better performance than others. Do you have a paper reference? As @LysandreJik indicates, the BERT paper indicates what the authors thing is the most efficient approach.", "Follow up on this thread: in practice, if we have a small training set, would it make sense to only finetune the top n layers?\r\nIn that case, do we want to consider adding an option that controls the number of trainable layers?", "You can also consider replacing the classification head first if you find a pre-trained model suiting your task..", "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/1480?src=pr&el=h1) Report\n> Merging [#1480](https://codecov.io/gh/huggingface/transformers/pull/1480?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/036483fae538faff62f78448b38787f3adb94f97?src=pr&el=desc) will **increase** coverage by `0.06%`.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/1480?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #1480 +/- ##\n==========================================\n+ Coverage 85.53% 85.59% +0.06% \n==========================================\n Files 91 91 \n Lines 13539 13526 -13 \n==========================================\n- Hits 11580 11578 -2 \n+ Misses 1959 1948 -11\n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/1480?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [transformers/tokenization\\_ctrl.py](https://codecov.io/gh/huggingface/transformers/pull/1480/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL3Rva2VuaXphdGlvbl9jdHJsLnB5) | `96.03% <ø> (+7.64%)` | :arrow_up: |\n| [transformers/tokenization\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/1480/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL3Rva2VuaXphdGlvbl91dGlscy5weQ==) | `91.43% <ø> (+0.42%)` | :arrow_up: |\n| [transformers/configuration\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/1480/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL2NvbmZpZ3VyYXRpb25fdXRpbHMucHk=) | `97.29% <ø> (ø)` | :arrow_up: |\n| [transformers/modeling\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/1480/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL21vZGVsaW5nX3V0aWxzLnB5) | `92.44% <ø> (ø)` | :arrow_up: |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/1480?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/1480?src=pr&el=footer). Last update [036483f...177a721](https://codecov.io/gh/huggingface/transformers/pull/1480?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n", "Updating this, upon deeper inspection, `fastBPE` tokenizer just basically [split on spaces only](https://github.com/glample/fastBPE/blob/master/fastBPE/fastBPE.hpp?fbclid=IwAR1Vp2WMLxDjpmBIpU6mkddeyxzi2vpvHOcm8fL4iaWL1m3tVbSfz-yZAcE#L652).\r\n\r\nThis tokenizer was used in CTRL which is confirmed by the fact that many vocabulary tokens in CTRL vocabulary contains end or start punctuation (see CTRL vocabulary [here](https://raw.githubusercontent.com/salesforce/ctrl/master/ctrl-vocab.json)).\r\n\r\nSo the most logical solution is thus just to split on spaces which is also the easiest solution :-)", "Ok, merging." ]

[]

[]

[ "The models we use inherit directly from `torch.nn.Module` for our pytorch models and `tf.keras.layers.Layer` for tensorflow modules. You can therefore get the total number of parameters as you would do with any other pytorch/tensorflow modules:\r\n\r\n`sum(p.numel() for p in model.parameters() if p.requires_grad)` for pytorch and\r\n`np.sum([np.prod(v.shape) for v in tf.trainable_variables])` for tensorflow, for example.", "Got it!" ]

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

[]

[ "Hey @jeffxtang in your last line you are asking for 3 outputs, but only index from [:2]. You need to change it to \r\n```\r\nloss, start_scores, end_scores = outputs[:3]\r\n```\r\nThe documentation is off in that example. As for your last question, I don't entirely understand it; however, BertForQuestionAnswering is the architecture you are using and bert-large-uncased-whole-word-masking-finetuned-squad is the weights (fine tuned on Squad 1.1) you are using in that architecture.\r\n\r\nHope that helps!", "Thanks @cformosa ! My bad, I should've checked the value of outputs instead of just asking for help :) \r\n\r\nSo my last question is how I can use the Bert model fine tuned on Squad in Python the same way as it's used in [iOS](https://developer.apple.com/machine-learning/models/#text), [which](https://developer.apple.com/documentation/coreml/finding_answers_to_questions_in_a_text_document) expects a text and a question as input then outputs a possible answer from the text. From your answer, BertForQuestionAnswering uses the pre-trained finetuned-on-squad weights so I should be able to just use the BertForQuestionAnswering class? ", "I think I'm getting closer to the solution - the code below returns `predictions` with shape [1, 14, 1024]:\r\n\r\n```\r\nmodel = BertModel.from_pretrained('bert-large-uncased-whole-word-masking-finetuned-squad')\r\nmodel.eval()\r\n\r\ntokenizer = BertTokenizer.from_pretrained('bert-base-uncased')\r\ntext = \"[CLS] Who was Jim Henson ? [SEP] Jim Henson was a nice puppet [SEP]\"\r\ntokenized_text = tokenizer.tokenize(text)\r\n\r\nindexed_tokens = tokenizer.convert_tokens_to_ids(tokenized_text)\r\nsegments_ids = [0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1]\r\n\r\ntokens_tensor = torch.tensor([indexed_tokens])\r\nsegments_tensors = torch.tensor([segments_ids])\r\n\r\nwith torch.no_grad():\r\n outputs = model(tokens_tensor, token_type_ids=segments_tensors)\r\n predictions = outputs[0]\r\n```\r\n\r\nSo the model with the pre-trained weights `bert-large-uncased-whole-word-masking-finetuned-squad` gets an input with the question \"Who was Jim Henson ?\" and the text \"Jim Henson was a nice puppet\" and outputs info that can be used to get the \"a nice puppet\" answer's indexes (10 and 12) from the `text` value in the code. But why 1024 in the predictions's shape? (14 is the length of the text) I think I'd use argmax on predictions to find out the begin and end indexes of the answer, but how exactly? Thanks!", "OK after a lot of reading and testing, I got my final complete little working program that ends up using `bert-large-uncased-whole-word-masking-finetuned-squad` with `BertForQuestionAnswering`:\r\n```\r\nimport torch\r\nfrom transformers import BertTokenizer, BertForQuestionAnswering\r\n\r\ntokenizer = BertTokenizer.from_pretrained('bert-base-uncased')\r\nmodel = BertForQuestionAnswering.from_pretrained('bert-large-uncased-whole-word-masking-finetuned-squad')\r\n\r\nquestion, text = \"Who was Jim Henson?\", \"Jim Henson was a nice puppet\"\r\ninput_text = \"[CLS] \" + question + \" [SEP] \" + text + \" [SEP]\"\r\ninput_ids = tokenizer.encode(input_text)\r\ntoken_type_ids = [0 if i <= input_ids.index(102) else 1 for i in range(len(input_ids))] \r\n\r\nstart_scores, end_scores = model(torch.tensor([input_ids]), token_type_ids=torch.tensor([token_type_ids]))\r\nall_tokens = tokenizer.convert_ids_to_tokens(input_ids) \r\nprint(' '.join(all_tokens[torch.argmax(start_scores) : torch.argmax(end_scores)+1]))\r\n# a nice puppet\r\n``` \r\n\r\nThanks huggingface for the cool stuff, although your documentation could be cooler :)", "Yes we are always a bit behind on documentation, just too many projects at the same time.\r\n\r\nIf you want to submit a PR fixing this part of the documentation that you noticed was wrong, that would be the most awesome thing!", "Totally understandable :) and would love to do a PR, but first, I'd like to understand whether what I did is THE right way or one of the right ways to use the `bert-large-uncased-whole-word-masking-finetuned-squad` model. \r\n\r\nTo be more specific: Can I use also `model = BertModel.from_pretrained('bert-large-uncased-whole-word-masking-finetuned-squad')` to get the right `start_score` and `end_score`? Or dp I have to use `model = BertForQuestionAnswering.from_pretrained('bert-large-uncased-whole-word-masking-finetuned-squad')`?\r\n", "Use `BertForQuestionAnswering`, otherwise your model will not initialize its final span classification layer. ", "Thanks for the info. PR created https://github.com/huggingface/transformers/pull/1502", "> OK after a lot of reading and testing, I got my final complete little working program that ends up using `bert-large-uncased-whole-word-masking-finetuned-squad` with `BertForQuestionAnswering`:\r\n> \r\n> ```\r\n> import torch\r\n> from transformers import BertTokenizer, BertForQuestionAnswering\r\n> \r\n> tokenizer = BertTokenizer.from_pretrained('bert-base-uncased')\r\n> model = BertForQuestionAnswering.from_pretrained('bert-large-uncased-whole-word-masking-finetuned-squad')\r\n> \r\n> question, text = \"Who was Jim Henson?\", \"Jim Henson was a nice puppet\"\r\n> input_text = \"[CLS] \" + question + \" [SEP] \" + text + \" [SEP]\"\r\n> input_ids = tokenizer.encode(input_text)\r\n> token_type_ids = [0 if i <= input_ids.index(102) else 1 for i in range(len(input_ids))] \r\n> \r\n> start_scores, end_scores = model(torch.tensor([input_ids]), token_type_ids=torch.tensor([token_type_ids]))\r\n> all_tokens = tokenizer.convert_ids_to_tokens(input_ids) \r\n> print(' '.join(all_tokens[torch.argmax(start_scores) : torch.argmax(end_scores)+1]))\r\n> # a nice puppet\r\n> ```\r\n> \r\n> Thanks huggingface for the cool stuff, although your documentation could be cooler :)\r\n\r\n@jeffxtang , thanks for sharing this. \r\nThere may be an issue with your output. For instance, question, text = \"Was Jim Henson a nice puppet?\", \"Jim Henson was a nice puppet\". You answer text could be part of question, because you are using the start_scores/end_scores of all_tokens. It is possible that highest score is within the question. \r\n\r\nThanks.\r\nLuke\r\n", "Thanks @luke4u but I think that's what the Squad-fine-tuned Bert model is supposed to do - its iOS version also returns \"Jim Henson was a nice puppet\" for the question \"Was Jim Henson a nice puppet?\", although ideally the answer should be simply \"yes\". My understanding is that answers returned by the model always have the highest start and end scores located in the text (not the question) - maybe @thomwolf or @julien-c can please verify this?", "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", "How I can further fine tune bert-large-uncased-whole-word-masking-finetuned-squad with our domain specific data set?", "@sanigam, please take a look at the [fine-tuning/training](https://huggingface.co/transformers/training.html) documentation. If you're having trouble, please open a new thread with your specific issue on the [forum](https://discuss.huggingface.co). Thanks!", "I tried using the suggested code for using BertForQuestionAnswering but got an error at the end\r\n<img width=\"980\" alt=\"Screenshot 2021-05-25 at 11 37 56 AM\" src=\"https://user-images.githubusercontent.com/27727185/119447528-a8611c00-bd4d-11eb-943e-fc7b64f70e30.png\">\r\n" ]

[]

[]

[ "Can you fix this sentence? It seems some error slipped in there\r\n\r\n> One of the primary differences I can think of is that now I am padding all up to max-seq length, and it does increase performance a lot decrease this.\r\n\r\nAs far as I know, you don't have to pad up to the max sequence length manually, and you can just pad up to the max sequence length _per batch_. That might save you some time.", "Yeah sorry I meant it increases performance a lot **to** decrease the max-seq-len.\r\n\r\nGood point.. I should definitely padding up to max length per batch, although I am not sure this will make huge difference as most of my inputs are of similar length and close to the max. \r\n\r\nI guess before I dive deeper I'm looking for a starting place into an investigation of why, say, the implementation of roberta here https://github.com/pytorch/fairseq/tree/master/examples/roberta would be 2x faster on the same GPU than the implementation in transformers.\r\n\r\nDoes transformers make a conscious performance sacrifice in the name of modularity and extensibility? Or are there specific optimizations in fairseq (for example) that I am observing that have not been ported.\r\n\r\nWould updating the new pytorch modules from 1.12 discussed in #1451 make a difference (it seems like there can be performance improvements by fusing kernels so pytorch requires fewer to run the same model, although I do not fully understand this https://pytorch.org/blog/optimizing-cuda-rnn-with-torchscript/)", "I am not sure about any of this, but I do remember that the PyTorch developers do an effort to implement as much parity between CPU and CUDA with specific optimisations for both. As an example, their C++ implementation for activation functions does specific things when MKL is available and otherwise. I'm not sure whether `nn.Transformer` and `nn.MultiheadAttention` got optimised intensively as well.", "@pertschuk these benchmarks are usually mostly dependant on stuff like data-processing, selected float precision, specific inference code (are you in a `torch.no_grad` context for instance) and basically all these things that are outside of the models themselves (which computational graphs are pretty much identical across frameworks).\r\n\r\nIf you have a (not too big) codebase for benchmarking and clear numbers, we can have a look.", "Yeah, thank you - I am not directly creating a torch.no_grad context, I (perhaps wrongly) assumed this would be handled with a call to .eval(). Also it seems that in the new release pretrained models are by default not loaded in a trainable state? Aka no grad? But perhaps I don't understand correctly.\r\n\r\n**Load the model** (takes ~10) and then trace **~2 seconds**\r\n```python \r\nself.model_config = config.from_pretrained(self.checkpoint, cache_dir=self.dir)\r\nself.model_config.num_labels = len(self.config.labels)\r\nself.model_config.torchscript = True\r\nself.model = model.from_pretrained(self.checkpoint, config=self.model_config,\r\n cache_dir=self.dir, **kwargs)\r\nself.tokenizer = tokenizer.from_pretrained(self.checkpoint, cache_dir=self.dir)\r\nself.model.eval()\r\nself.trace_model()\r\n```\r\nTrace function:\r\n```python\r\n def trace_model(self):\r\n examples = [\r\n InputExample(\r\n guid=1,\r\n text_a=\"Once upon a time there was a boy\",\r\n text_b=\"He liked to write code all day long\"\r\n )\r\n ]\r\n features = [self.example_to_feature(example) for example in examples]\r\n all_inputs = self.features_to_inputs(features, True)\r\n inputs = self.inputs_from_batch(all_inputs)\r\n self.model = torch.jit.trace(self.model, self.tuple_inputs(inputs))\r\n```\r\n\r\n**Run inference** Runs ~18/samples per second or ~2.25 batches (each call to run) with batch size = 8 (helper functions are below). Max_seq_len = 256:\r\n```python\r\n def run(self, *args):\r\n examples = [\r\n InputExample(\r\n guid=str(i),\r\n text_a=arg[0],\r\n text_b=None if len(arg) < 2 else arg[1]\r\n ) for i, arg in enumerate(zip(*args))\r\n ]\r\n features = [self.example_to_feature(example) for example in examples]\r\n all_inputs = self.features_to_inputs(features, True)\r\n inputs = self.inputs_from_batch(all_inputs)\r\n outputs = self.model(*self.tuple_inputs(inputs))\r\n return self.pred_from_output(outputs)\r\n```\r\n Convert examples to features:\r\n``` python\r\n def example_to_feature(self, example):\r\n inputs = self.tokenizer.encode_plus(\r\n example.text_a,\r\n example.text_b,\r\n add_special_tokens=True,\r\n max_length=self.max_length,\r\n truncate_first_sequence=True # We're truncating the first sequence in priority\r\n )\r\n input_ids, token_type_ids = inputs[\"input_ids\"][:self.max_length], \\\r\n inputs[\"token_type_ids\"][:self.max_length]\r\n\r\n attention_mask = [1] * len(input_ids)\r\n\r\n # Zero-pad up to the sequence length.\r\n if self.pad:\r\n padding_length = self.max_length - len(input_ids)\r\n if self.pad_on_left:\r\n input_ids = ([self.pad_token] * padding_length) + input_ids\r\n attention_mask = ([0] * padding_length) + attention_mask\r\n token_type_ids = ([self.pad_token_segment_id] * padding_length) + token_type_ids\r\n else:\r\n input_ids = input_ids + ([self.pad_token] * padding_length)\r\n attention_mask = attention_mask + ([0] * padding_length)\r\n token_type_ids = token_type_ids + ([self.pad_token_segment_id] * padding_length)\r\n\r\n if example.label is not None:\r\n if self.config.task == \"classification\":\r\n if example.label in self.label_map:\r\n label = self.label_map[example.label]\r\n else:\r\n logger.warning(\"UNKNOWN LABEL %s, ignoring\" % example.label)\r\n return\r\n elif self.config.task == \"regression\":\r\n label = float(example.label)\r\n else:\r\n logger.error(\"Only supported tasks are classification and regression\")\r\n raise NotImplementedError()\r\n else:\r\n label = None\r\n\r\n return InputFeatures(input_ids=input_ids,\r\n attention_mask=attention_mask,\r\n token_type_ids=token_type_ids,\r\n label=label)\r\n```\r\n\r\nConvert features to inputs:\r\n```python\r\n def features_to_inputs(self, features, inference):\r\n all_input_ids = torch.tensor([f.input_ids for f in features], dtype=torch.long).to(self.device)\r\n all_attention_mask = torch.tensor([f.attention_mask for f in features], dtype=torch.long).to(self.device)\r\n all_token_type_ids = torch.tensor([f.token_type_ids for f in features], dtype=torch.long).to(self.device)\r\n if not inference:\r\n if self.config.task == \"classification\":\r\n all_labels = torch.tensor([f.label for f in features], dtype=torch.long).to(self.device)\r\n elif self.config.task == \"regression\":\r\n all_labels = torch.tensor([f.label for f in features], dtype=torch.float).to(self.device)\r\n else:\r\n raise NotImplementedError()\r\n return all_input_ids, all_attention_mask, all_token_type_ids, all_labels\r\n else:\r\n return all_input_ids, all_attention_mask, all_token_type_ids\r\n```\r\nReturn inputs from batch:\r\n```python\r\n def inputs_from_batch(self, batch):\r\n inputs = {'input_ids': batch[0],\r\n 'attention_mask': batch[1]}\r\n if self.config.arch != 'distilbert':\r\n inputs['token_type_ids'] = batch[2] if self.config.arch in ['bert',\r\n 'xlnet'] else None\r\n if len(batch) > 3:\r\n inputs['labels'] = batch[3]\r\n return inputs\r\n```\r\nsource: https://github.com/koursaros-ai/koursaros/blob/master/koursaros/modeling/models/transformer_model.py", "I cleaned and consolidated my code with dynamic padding to current batch size and torch.no_grad() context. Output is below. It seems like the native fairseq/ torchub implementation is a little less than 2x as fast as transformers.\r\n\r\n```python\r\nimport transformers\r\nfrom fairseq.data.data_utils import collate_tokens\r\nimport time\r\nimport torch.nn.functional as F\r\nimport torch.hub\r\n\r\nMAX_LENGTH = 512\r\nPAD = True\r\n\r\n\r\ndef benchmark_mnli(samples):\r\n torch_hub_model = time_fn(torch.hub.load, 'pytorch/fairseq','roberta.large.mnli')\r\n torch_hub_model.eval()\r\n torch_hub_model.cuda()\r\n try:\r\n transformers_model = time_fn(transformers.RobertaModel.from_pretrained,\r\n 'roberta-large-mnli')\r\n except:\r\n transformers_model = time_fn(transformers.RobertaModel.from_pretrained,\r\n 'roberta-large-mnli', force_download=True)\r\n transformers_tokenizer = time_fn(transformers.RobertaTokenizer.from_pretrained, 'roberta-large-mnli')\r\n pred_functions = {\r\n 'transformers' : predict_transformers(transformers_model, transformers_tokenizer),\r\n 'torch_hub' : predict_roberta(torch_hub_model)\r\n }\r\n for framework, pred_fn in pred_functions.items():\r\n print(f'Benchmarking {framework} with {samples} samples')\r\n time_fn(benchmark, pred_fn, samples)\r\n\r\n\r\ndef predict_transformers(model, tokenizer):\r\n device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\r\n model = model.to(device)\r\n def predict_fn(*args):\r\n inputs = time_fn(transformers_encode_batch, tokenizer, *args)\r\n inputs_dict = {\r\n 'input_ids': torch.tensor(inputs[0], dtype=torch.long).to(device),\r\n 'attention_mask': torch.tensor(inputs[1], dtype=torch.long).to(device),\r\n # 'token_type_ids': torch.tensor(inputs[2], dtype=torch.long)\r\n }\r\n outputs = model(**inputs_dict)\r\n logits = outputs[0]\r\n preds = F.log_softmax(logits, dim=-1)\r\n return preds.tolist()\r\n return predict_fn\r\n\r\n\r\ndef predict_roberta(model):\r\n def pred_fn(*args):\r\n batch = time_fn(collate_tokens, [model.encode(*arg)[:MAX_LENGTH] for arg in zip(*args)], pad_idx=1)\r\n labels = model.predict('mnli', batch).tolist()\r\n return labels\r\n return pred_fn\r\n\r\n\r\ndef benchmark(pred_fn, n):\r\n args = ['All work and no play.'] * 8, ['Make jack a very dull boy.'] * 8\r\n for i in range(0, n):\r\n assert(type(pred_fn(*args)) == list)\r\n\r\n### HELPERS\r\n\r\ndef time_fn(fn, *args, **kwargs):\r\n start = time.time()\r\n res = fn(*args, **kwargs)\r\n print(f'Took {time.time() - start} seconds to run {fn.__name__}')\r\n return res\r\n\r\n\r\ndef transformer_to_features(tokenizer, *args):\r\n inputs = tokenizer.encode_plus(\r\n *args,\r\n add_special_tokens=True,\r\n max_length=MAX_LENGTH,\r\n truncate_first_sequence=True\r\n )\r\n input_ids = inputs[\"input_ids\"][:MAX_LENGTH]\r\n\r\n return input_ids\r\n\r\ndef pad_up(input_ids, max_length):\r\n padding_length = max_length - len(input_ids)\r\n input_ids = ([0] * padding_length) + input_ids\r\n attention_mask = ([0] * padding_length) + [1] * len(input_ids)\r\n return (input_ids, attention_mask)\r\n\r\n\r\ndef transformers_encode_batch(tokenizer, *args):\r\n assert(type(args[0]) == list)\r\n all_input_ids = []\r\n max_batch_len = 0\r\n\r\n for sample in zip(*args):\r\n input_ids = transformer_to_features(tokenizer, *sample)\r\n all_input_ids.append(input_ids)\r\n max_batch_len = max(max_batch_len, len(input_ids))\r\n\r\n all_input_ids, all_attention_masks = zip(*[\r\n pad_up(input_ids, max_batch_len) for input_ids in all_input_ids\r\n ])\r\n return all_input_ids, all_attention_masks\r\n\r\n\r\nif __name__ == '__main__':\r\n with torch.no_grad():\r\n benchmark_mnli(10)\r\n\r\n```\r\nHere is the output:\r\n```\r\nTook 11.221294641494751 seconds to run load\r\nTook 10.316125392913818 seconds to run from_pretrained\r\nTook 0.3631258010864258 seconds to run from_pretrained\r\nBenchmarking transformers with 10 samples\r\nTook 0.00434112548828125 seconds to run transformers_encode_batch\r\nTook 0.0039653778076171875 seconds to run transformers_encode_batch\r\nTook 0.003747701644897461 seconds to run transformers_encode_batch\r\nTook 0.0035974979400634766 seconds to run transformers_encode_batch\r\nTook 0.0037157535552978516 seconds to run transformers_encode_batch\r\nTook 0.003725767135620117 seconds to run transformers_encode_batch\r\nTook 0.0038688182830810547 seconds to run transformers_encode_batch\r\nTook 0.004169464111328125 seconds to run transformers_encode_batch\r\nTook 0.003767728805541992 seconds to run transformers_encode_batch\r\nTook 0.003550291061401367 seconds to run transformers_encode_batch\r\nTook 0.7687280178070068 seconds to run benchmark\r\nBenchmarking torch_hub with 10 samples\r\nTook 0.0001957416534423828 seconds to run collate_tokens\r\nTook 8.797645568847656e-05 seconds to run collate_tokens\r\nTook 6.890296936035156e-05 seconds to run collate_tokens\r\nTook 6.961822509765625e-05 seconds to run collate_tokens\r\nTook 6.914138793945312e-05 seconds to run collate_tokens\r\nTook 6.961822509765625e-05 seconds to run collate_tokens\r\nTook 7.05718994140625e-05 seconds to run collate_tokens\r\nTook 9.202957153320312e-05 seconds to run collate_tokens\r\nTook 6.961822509765625e-05 seconds to run collate_tokens\r\nTook 7.700920104980469e-05 seconds to run collate_tokens\r\nTook 0.4018120765686035 seconds to run benchmark\r\n```\r\nOr with a longer sample input:\r\n```\r\nTook 10.34562063217163 seconds to run load\r\nTook 10.523965835571289 seconds to run from_pretrained\r\nTook 0.4653303623199463 seconds to run from_pretrained\r\nBenchmarking transformers with 10 samples\r\nTook 0.007193565368652344 seconds to run transformers_encode_batch\r\nTook 0.005567789077758789 seconds to run transformers_encode_batch\r\nTook 0.005621671676635742 seconds to run transformers_encode_batch\r\nTook 0.006003141403198242 seconds to run transformers_encode_batch\r\nTook 0.0061550140380859375 seconds to run transformers_encode_batch\r\nTook 0.005508899688720703 seconds to run transformers_encode_batch\r\nTook 0.005594730377197266 seconds to run transformers_encode_batch\r\nTook 0.005545854568481445 seconds to run transformers_encode_batch\r\nTook 0.005563259124755859 seconds to run transformers_encode_batch\r\nTook 0.0059223175048828125 seconds to run transformers_encode_batch\r\nTook 1.5394785404205322 seconds to run benchmark\r\nBenchmarking torch_hub with 10 samples\r\nTook 0.0001571178436279297 seconds to run collate_tokens\r\nTook 9.131431579589844e-05 seconds to run collate_tokens\r\nTook 9.322166442871094e-05 seconds to run collate_tokens\r\nTook 8.7738037109375e-05 seconds to run collate_tokens\r\nTook 8.726119995117188e-05 seconds to run collate_tokens\r\nTook 8.726119995117188e-05 seconds to run collate_tokens\r\nTook 8.869171142578125e-05 seconds to run collate_tokens\r\nTook 8.96453857421875e-05 seconds to run collate_tokens\r\nTook 8.58306884765625e-05 seconds to run collate_tokens\r\nTook 8.869171142578125e-05 seconds to run collate_tokens\r\nTook 0.9851493835449219 seconds to run benchmark\r\n```\r\nI benchmarked the traced transformer model and it's about the same.", "You closed this, but I'm curious to hear about your result and thoughts. So fairseq/HUB implementation is twice as fast as the transformers implementation? Do you have any intuition about why?" ]

[ { "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", "try:\r\ndevice = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\r\nmodel_path = None\r\nmodel = my_model.load_state_dict(torch.load(model_path, map_location=device))" ]

[]

[]

[]

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

[ { "login": "rlouf", "id": 3885044, "node_id": "MDQ6VXNlcjM4ODUwNDQ=", "avatar_url": "https://avatars.githubusercontent.com/u/3885044?v=4", "gravatar_id": "", "url": "https://api.github.com/users/rlouf", "html_url": "https://github.com/rlouf", "followers_url": "https://api.github.com/users/rlouf/followers", "following_url": "https://api.github.com/users/rlouf/following{/other_user}", "gists_url": "https://api.github.com/users/rlouf/gists{/gist_id}", "starred_url": "https://api.github.com/users/rlouf/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/rlouf/subscriptions", "organizations_url": "https://api.github.com/users/rlouf/orgs", "repos_url": "https://api.github.com/users/rlouf/repos", "events_url": "https://api.github.com/users/rlouf/events{/privacy}", "received_events_url": "https://api.github.com/users/rlouf/received_events", "type": "User", "site_admin": false } ]

[ "Hi @chiyuzhang94 . What do you get if you remove everything related to `cls` in\r\n\r\n```\r\n pointer = getattr(pointer, 'cls')\r\n pointer = getattr(pointer, 'bias')\r\n elif l[0] == 'output_weights':\r\n pointer = getattr(pointer, 'cls')\r\n pointer = getattr(pointer, 'weight')\r\n```\r\n\r\n?", "> Hi @chiyuzhang94 . What do you get if you remove everything related to `cls` in\r\n> \r\n> ```\r\n> pointer = getattr(pointer, 'cls')\r\n> pointer = getattr(pointer, 'bias')\r\n> elif l[0] == 'output_weights':\r\n> pointer = getattr(pointer, 'cls')\r\n> pointer = getattr(pointer, 'weight')\r\n> ```\r\n> \r\n> ?\r\n\r\n", "Hi @rlouf \r\nWhen I removed `cls` parts, I got the following error: \r\n```\r\nI1009 17:18:05.159223 47655893610048 modeling_bert.py:81] Skipping bert/encoder/layer_9/output/dense/bias/adam_m\r\nI1009 17:18:05.159870 47655893610048 modeling_bert.py:81] Skipping bert/encoder/layer_9/output/dense/bias/adam_v\r\nI1009 17:18:05.160566 47655893610048 modeling_bert.py:115] Initialize PyTorch weight ['bert', 'encoder', 'layer_9', 'output', 'dense', 'kernel']\r\nI1009 17:18:05.161365 47655893610048 modeling_bert.py:81] Skipping bert/encoder/layer_9/output/dense/kernel/adam_m\r\nI1009 17:18:05.162042 47655893610048 modeling_bert.py:81] Skipping bert/encoder/layer_9/output/dense/kernel/adam_v\r\nI1009 17:18:05.162718 47655893610048 modeling_bert.py:115] Initialize PyTorch weight ['bert', 'pooler', 'dense', 'bias']\r\nI1009 17:18:05.163373 47655893610048 modeling_bert.py:81] Skipping bert/pooler/dense/bias/adam_m\r\nI1009 17:18:05.164020 47655893610048 modeling_bert.py:81] Skipping bert/pooler/dense/bias/adam_v\r\nI1009 17:18:05.164690 47655893610048 modeling_bert.py:115] Initialize PyTorch weight ['bert', 'pooler', 'dense', 'kernel']\r\nI1009 17:18:05.165374 47655893610048 modeling_bert.py:81] Skipping bert/pooler/dense/kernel/adam_m\r\nI1009 17:18:05.166048 47655893610048 modeling_bert.py:81] Skipping bert/pooler/dense/kernel/adam_v\r\nI1009 17:18:05.166695 47655893610048 modeling_bert.py:81] Skipping global_step\r\n---------------------------------------------------------------------------\r\nAttributeError Traceback (most recent call last)\r\n<ipython-input-1-a2cbceaf2173> in <module>\r\n 15 \r\n 16 # Load weights from tf checkpoint\r\n---> 17 load_tf_weights_in_bert(model, config, tf_checkpoint_path)\r\n 18 \r\n 19 # Save pytorch-model\r\n\r\n~/py3.6/lib/python3.6/site-packages/transformers-2.1.0-py3.6.egg/transformers/modeling_bert.py in load_tf_weights_in_bert(model, config, tf_checkpoint_path)\r\n 90 pointer = getattr(pointer, 'weight')\r\n 91 elif l[0] == 'output_bias' or l[0] == 'beta':\r\n---> 92 pointer = getattr(pointer, 'bias')\r\n 93 elif l[0] == 'output_weights':\r\n 94 pointer = getattr(pointer, 'weight')\r\n\r\n~/py3.6/lib/python3.6/site-packages/torch/nn/modules/module.py in __getattr__(self, name)\r\n 589 return modules[name]\r\n 590 raise AttributeError(\"'{}' object has no attribute '{}'\".format(\r\n--> 591 type(self).__name__, name))\r\n 592 \r\n 593 def __setattr__(self, name, value):\r\n```", "Hi @chiyuzhang94, as a side question which TensorFlow version did you use to train your bert model ?\r\n\r\nDo you observe the same behavior by loading the .index file directly using:\r\n\r\n```python\r\nconfig = BertConfig.from_json_file('your/tf_model/config.json')\r\nmodel = BertForSequenceClassification.from_pretrained('your/tf_model/xxxx.ckpt.index', from_tf=True, config=config)\r\n```", "Hi @mfuntowicz ,\r\nI trained the model with tensorflow 1.12.0.\r\n\r\nMy am currently using tensorflow 1.13.1 and torch 1.2.0 for this converting task. \r\n\r\nIf I use your suggestion code, it is also same issue:\r\n```\r\n---------------------------------------------------------------------------\r\nAttributeError Traceback (most recent call last)\r\n<ipython-input-2-35a5c9b35e78> in <module>\r\n 12 \r\n 13 print(\"Building PyTorch model from configuration: {}\".format(str(config)))\r\n---> 14 model = BertForSequenceClassification.from_pretrained(tf_checkpoint_path, from_tf=True, config=config)\r\n 15 # Load weights from tf checkpoint\r\n 16 # load_tf_weights_in_bert(model, config, tf_checkpoint_path)\r\n\r\n~/py3.6/lib/python3.6/site-packages/transformers-2.1.0-py3.6.egg/transformers/modeling_utils.py in from_pretrained(cls, pretrained_model_name_or_path, *model_args, **kwargs)\r\n 352 if resolved_archive_file.endswith('.index'):\r\n 353 # Load from a TensorFlow 1.X checkpoint - provided by original authors\r\n--> 354 model = cls.load_tf_weights(model, config, resolved_archive_file[:-6]) # Remove the '.index'\r\n 355 else:\r\n 356 # Load from our TensorFlow 2.0 checkpoints\r\n\r\n~/py3.6/lib/python3.6/site-packages/transformers-2.1.0-py3.6.egg/transformers/modeling_bert.py in load_tf_weights_in_bert(model, config, tf_checkpoint_path)\r\n 90 pointer = getattr(pointer, 'weight')\r\n 91 elif l[0] == 'output_bias' or l[0] == 'beta':\r\n---> 92 pointer = getattr(pointer, 'bias')\r\n 93 elif l[0] == 'output_weights':\r\n 94 pointer = getattr(pointer, 'weight')\r\n\r\n~/py3.6/lib/python3.6/site-packages/torch/nn/modules/module.py in __getattr__(self, name)\r\n 589 return modules[name]\r\n 590 raise AttributeError(\"'{}' object has no attribute '{}'\".format(\r\n--> 591 type(self).__name__, name))\r\n 592 \r\n 593 def __setattr__(self, name, value):\r\n\r\nAttributeError: 'BertForSequenceClassification' object has no attribute 'bias'\r\n```", "This happens because you are trying to load weights the functions wasn't designed for. Unfortunately we cannot support every possible file. You will have to modify `modeling_bert.py` manually to support your file. The part you need to modify is:\r\n\r\n```\r\n for name, array in zip(names, arrays):\r\n name = name.split('/')\r\n # adam_v and adam_m are variables used in AdamWeightDecayOptimizer to calculated m and v\r\n # which are not required for using pretrained model\r\n if any(n in [\"adam_v\", \"adam_m\", \"global_step\"] for n in name):\r\n logger.info(\"Skipping {}\".format(\"/\".join(name)))\r\n continue\r\n pointer = model\r\n for m_name in name:\r\n if re.fullmatch(r'[A-Za-z]+_\\d+', m_name):\r\n l = re.split(r'_(\\d+)', m_name)\r\n else:\r\n l = [m_name]\r\n if l[0] == 'kernel' or l[0] == 'gamma':\r\n pointer = getattr(pointer, 'weight')\r\n elif l[0] == 'output_bias' or l[0] == 'beta':\r\n pointer = getattr(pointer, 'bias')\r\n elif l[0] == 'output_weights':\r\n pointer = getattr(pointer, 'weight')\r\n elif l[0] == 'squad':\r\n pointer = getattr(pointer, 'classifier')\r\n else:\r\n try:\r\n pointer = getattr(pointer, l[0])\r\n except AttributeError:\r\n logger.info(\"Skipping {}\".format(\"/\".join(name)))\r\n continue\r\n if len(l) >= 2:\r\n num = int(l[1])\r\n pointer = pointer[num]\r\n if m_name[-11:] == '_embeddings':\r\n pointer = getattr(pointer, 'weight')\r\n elif m_name == 'kernel':\r\n array = np.transpose(array)\r\n try:\r\n assert pointer.shape == array.shape\r\n except AssertionError as e:\r\n e.args += (pointer.shape, array.shape)\r\n raise\r\n logger.info(\"Initialize PyTorch weight {}\".format(name))\r\n pointer.data = torch.from_numpy(array)\r\n```", "Hi @rlouf ,\r\nThanks for your answer. I think adding `pointer = getattr(pointer, 'cls')` to the two if-so section make sense. But I am wondering how I can deal with the question of 'LayerNorm' object has no attribute 'cls'. Could you please provide me any hint?", "Separating out the conditional statement:\r\n`elif l[0] == 'output_bias' or l[0] == 'beta':`\r\ninto two, while maintaining the original functionality in the beta conditional should work?", "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" ]

[]

[]

[ "Yes we have observed there was a difference in tokenization. We've temporarily fixed in 036483f, could you install from source and tell us if you manage to have good generations?\r\n\r\nBy following the recommended specs (temperature=0.2, top_k=5 and repetition_penalty=1.2), with the following input sentence: `Reviews Rating 4.0`, I obtained the following completion:\r\n\r\nReviews Rating 4.0 < GENERATION > \"out of 5 stars. I received a copy from the author in exchange for an honest review. \r\nRating: 4.0 \r\n This is one book that you will not want to put down. It was very well written and kept me on my toes. The characters were so real it made them seem like people you know. You could feel their pain as they struggled with what happened to them. There are some twists and turns along the way but nothing too surprising. All in all this story had everything needed to make it a great book. If\"\r\n", "Looks like that fixed it. Getting a similar completion to what you got for `Reviews Rating 4.0`:\r\n\r\n```\r\nReviews Rating 4.0 out of 5 starsI received a copy from the author in exchange for an honest review. \r\n Rating: 4.0 \r\n This was a great book that kept me interested and wanting to read more. It is about two people who have been together forever but are not\r\n```\r\n\r\nAnd for `Links https://www.cnn.com/2018/09/20/us-president-meets-british-pm`, I get:\r\n\r\n```\r\nLinks https://www.cnn.com/2018/09/20/us-president-meets-british-pm (CNN)President Donald Trump said Friday he would meet with British Prime Minister Theresa May in Washington next week, a move that could help ease tensions between the two countries after months of escalating trade tensions. \r\n \r\n The White House announced Trump's decision to hold talks\r\n```\r\n\r\nShould I leave this issue open, since you mentioned this is a temporary fix?", "Following deeper investigations, this temporary solution is actually the correct one for CTRL.\r\n\r\nSee details in #1480." ]

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

[]

[ "https://github.com/huggingface/transformers/issues/940", "@ahotrod you have any fix for this bug?", "> @ahotrod you have any fix for this bug?\r\n\r\n@a-maci no unfortunately not, still searching. I'm considering rolling back to Transformers 2.0.0 or even pytorch-transformers 1.2.0, one or both of which didn't spawn this error in my earlier SQuAD replications.", "> @ahotrod you have any fix for this bug?\r\n\r\n@a-maci I needed XLNet fine-tuned on SQuAD 2.0 with 512 max_seq_length. I found \"**A**\" solution: went back to the original XLNet paper's github for the \"native\" code. I could fit 1 batch on each of (2) 1080Ti GPUs, 85,000 steps, ~14.5 hr of fine-tuning with results EM / F1: 84.5 / 87.1.\r\n\r\n`INFO:tensorflow:Result | best_exact 84.52792049187232 | best_exact_thresh -2.716632127761841 | best_f1 87.12844471348052 | best_f1_thresh -2.447098970413208 | has_ans_exact 0.8733130904183536 | has_ans_f1 0.9327569452896122 | `\r\n\r\nPossibly try the BERT paper's \"native\" code?", "I've described the bug here: https://github.com/huggingface/transformers/issues/940#issuecomment-547686206\r\n\r\nWorkaround is either to use DataParallel (remove `-m torch.distributed.launch --nproc_per_node=8`) or don't eval in the same run (remove `--do_eval`). You can evaluate the model after training with:\r\n\r\n```\r\npython examples/run_squad.py \\\r\n--model_type bert \\\r\n--model_name_or_path bert-base-cased \\\r\n--do_eval \\\r\n--do_lower_case \\\r\n--predict_file $SQUAD_DIR/dev-v1.1.json \\\r\n--train_file $SQUAD_DIR/train-v1.1.json \\\r\n--output_dir ./models/wwm_uncased_finetuned_squad/\r\n```", "As mentioned in #940, happy to welcome a PR to fix this case if someone from the community wants to contribute (I don't have the bandwidth for this issue at the moment).", "Maybe try changing `args.local_rank == -1` to `args.local_rank in [-1, 0]` at this line? https://github.com/huggingface/transformers/blob/079bfb32fba4f2b39d344ca7af88d79a3ff27c7c/examples/run_squad.py#L216\r\n\r\nI think evaluate is only used in the main process (local_rank==0) if you're using multiple gpus\r\n(reference: https://github.com/huggingface/transformers/blob/079bfb32fba4f2b39d344ca7af88d79a3ff27c7c/examples/run_squad.py#L543)", "It makes more sense to just remove the `DistributedSampler` case entirely. The problem is that `all_results` doesn't get gathered from all GPUs. Unless you also implement a gather you shouldn't use `DistributedSampler` at all.", "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", "Is there a fix for this ? Im seeing the same issue for running only evaluation on CPU too. ", "Are you trying to do multiprocess evaluation? A single CPU process should work, my WAR above is to run eval seperately as a single process. " ]

[]

[]

[ "You're right, the interface isn't well suited to mess with settings on mobile. It's on our roadmap!", "awesome! I found a dumb workaround; saved a copy of the page, changed the default values, then put it in my dropbox!", "Haha that’s a great hack!\r\n\r\nClosing this for now, thanks" ]

[]

[]

[ "Duplicate of #1370" ]

[]

[]

[ "With 4 GB you're bound to issues with a batch size of 12. You could figure out the total memory usage of the model + calculate the memory footprints of tensors to determine the biggest batch size that would fit on your GPU. \r\n\r\nSpecifying a smaller batch size (like 1 or 2) would let you run the script, though.", "> With 4 GB you're bound to issues with a batch size of 12. You could figure out the total memory usage of the model + calculate the memory footprints of tensors to determine the biggest batch size that would fit on your GPU.\r\n> \r\n> Specifying a smaller batch size (like 1 or 2) would let you run the script, though.\r\n\r\nHello @LysandreJik I am able to run the code with `batch size = 1`\r\n\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 } ]

[]

[ "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 } ]

[]

[ "Hi, what do you mean by different CPU systems? Do you mean that you tried it with different CPU architectures like ARM/x86? On which CPU did it fail?", "Thank you so much for your reply sir.\n** Different CPUs means** - normal CPU system it is not working. I tested two more CPU system for inferencing own model. It was arising this issue. \n\nCPU architecture is normal Ubuntu 16.04 & 18.04 64 bit 8 GB RAM. \n\nI don't know much about CPU architecture sir. :relaxed:", "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 } ]

[]

[ "Have you figured out what the problem is? I'm facing the same thing....", "Have you figured out what the problem is? I'm facing the same thing....", "Hello, did you figure out the solution here?", "Hey guys, sorry for the late update. Here's my solution: I set a lower learning rate and the problem is fixed. It seems that when we do transfer learning, we cannot set a high learning rate because the model is not well connected to the softmax layer you add.(Just some intuition) In addition, it's also possible that you forget to call model.eval() to invalidate the Dropout layer or something. But this is not the case for me.", "> Hey guys, sorry for the late update. Here's my solution: I set a lower learning rate and the problem is fixed. It seems that when we do transfer learning, we cannot set a high learning rate because the model is not well connected to the softmax layer you add.(Just some intuition) In addition, it's also possible that you forget to call model.eval() to invalidate the Dropout layer or something. But this is not the case for me.\r\n\r\nDo you also use TFBertForSequenceClassification for multi-label classification?Multi-label classification requires sigmoid function.", "> Hey guys, sorry for the late update. Here's my solution: I set a lower learning rate and the problem is fixed. It seems that when we do transfer learning, we cannot set a high learning rate because the model is not well connected to the softmax layer you add.(Just some intuition) In addition, it's also possible that you forget to call model.eval() to invalidate the Dropout layer or something. But this is not the case for me.\r\n\r\nI used your method but it was unsuccessful", "> > Hey guys, sorry for the late update. Here's my solution: I set a lower learning rate and the problem is fixed. It seems that when we do transfer learning, we cannot set a high learning rate because the model is not well connected to the softmax layer you add.(Just some intuition) In addition, it's also possible that you forget to call model.eval() to invalidate the Dropout layer or something. But this is not the case for me.\r\n> \r\n> Do you also use TFBertForSequenceClassification for multi-label classification?Multi-label classification requires sigmoid function.\r\n\r\nTypically, **sigmoid** function is used in binary classification problems, instead **softmax** function is used in multi-class classification problems", "> > > Hey guys, sorry for the late update. Here's my solution: I set a lower learning rate and the problem is fixed. It seems that when we do transfer learning, we cannot set a high learning rate because the model is not well connected to the softmax layer you add.(Just some intuition) In addition, it's also possible that you forget to call model.eval() to invalidate the Dropout layer or something. But this is not the case for me.\r\n> > \r\n> > \r\n> > Do you also use TFBertForSequenceClassification for multi-label classification?Multi-label classification requires sigmoid function.\r\n> \r\n> Typically, **sigmoid** function is used in binary classification problems, instead **softmax** function is used in multi-class classification problems\r\nI use the sigmoid function, but the output of different words is the same\r\n", "@thomwolf Is this problem solved", "I have facing the same issue", "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", "Try adjusting learning rate, the dropout probability(`config.hidden_dropout_prob`) and batch size.", "@venkatasg Can you share the hyperparameters that worked for your experiments? Trying to get a sense of order-of-magnitude for each of those.", "The problem is probably that the model is overfitting to the data. For BERT, the following hyperparameters worked for me:\r\n\r\n- batch size: 16/32\r\n- learning rate: 1e-5, 2e-5\r\n- dropouts: 0.1\r\n- weight decay: 0\r\n\r\nHowever, what works for me might not work for you. Keep experimenting with the hyperparameters(and random seeds)", "I just finished a multi label classifier training and got the exact same result:\r\n\r\n> same output no matter the input that I put in\r\n", "I've been facing a similar issue as described here - I got a good accuracy but the predictions just would not make any sense.\r\n\r\nLuckily, I've received help from the [HuggingFace.co community](https://discuss.huggingface.co/t/fine-tune-for-multiclass-or-multilabel-multiclass/4035) and it turned out that one has to initialize the model with the correct labels because the model is otherwise learning something but its just not clear what numeric label represents what string...\r\n\r\n`bert = TFAutoModel.from_pretrained(tranformersPreTrainedModelName, label2id=label2Index, id2label=index2label)\r\n`\r\n\r\nThe full code for the solution that works for me with public data can [be found here](https://github.com/Dirkster99/PyNotes/blob/master/Transformers/LocalModelUsage_Finetuning/66_Transformer_4_Language_Classification_MultiClass.ipynb).\r\n\r\nHope this helps..." ]

[ { "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 on our mid-term roadmap (seq2seq models).", "@Hannabrahman In the original GPT2 paper (section 3.7 Translation) the authors used the format \"english sentence = french sentence\" to produce translations. You can definitely fine tune the model using this format to produce translations using the existing scripts if you structure your seq2seq data this way.", "@dvaltchanov and @thomwolf thanks for pointing out to me.\r\nDo you think for that, I need to pass another input to the forward method of GPTLMHead method which is a list containing the length of source sequence, so that I will be able to zero out the loss calculated for the tokens in source?\r\nI mean did I have to zero out the lm_logits associated with source sequence tokens so that I do not count them in loss calculation?\r\n\r\nOr it doesn't matter if we include the source tokens loss in our total loss?", "@Hannabrahman Based on my tests, it doesn't matter if you include them. Your total loss will be higher but you're mainly interested in the validation loss on the translations anyway. As long as you use the \"start of text\" and \"end of text\" tokens to wrap your \"sequence = sequence\" text the model seems to be able to figure it out after a little bit of fine tuning.", "@dvaltchanov Thanks. \r\nJust one question since you had experimented this. \r\nI want to finetune gpt on a new dataset using the format you said and [this script.](https://github.com/huggingface/transformers/blob/master/examples/run_lm_finetuning.py) which is for finetuning pretained model on new dataset.\r\n\r\n1- should I add special tokens ( [SOS], some separator token for source and target, [EOS]) and train it like this:\r\n```\r\n# Add a [SOS], [SEP] and [EOS] to the vocabulary (we should train it also!)\r\n tokenizer.add_special_tokens({'start_token': '[CLS]', 'sep_token': '[SEP]', 'end_token': '[EOS]'})\r\n model.resize_token_embeddings(len(tokenizer)) # Update the model embeddings with the new vocabulary size\r\n```\r\n\r\n2- The instances in my dataset have different length ( 60-85 tokens). I have to either trim them to be the same size (it is not really good for my usecase), or use padding to pad them to same size. However, I read somewhere in this repo that gpt and gpt-2 doesnt handle right padding, how did you solve this issue while finetuning gpt on your own usecase and dataset?\r\n\r\nMany thanks in advance.", "@Hannabrahman Great questions:\r\n\r\n1. This is up to you. The model can learn the sequence of known tokens (e.g. \"[\", \"E\", \"OS\", \"]\") and use that as a prompt. I used a sequence and found that it worked well enough so I did not try adding extra tokens. There is already an \"<|endoftext|>\" token in the vocabulary which you can leverage.\r\n\r\n2. I created a custom data loader which concatenated the desired sample with randomly selected sequences from the data up to the desired length. E.g., A training sample may be a concat of sample translation #1 and #32 which would look like this: \"[SOS] something in English_#1 = something in French_#1 [EOS] [SOS] something in English_#32 = something in French_#32 [EOS] [SOS] .. etc\" \r\n\r\nThis then gets tokenized and truncated to the max length. This will allow the model to learn variable length sequences.\r\n\r\nYou can accomplish the same effect by concatenating all of your text into a single string and sampling sections of it. However, if you do this the model will learn associations between neighbouring samples over multiple epochs, so I recommend having something that shuffles the order of concatenated samples each epoch.\r\n\r\nDuring generation you prompt with \"[SOS] something in English = \" and stop generating when it produces an [EOS] token.\r\n", "@dvaltchanov \r\nregarding 2 - I didn't get it completely. \r\nWhere is the padding in your given batch example? Also, did you mean you concat all the instances back to back to create a single instance when you have #32 after #1 or #32 is probably another instance in the same batch? that being said the input is [bs, max_seq_len]? (bs = 2 in this example)\r\nAlso did you add a [pad] token to the vocabulary? because gpt and gpt2 doesnt have padding token. Or you follow the same strategy as in question 1\r\n\r\nDo you have your custom data loader code somewhere so that I can take a look?", "@Hannabrahman See my edited response above. I hope my clarification helps. ", "@dvaltchanov Thankss. Basically you followed the same approach as in [here](https://github.com/huggingface/transformers/blob/master/examples/run_lm_finetuning.py) . They read all the input into one long string and then truncate it in max_len. However it doesn't have any sampling or shuffling.\r\nMy data is stories and each story is around 60-80 tokens. I read all the stories in one long string and truncate each section to 128 tokens. The problem is sometimes the beginning of an story may goes into previous sample section. and the rest goes in to next section.", "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", "Hi, is there a seq2seq example of GPT2 now?", "Hi, any updates?", "Hi everyone,\r\n\r\nGiven that [Alpaca](https://github.com/tatsu-lab/stanford_alpaca) (decoder-only model like GPT) was trained in a seq2seq manner, I realised we can learn from their code (cheers to OS!).\r\n\r\n## Approach\r\nThe naive solution is to concatenate the source and target strings. However, the main issue here is that the loss is incurred in the next-word-prediction of the source strings. \r\n\r\nTo circumvent this, [Alpaca](https://github.com/tatsu-lab/stanford_alpaca/blob/main/train.py) simply ignored the loss in the source strings. Concretely:\r\n\r\n```\r\ndef preprocess(\r\n sources: Sequence[str],\r\n targets: Sequence[str],\r\n tokenizer: transformers.PreTrainedTokenizer,\r\n) -> Dict:\r\n \"\"\"Preprocess the data by tokenizing.\"\"\"\r\n examples = [s + t for s, t in zip(sources, targets)] # concatenate source and target strings\r\n examples_tokenized, sources_tokenized = [_tokenize_fn(strings, tokenizer) for strings in (examples, sources)]\r\n input_ids = examples_tokenized[\"input_ids\"]\r\n labels = copy.deepcopy(input_ids)\r\n for label, source_len in zip(labels, sources_tokenized[\"input_ids_lens\"]):\r\n label[:source_len] = IGNORE_INDEX # the source string's loss is ignored with IGNORE_INDEX\r\n return dict(input_ids=input_ids, labels=labels)\r\n\r\n```\r\n\r\nNote how the source string's loss is ignored with `IGNORE_INDEX`\r\n\r\n## Implications\r\n\r\n**Seq2Seq prompting.**\r\n\r\nIn concatenating the source and target strings, it may not be obvious to the model how to differentiate the source from target strings. I suspect that Alpaca/self-instruct circumvented this by making the differentiation clear via prompts:\r\n\r\n```\r\nPROMPT_DICT = {\r\n \"prompt_input\": (\r\n \"Below is an instruction that describes a task, paired with an input that provides further context. \"\r\n \"Write a response that appropriately completes the request.\\n\\n\"\r\n \"### Instruction:\\n{instruction}\\n\\n### Input:\\n{input}\\n\\n### Response:\"\r\n ),\r\n \"prompt_no_input\": (\r\n \"Below is an instruction that describes a task. \"\r\n \"Write a response that appropriately completes the request.\\n\\n\"\r\n \"### Instruction:\\n{instruction}\\n\\n### Response:\"\r\n ),\r\n}\r\n```\r\n\r\nNotice how `### Instruction:` tells the model where the source string is while `### Response:` tells the model where the target string is.\r\n\r\n**Increased GPU Memory usage**. To my understanding, the `input` and `labels` will now both be the concatenated source and target strings. In contrast for seq2seq models, the `input` will only be the source strings while the `labels` will only be the target strings. Thus this neat trick incurs additional GPU memory.\r\n\r\n**Packing is more intuitive with causal LM.** Packing is the act of packing training examples together to avoid padding. In causal LM, we can pack via\r\n\r\n```\r\n(source->target)[IGNORE_INDEX](source->target)[IGNORE_INDEX]...(source->target)[IGNORE_INDEX])\r\n```\r\n\r\nNotice how the target string immediately comes after the source. In contrast, packing for seq2seq LM will look like\r\n\r\n```\r\nInput: (source)[IGNORE_INDEX](source)[IGNORE_INDEX]...(source)[IGNORE_INDEX]\r\nTarget: (target)[IGNORE_INDEX](target)[IGNORE_INDEX]...(target)[IGNORE_INDEX]\r\n```\r\n\r\nTo me, it's not intuitive that the model can match the ith target to the ith source string. \r\n\r\n## Credits\r\nCheers to Alpaca, LlaMMA, and OS for finally solving this engineering puzzle for me! Do LMK if any parts don't make sense to you - I'm still learning myself.", "Created training examples by concatenating inputs and targets like this: 'Document:{document}\\nSummary:{Summary}'\r\nand created text summary model with this. But the problem here is the model starts generating from Document not from Summary. Would be there anyway to handle this problem?" ]

[]

[]

[ "Hello, you should take a look at the `encode` and `decode` methods in the [documentation](https://huggingface.co/transformers/main_classes/tokenizer.html#transformers.PreTrainedTokenizer.decode).", "Thank you so much that was very helpful", "Glad I could help!", "Hi \r\nI am trying to use the code in this link (https://colab.research.google.com/drive/1pS-eegmUz9EqXJw22VbVIHlHoXjNaYuc#scrollTo=JggjeDC9m2MH) to plot my trained model\r\nBut, I am getting an error. any idea please??\r\n\r\n\r\n\r\n\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 } ]

[]

[ "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 } ]

[]

[ "Exactly the same question here! Can someone please provide us with a small tutorial or even some general guidelines?", "Any response here? I was looking for something similar", "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" ]

[]

[]

[ "Hi, I believe that means that the decoder is a linear layer that has the same weights as the word embedding matrix. However, that embedding matrix does not have a bias, whereas the decoder does have a bias. \r\n\r\nIt is initialized to a vector of zeros here, but it can update its weights during training and has actual values in pre-trained models. For example:\r\n\r\n```py\r\nfrom transformers import BertForMaskedLM\r\n\r\nbert = BertForMaskedLM.from_pretrained(\"bert-base-cased\")\r\nprint(bert.cls.predictions.bias)\r\n\r\n# tensor([-0.1788, -0.1758, -0.1752, ..., -0.3448, -0.3574, -0.3483], requires_grad=True)\r\n```", "Oh, I see. That makes sense. So it should share parameters with the embedding weights? Where is that enforced in the code?", "Yes, exactly. You can see it in the [bert_modeling.py file, inside the BertForMaskedLM class](https://github.com/huggingface/transformers/blob/master/transformers/modeling_bert.py#L754-L759).", "Thanks for the clarification!", "hi, \r\nI know it's an old issue but I had the same questions:\r\n* it's not clear to me in the code where this parameter sharing is enforced\r\n* is there any intuition why it is done ? \r\n\r\nthanks in advance ", "@thibault-formal Hey, I had similar questions and asked them in the huggingface forum, I think my [post there](https://discuss.huggingface.co/t/understanding-bertlmpredictionhead/3618) could be helpful for you (if its still relevant). Your first point should be adressed by the explanation of my understanding and the second one is adressed by both replies.\r\nCheers!" ]

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

[]

[ "Getting the same error here", "Will investigate, thanks for reporting. (And Hi, @louismartin :)", "I had a similar issue trying to load BioBERT and I figured out what was going on in my case, sharing just in case that's what's going on in your case. In my case I converted TF BioBERT checkpoint to pytorch model. In my case the (first) problem was that I didn't provide a path to the config file. \r\n\r\nMy local scripts are adapted from the python code that runs Glue. I have a `--config_name` parameter that specifies the json file from which the configuration is loaded. If you don't provide that one, it tries to infer it by using the model_name_or_path - and that's what caused my problem. \r\n\r\nOnce I specified the config file, I had another problem that had to do with the following: `model_name_or_path` is supposed to be the path where you store the other info for the models and the model files are expected to follow a certain naming convention (e.g., in my case, it was looking for a `pytorch_model.bin` file; there is a similar file name for TF). \r\n\r\nHope this helps!\r\n\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", "This was superseded, and closed, by #1512" ]

[]

[]

[ "Hi, indeed GPT-2 has a slightly different implementation than BERT. In order to have access to the embeddings, you would have to do the following:\r\n\r\n```py\r\nfrom transformers import GPT2LMHeadModel\r\n\r\nmodel = GPT2LMHeadModel.from_pretrained('gpt2') # or any other checkpoint\r\nword_embeddings = model.transformer.wte.weight # Word Token Embeddings \r\nposition_embeddings = model.transformer.wpe.weight # Word Position Embeddings \r\n```", "> Hi, indeed GPT-2 has a slightly different implementation than BERT. In order to have access to the embeddings, you would have to do the following:\r\n> \r\n> ```python\r\n> from transformers import GPT2LMHeadModel\r\n> \r\n> model = GPT2LMHeadModel.from_pretrained('gpt2') # or any other checkpoint\r\n> word_embeddings = model.transformer.wte.weight # Word Token Embeddings \r\n> position_embeddings = model.transformer.wpe.weight # Word Position Embeddings \r\n> ```\r\n\r\nHi,\r\n\r\nThank you for your reply! So if I want to get the vector for 'man', it would be like this:\r\n\r\n>tokenizer = GPT2Tokenizer.from_pretrained('gpt2')\r\n>text_index = tokenizer.encode('man',add_prefix_space=True)\r\n>vector = model.transformer.wte.weight[text_index,:]\r\n\r\nIs it correct?\r\n", "Just wondering, how to transform word_vector to word? Imagine a word vector and change a few elements, how can I find closest word from gpt2 model?", "> Just wondering, how to transform word_vector to word? Imagine a word vector and change a few elements, how can I find closest word from gpt2 model?\r\n\r\nSo for each token in dictionary there is a static embedding(on layer 0). You can use cosine similarity to find the closet static embedding to the transformed vector. That should help you find the word.", "> > Just wondering, how to transform word_vector to word? Imagine a word vector and change a few elements, how can I find closest word from gpt2 model?\r\n> \r\n> So for each token in dictionary there is a static embedding(on layer 0). You can use cosine similarity to find the closet static embedding to the transformed vector. That should help you find the word.\r\n\r\nThanks. It means that for every word_vector I have to calculate vocab_size (~50K) cosine_sim manipulation. Is that right? ", "> > > Just wondering, how to transform word_vector to word? Imagine a word vector and change a few elements, how can I find closest word from gpt2 model?\r\n> > \r\n> > \r\n> > So for each token in dictionary there is a static embedding(on layer 0). You can use cosine similarity to find the closet static embedding to the transformed vector. That should help you find the word.\r\n> \r\n> Thanks. It means that for every word_vector I have to calculate vocab_size (~50K) cosine_sim manipulation. Is that right?\r\n\r\nI guess so. Unless you can use some property to first tighten the range.", "> > > > Just wondering, how to transform word_vector to word? Imagine a word vector and change a few elements, how can I find closest word from gpt2 model?\r\n> > > \r\n> > > \r\n> > > So for each token in dictionary there is a static embedding(on layer 0). You can use cosine similarity to find the closet static embedding to the transformed vector. That should help you find the word.\r\n> > \r\n> > \r\n> > Thanks. It means that for every word_vector I have to calculate vocab_size (~50K) cosine_sim manipulation. Is that right?\r\n> \r\n> I guess so. Unless you can use some property to first tighten the range.\r\n\r\nOk. Three more questions, 1) is there any resource on how to generate fixed length sentence (a sentence with N words that ends with \".\" or \"!\" )? 2) what is the most effective underlying parameter for hyper-parameter tuning (eg. Temperature)? 3) Is there any slack channel to discuss these types of questions? ", "> > > > > Just wondering, how to transform word_vector to word? Imagine a word vector and change a few elements, how can I find closest word from gpt2 model?\r\n> > > > \r\n> > > > \r\n> > > > So for each token in dictionary there is a static embedding(on layer 0). You can use cosine similarity to find the closet static embedding to the transformed vector. That should help you find the word.\r\n> > > \r\n> > > \r\n> > > Thanks. It means that for every word_vector I have to calculate vocab_size (~50K) cosine_sim manipulation. Is that right?\r\n> > \r\n> > \r\n> > I guess so. Unless you can use some property to first tighten the range.\r\n> \r\n> Ok. Three more questions, 1) is there any resource on how to generate fixed length sentence (a sentence with N words that ends with \".\" or \"!\" )? 2) what is the most effective underlying parameter for hyper-parameter tuning (eg. Temperature)? 3) Is there any slack channel to discuss these types of questions?\r\n\r\nabout 1) I don't think that there is any. You can use Web Scraping for such specified sentences. Also, you can download a corpus and use Regex to extract desired sentences.\r\n\r\n2) I don't really know\r\n\r\n3) If you find any, please share it with me too. Thanks! 😄 ", "> > Hi, indeed GPT-2 has a slightly different implementation than BERT. In order to have access to the embeddings, you would have to do the following:\r\n> > ```python\r\n> > from transformers import GPT2LMHeadModel\r\n> > \r\n> > model = GPT2LMHeadModel.from_pretrained('gpt2') # or any other checkpoint\r\n> > word_embeddings = model.transformer.wte.weight # Word Token Embeddings \r\n> > position_embeddings = model.transformer.wpe.weight # Word Position Embeddings \r\n> > ```\r\n> \r\n> Hi,\r\n> \r\n> Thank you for your reply! So if I want to get the vector for 'man', it would be like this:\r\n> \r\n> > tokenizer = GPT2Tokenizer.from_pretrained('gpt2')\r\n> > text_index = tokenizer.encode('man',add_prefix_space=True)\r\n> > vector = model.transformer.wte.weight[text_index,:]\r\n> \r\n> Is it correct?\r\n\r\nDid you succeed? I'm pursuing the same goal and I don't know how to validate my findings. I have tested some king - man + woman stuff, but it didn't work.", "> > > Hi, indeed GPT-2 has a slightly different implementation than BERT. In order to have access to the embeddings, you would have to do the following:\r\n> > > ```python\r\n> > > from transformers import GPT2LMHeadModel\r\n> > > \r\n> > > model = GPT2LMHeadModel.from_pretrained('gpt2') # or any other checkpoint\r\n> > > word_embeddings = model.transformer.wte.weight # Word Token Embeddings \r\n> > > position_embeddings = model.transformer.wpe.weight # Word Position Embeddings \r\n> > > ```\r\n> > \r\n> > \r\n> > Hi,\r\n> > Thank you for your reply! So if I want to get the vector for 'man', it would be like this:\r\n> > > tokenizer = GPT2Tokenizer.from_pretrained('gpt2')\r\n> > > text_index = tokenizer.encode('man',add_prefix_space=True)\r\n> > > vector = model.transformer.wte.weight[text_index,:]\r\n> > \r\n> > \r\n> > Is it correct?\r\n> \r\n> Did you succeed? I'm pursuing the same goal and I don't know how to validate my findings. I have tested some king - man + woman stuff, but it didn't work.\r\n\r\nHow did it go? I am stuck here too.", "> > Hi, indeed GPT-2 has a slightly different implementation than BERT. In order to have access to the embeddings, you would have to do the following:\r\n> > ```python\r\n> > from transformers import GPT2LMHeadModel\r\n> > \r\n> > model = GPT2LMHeadModel.from_pretrained('gpt2') # or any other checkpoint\r\n> > word_embeddings = model.transformer.wte.weight # Word Token Embeddings \r\n> > position_embeddings = model.transformer.wpe.weight # Word Position Embeddings \r\n> > ```\r\n> \r\n> Hi,\r\n> \r\n> Thank you for your reply! So if I want to get the vector for 'man', it would be like this:\r\n> \r\n> > tokenizer = GPT2Tokenizer.from_pretrained('gpt2')\r\n> > text_index = tokenizer.encode('man',add_prefix_space=True)\r\n> > vector = model.transformer.wte.weight[text_index,:]\r\n> \r\n> Is it correct?\r\n\r\nHow did it go?", "> > > Hi, indeed GPT-2 has a slightly different implementation than BERT. In order to have access to the embeddings, you would have to do the following:\r\n> > > ```python\r\n> > > from transformers import GPT2LMHeadModel\r\n> > > \r\n> > > model = GPT2LMHeadModel.from_pretrained('gpt2') # or any other checkpoint\r\n> > > word_embeddings = model.transformer.wte.weight # Word Token Embeddings \r\n> > > position_embeddings = model.transformer.wpe.weight # Word Position Embeddings \r\n> > > ```\r\n> > \r\n> > \r\n> > Hi,\r\n> > Thank you for your reply! So if I want to get the vector for 'man', it would be like this:\r\n> > > tokenizer = GPT2Tokenizer.from_pretrained('gpt2')\r\n> > > text_index = tokenizer.encode('man',add_prefix_space=True)\r\n> > > vector = model.transformer.wte.weight[text_index,:]\r\n> > \r\n> > \r\n> > Is it correct?\r\n> \r\n> How did it go?\r\n\r\nWell, it is working. However, these weights/embeddings are \"context-dependent\" so one should not expect \"king-queen+woman\" lead to anything. ", "The code already posted here is correct:\r\n\r\n```\r\nmodel.transformer.wte.weight[input_ids,:]\r\n```\r\n\r\nwhere `input_ids` is a tensor of shape `(batch_size, sequence_length)`. This will give you a tensor of shape `(batch_size, sequence_length, embedding_dimension)`. For example, you can do this with the output of the tokenizer:\r\n\r\n```\r\ninputs = tokenizer([\"Hello, my name\"], return_tensors=\"pt\")\r\nembeds = model.transformer.wte.weight[input_ids, :]\r\n```\r\n\r\nYou can validate that this is correct by passing the embeds into the model and checking that you get the same thing as when passing in the inputs:\r\n\r\n```\r\noutputs1 = model(input_ids=inputs.input_ids)\r\noutputs2 = model(inputs_embeds=embeds)\r\nassert torch.allclose(outputs1.logits, outputs2.logits)\r\n```\r\n\r\nor even\r\n\r\n```\r\nfor layer1, layer2 in zip(outputs1.hidden_states, outputs2.hidden_states):\r\n assert torch.allclose(layer1, layer2)\r\n```", "it's a bit late, but might help someone, despite not being static, contextual embeddings still gave me reasonable results here\r\n\r\n```\r\nmodel_id = \"gpt2-large\"\r\nmodel = GPT2LMHeadModel.from_pretrained(model_id, output_attentions=True).to(device)\r\nmodel.eval()\r\ntokenizer = GPT2TokenizerFast.from_pretrained(model_id)\r\ndef get_word_embedding(word, model, tokenizer):\r\n # Encode the word to get token IDs\r\n token_ids = tokenizer.encode(word, add_special_tokens=False)\r\n \r\n # Convert token IDs to tensor and move it to the model's device\r\n tokens_tensor = torch.tensor([token_ids], device=model.device)\r\n \r\n with torch.no_grad():\r\n # Forward pass through the model\r\n outputs = model(tokens_tensor)\r\n # Retrieve the hidden states from the model output\r\n hidden_states = outputs[0] # 'outputs' is a tuple, the first element is the hidden states\r\n\r\n # Averaging over the sequence length\r\n return hidden_states[0].mean(dim=0)\r\n\r\nking_emb = get_word_embedding('King', model, tokenizer)\r\nman_emb = get_word_embedding('Man', model, tokenizer)\r\nwoman_emb = get_word_embedding('Woman', model, tokenizer)\r\nqueen_emb = get_word_embedding('Queen', model, tokenizer)\r\n\r\n# print all the embeddings\r\nprint(\"king embedding: \", king_emb)\r\nprint(\"man embedding:\", man_emb)\r\nprint(\"woman embedding: \", woman_emb)\r\nprint(\"queen embedding:\", queen_emb)\r\nfrom torch.nn.functional import cosine_similarity\r\nanalogy_emb = king_emb - man_emb + woman_emb\r\nsimilarity = cosine_similarity(analogy_emb.unsqueeze(0), queen_emb.unsqueeze(0))\r\nprint(\"Cosine similarity: \", similarity.item())\r\n```\r\ngave me: \r\n```\r\nking embedding: tensor([ 2.3706, 4.7613, -0.7195, ..., -8.0351, -3.0770, 2.2482],\r\n device='cuda:3')\r\nman embedding: tensor([ 2.8015, 3.5800, -0.1190, ..., -6.7876, -3.8558, 1.8777],\r\n device='cuda:3')\r\nwoman embedding: tensor([ 3.0411, 5.3653, 0.3071, ..., -6.2418, -3.3228, 2.6389],\r\n device='cuda:3')\r\nqueen embedding: tensor([ 2.5185, 5.2505, -0.6024, ..., -7.1251, -2.5000, 1.6070],\r\n device='cuda:3')\r\nCosine similarity: 0.9761547446250916\r\n```\r\n\r\nand regarding @fqassemi 's question: \r\n```\r\nfrom torch.nn.functional import cosine_similarity\r\nimport torch\r\n\r\nfrom tqdm import tqdm # Import tqdm\r\n\r\n# Iterate over the entire vocabulary\r\nvocab = tokenizer.get_vocab()\r\ntop_matches = []\r\ntop_similarities = []\r\ndef get_word_embedding(word, model, tokenizer):\r\n if word in embeddings_dict:\r\n # Return the embedding if already in the dictionary\r\n return embeddings_dict[word]\r\n \r\n # Encode the word to get token IDs\r\n token_ids = tokenizer.encode(word, add_special_tokens=False)\r\n \r\n # Convert token IDs to tensor and move it to the model's device\r\n tokens_tensor = torch.tensor([token_ids], device=model.device)\r\n \r\n with torch.no_grad():\r\n # Forward pass through the model\r\n outputs = model(tokens_tensor)\r\n # Retrieve the hidden states from the model output\r\n hidden_states = outputs[0] # 'outputs' is a tuple, the first element is the hidden states\r\n word_emb = hidden_states[0].mean(dim=0)\r\n \r\n # Store the new embedding in the dictionary\r\n embeddings_dict[word] = word_emb\r\n return word_emb\r\n \r\n\r\nfor word, token_id in tqdm(vocab.items(), desc=\"Processing vocabulary\"):\r\n word_emb = get_word_embedding(word, model, tokenizer)\r\n sim = cosine_similarity(analogy_emb.unsqueeze(0), word_emb.unsqueeze(0)).item()\r\n \r\n # Keep track of top matches\r\n if len(top_matches) < 5 or sim > min(top_similarities):\r\n if len(top_matches) >= 5:\r\n # Remove the current lowest similarity\r\n min_index = top_similarities.index(min(top_similarities))\r\n top_matches.pop(min_index)\r\n top_similarities.pop(min_index)\r\n \r\n top_matches.append(word)\r\n top_similarities.append(sim)\r\n\r\n# Sort the top matches by similarity\r\nsorted_top_matches = sorted(zip(top_matches, top_similarities), key=lambda x: x[1], reverse=True)\r\n\r\nprint(sorted_top_matches)\r\n```\r\ngave me reasonable result for the nearest vectors:\r\n```\r\nProcessing vocabulary: 100%|██████████| 50257/50257 [22:23<00:00, 37.41it/s]\r\n[('Woman', 0.9765560626983643), ('Queen', 0.9761547446250916), ('Lady', 0.9727475643157959), ('ishop', 0.9681873917579651), ('!\"', 0.9671139717102051)]\r\n```", "Thanks for your code @ish3lan . I assume you can extend that to include sentences not just words correct?\r\n\r\nOn another note, I've been using that first code snippet and trying a bunch of different words to get different cosine similarities but all the similarities ended up being very high (>0.95). Is this normal/expected?" ]

[]

[]

[ "when i use \r\nexport SQUAD_DIR=/path/to/SQUAD \r\npython run_squad.py \\ --model_type bert \\ --model_name_or_path bert-base-cased \\ --do_train \\ --do_eval \\ --do_lower_case \\ --train_file $SQUAD_DIR/train-v1.1.json \\ --predict_file $SQUAD_DIR/dev-v1.1.json \\ --per_gpu_train_batch_size 12 \\ --learning_rate 3e-5 \\ --num_train_epochs 2.0 \\ --max_seq_length 384 \\ --doc_stride 128 \\ --output_dir /tmp/debug_squad/\r\n\r\nit is showing no progress. no gpu utilization.\r\n\r\n", "like this\r\n\r\n\r\n\r\n 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\r\n10/08/2019 15:32:45 - INFO - utils_squad - input_mask: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\r\n10/08/2019 15:32:45 - INFO - utils_squad - segment_ids: 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\r\n10/08/2019 15:32:45 - INFO - utils_squad - start_position: 49\r\n10/08/2019 15:32:45 - INFO - utils_squad - end_position: 50\r\n10/08/2019 15:32:45 - INFO - utils_squad - answer: the 1870s\r\n", "what is the meaning of this???? stuck somewhere in tokenization_bert.py", "wait for 10-15 min ...it will work.", "Hi, the program probably didn't hang but was still converting the examples to features, which can be a timely process.\r\nIf you want to have more information, you could always add a print statement notifying you of the current index it is converting to feature.\r\n\r\nPlease note that once you have done this conversion to features, these will be cached on your disk to be used the next time. This conversion is only done once.", "I observed exactly the same, it took 18 minutes to log the next line in a p3.2xlarge host. Would be great to parallelize this portion(I notice only one cpu is running in this period.), and show a progress bar for converting the examples to features.\r\n\r\n```\r\n10/27/2019 02:45:46 - INFO - utils_squad - start_position: 47\r\n10/27/2019 02:45:46 - INFO - utils_squad - end_position: 48\r\n10/27/2019 02:45:46 - INFO - utils_squad - answer: the 1870s\r\n10/27/2019 03:03:05 - INFO - __main__ - Saving features into cached file /home/ubuntu/SQuAD-explorer/dataset/cached_train_bert-base-uncased_384\r\n10/27/2019 03:05:09 - INFO - __main__ - ***** Running training *****\r\n```\r\n\r\n\r\n", "@cockroachzl @vikrant094\r\nIf you're running on a Linux variant OS you might try adding **`export OMP_NUM_THREADS=x`** at the top of your script file, where x is the number of cores, not threads, of your CPU. With this script file addition on my Ubuntu 18.04 machine, examples-to-features uses 2 of my 6 CPUs @ 100%, instead of just a single CPU." ]

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

[]

[ "The model saved can be loaded by using the `model.from_pretrained(directory)` method. The training arguments are saved so that they can be re-used later. You can load them using the `torch.load(directory/training_args.bin)` method.", "thanks for reply. ", "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/1455?src=pr&el=h1) Report\n> Merging [#1455](https://codecov.io/gh/huggingface/transformers/pull/1455?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/ae1d03fc51bb22ed59517ee6f92c560417fdb049?src=pr&el=desc) will **decrease** coverage by `1.92%`.\n> The diff coverage is `53.35%`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/1455?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #1455 +/- ##\n==========================================\n- Coverage 85.9% 83.97% -1.93% \n==========================================\n Files 91 87 -4 \n Lines 13653 12866 -787 \n==========================================\n- Hits 11728 10804 -924 \n- Misses 1925 2062 +137\n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/1455?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [transformers/modeling\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/1455/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL21vZGVsaW5nX3V0aWxzLnB5) | `92.44% <ø> (ø)` | :arrow_up: |\n| [transformers/modeling\\_beam\\_search.py](https://codecov.io/gh/huggingface/transformers/pull/1455/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL21vZGVsaW5nX2JlYW1fc2VhcmNoLnB5) | `0% <0%> (ø)` | |\n| [transformers/modeling\\_xlm.py](https://codecov.io/gh/huggingface/transformers/pull/1455/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL21vZGVsaW5nX3hsbS5weQ==) | `88.42% <100%> (ø)` | :arrow_up: |\n| [transformers/configuration\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/1455/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL2NvbmZpZ3VyYXRpb25fdXRpbHMucHk=) | `97.33% <100%> (-1.34%)` | :arrow_down: |\n| [transformers/tests/modeling\\_common\\_test.py](https://codecov.io/gh/huggingface/transformers/pull/1455/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL3Rlc3RzL21vZGVsaW5nX2NvbW1vbl90ZXN0LnB5) | `74.68% <100%> (-1.34%)` | :arrow_down: |\n| [transformers/tests/modeling\\_bert\\_test.py](https://codecov.io/gh/huggingface/transformers/pull/1455/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL3Rlc3RzL21vZGVsaW5nX2JlcnRfdGVzdC5weQ==) | `96.92% <100%> (+0.53%)` | :arrow_up: |\n| [transformers/modeling\\_encoder\\_decoder.py](https://codecov.io/gh/huggingface/transformers/pull/1455/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL21vZGVsaW5nX2VuY29kZXJfZGVjb2Rlci5weQ==) | `67.69% <67.69%> (ø)` | |\n| [transformers/modeling\\_bert.py](https://codecov.io/gh/huggingface/transformers/pull/1455/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL21vZGVsaW5nX2JlcnQucHk=) | `87.59% <86.79%> (-0.59%)` | :arrow_down: |\n| [...ransformers/tests/modeling\\_encoder\\_decoder\\_test.py](https://codecov.io/gh/huggingface/transformers/pull/1455/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL3Rlc3RzL21vZGVsaW5nX2VuY29kZXJfZGVjb2Rlcl90ZXN0LnB5) | `96.29% <96.29%> (ø)` | |\n| [transformers/modeling\\_tf\\_pytorch\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/1455/diff?src=pr&el=tree#diff-dHJhbnNmb3JtZXJzL21vZGVsaW5nX3RmX3B5dG9yY2hfdXRpbHMucHk=) | `76.92% <0%> (-16.04%)` | :arrow_down: |\n| ... and [45 more](https://codecov.io/gh/huggingface/transformers/pull/1455/diff?src=pr&el=tree-more) | |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/1455?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/1455?src=pr&el=footer). Last update [ae1d03f...a88a0e4](https://codecov.io/gh/huggingface/transformers/pull/1455?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n", "Regarding the initialization of `Bert2Rnd` with pretrained-weights for the encoder and random initialization for the decoder (as per the name), I see two potential solutions:\r\n\r\n1. Patch `from_pretrained` in `modeling_utils.py` by not attempting to load weights if `Decoder` is in the name. I don't like this solution at all: the burden of initialization should be borne by the instantiating class, and adding model-specific logic in this function will bite us in the 🍑 with almost 100% certainty at some point in the future.\r\n\r\n2. Override `from_pretrained` in `Bert2Rnd`. This would be fairly simple if there was a way to fetch the config created by `from_pretrained` in the Base class. I imagined the following:\r\n\r\n```python\r\n@classmethod\r\ndef from_pretrained(cls, pretrained_model_or_path, *args, **kwargs):\r\n pretrained_encoder, config = BertEncoder.from_pretrained(pretrained_model_or_path, *args, **kwargs)\r\n model = cls(config)\r\n model.encoder = pretrained_encoder\r\n return model\r\n```\r\n\r\nChanging `PretrainedModel`'s `from_pretrained` to output the config as well as the model is a non-breaking change, thanks to python's magic ✨. The following code runs without problem:\r\n\r\n```python\r\ndef magic_function(number):\r\n return number, number+1\r\n\r\na = magic_function(10)\r\n```\r\n\r\nI'd appreciate your opinion since this is a central part of the library. In the meantime I'll dive into the way parameter loading works and see if I can find another solution.", "Not sure I get the point of your `magic_function` stuff but yes solution 2 is the way to go.\r\nYou'll have to write a specific `from_pretrained` function for the seq2seq models.", "Now I realize there was strictly zero point :smile: \r\n\r\nI hope to have a functioning version by noon :crossed_fingers: ", "Here is something that \"works\" in the sense that:\r\n\r\n1. All tests pass (with a new one that tests the initialization)\r\n2. I can add an LM head at the top of the decoder and have a working `text -> Bert2Rand -> text` pipeline; the output is rubbish since the decoder is initialized randomly.\r\n\r\n*Edit:* I just re-read the paper and it turns out they initialized the decoder with pretrained embeddings and not random embeddings. I’ll make the change.\r\n\r\nTo be able to generate meaningful text we would need to fine-tune the model. From here I can either:\r\n\r\n- fine-tune the model for text generation (create `run_seq2seq_finetuning.py` and use `run_generation.py`)\r\n- fine-tune for abstractive summarization (and create `run_abstractive_summarization.py`);\r\n\r\nI’d vote for text generation for now as it is narrower in scope and won’t add yet another concept in the PR. Then we can finalize the API of the model and ship it + examples.\r\n\r\nTentative plan:\r\n1. `Bert2Rnd` finetuning + text generation;\r\n2. `UniLM`+ finetuning + text generation in a separate PR;\r\n3. Abstractive summarization using `Bert2Rnd`and `UniLM`.\r\n\r\nI am also strangely fascinated by the `BertShare`architecture (decoder sharing weights with encoder, asymmetry between them due to encoder-decoder attention only & outperforming everything else), but we can keep this one for later.", "I implemented all elements necessary to reproduce the results from Lapata & Liu:\r\n\r\n* Separate sentences in the document by `[SEP] [CLS]`. I currently did this in the `run_seq2seq_finetuning.py` file, but I could instead add a `add_special_tokens_sentence_representation` function in `tokenizer_bert.py` if you think it is cleaner.\r\n* Add alternating `token_type_ids` for each sentence. Same remark as the previous point.\r\n* Add a custom Optimizer class: they use separate optimizers for encoder & decoder + different learning schedules.\r\n* Add the beloved beam-search in `modeling_beam_search.py.` It is a bit awkward, and I would like to have it well tested.\r\n\r\nThings I need input on:\r\n- [ ] Any mistake\r\n- [ ] First and second point: is it worth adding two functions in `bert_tokenizer`?\r\n- [ ] What do we do about beam search?", "Ok LGTM, let's merge this and continue the work on summarization and T5 on separate PRs.", "@rlouf This is a really great addition! Any plan to complete the run_summarization_finetuning.py end-to-end soon? Or any psuedo code to point me to the right direction would be great too. ", "> @rlouf This is a really great addition! Any plan to complete the run_summarization_finetuning.py end-to-end soon? Or any psuedo code to point me to the right direction would be great too.\r\n\r\nWould something like this work? \r\n1. Initialize a `TransformerBeamSearch` from a `PreTrainedEncoderDecoder`\r\n2. Call the `forward` method of `TransformerBeamSearch` with encoder_input_ids and other necessary arguments\r\n3. Use the tokenizer to convert results from step 2 back to text. \r\n\r\nYou mentioned `TransformerBeamSearch` is a draft version. Not sure how much more work is needed on it. Looks OK to me, but I'm new to seq2seq models. :)", "@hlums Thanks! You can follow the `example-summarization` branch where we are currently completing the example (and solidifying the Beam Search).\r\n\r\nThe answers to your questions are in the `evaluate` function of the `run_summarization.py` example, and you are essentially right :) We will soon release the example with a short example of how to you use BeamSearch.", "@rlouf , great addition. Is that possible to initialize Model2Model class with both the encoder/ decoder are going to be XLMRoberta model and pre-train it with my own data?" ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/1454?src=pr&el=h1) Report\n> Merging [#1454](https://codecov.io/gh/huggingface/transformers/pull/1454?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/d688af19e5ce92c1395820a89e3f3b635eacc2ba?src=pr&el=desc) will **not change** coverage.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/1454?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #1454 +/- ##\n=======================================\n Coverage 84.72% 84.72% \n=======================================\n Files 84 84 \n Lines 12591 12591 \n=======================================\n Hits 10668 10668 \n Misses 1923 1923\n```\n\n\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/1454?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/1454?src=pr&el=footer). Last update [d688af1...5ce8d29](https://codecov.io/gh/huggingface/transformers/pull/1454?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n", "Fine with me, thanks!", "My critique would be that just writing a general 'except' is not PEP-y. The correct error test should be checked. Then again, large parts of the whole package are not PEP-y so it might not be important for the developers. " ]

[]

[]

[ "I have been experiencing the same issue #1378.", "Fixed on master with 23b7138, thanks.\r\nWill be in this week's new release 2.1", "thanks a lot" ]

[ { "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" ]

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

[]

[ "Even though I am in favour of using as many built-ins as possible, I wonder whether it is not too early to do this. You will end up with a lot of pseudo-duplicate code: for those who are on 1.0 (no transformer-like support), 1.1 (only nn.*Attention), and 1.2 (full transformer). I don't know any statistics about people using `transformers` but I can imagine that many are still on PyTorch 1.0. ", "We have a small codebase on the side where we use `nn.Transformer` to build both a BERT-style and a GPT2-style model that are compatible with our pretrained weights, but we still think it's a bit too early to refactor/freeze the lib's internals. A lot of research is still going to focus on the models' internals so we don't want to overfit to the current architecture.", "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", "@julien-c @BramVanroy just curious, do you guys still think it’s too early to use nn.Transformer?", "A friendly ping to the maintainers to implement the in-built modules.", "That's not something we can do because it will break all existing checkpoints." ]

[]

[]

[ " @evanweissburg Hi, i got the same error..have you found any solution???\r\n ", "Hello! I believe you must have sklearn installed in order to pass these tests. Please let me know if it doesn't work while having sklearn installed.", "yeah got it .... i guess we need to run pip install -r ./examples/requirements.txt", "Indeed!", "Could we add this to the getting started documentation?\r\n\r\npip install -r ./examples/requirements.txt", "Had the same issue. Probably a problem with sklearn. Installed with conda and it was fixed." ]

[]

[]

[ "Hello, this language model fine-tuning was community-maintained and is now deprecated. The example script to fine-tune on language modeling is now `run_lm_finetuning.py`." ]

[]

[ { "login": "julien-c", "id": 326577, "node_id": "MDQ6VXNlcjMyNjU3Nw==", "avatar_url": "https://avatars.githubusercontent.com/u/326577?v=4", "gravatar_id": "", "url": "https://api.github.com/users/julien-c", "html_url": "https://github.com/julien-c", "followers_url": "https://api.github.com/users/julien-c/followers", "following_url": "https://api.github.com/users/julien-c/following{/other_user}", "gists_url": "https://api.github.com/users/julien-c/gists{/gist_id}", "starred_url": "https://api.github.com/users/julien-c/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/julien-c/subscriptions", "organizations_url": "https://api.github.com/users/julien-c/orgs", "repos_url": "https://api.github.com/users/julien-c/repos", "events_url": "https://api.github.com/users/julien-c/events{/privacy}", "received_events_url": "https://api.github.com/users/julien-c/received_events", "type": "User", "site_admin": false }, { "login": "thomwolf", "id": 7353373, "node_id": "MDQ6VXNlcjczNTMzNzM=", "avatar_url": "https://avatars.githubusercontent.com/u/7353373?v=4", "gravatar_id": "", "url": "https://api.github.com/users/thomwolf", "html_url": "https://github.com/thomwolf", "followers_url": "https://api.github.com/users/thomwolf/followers", "following_url": "https://api.github.com/users/thomwolf/following{/other_user}", "gists_url": "https://api.github.com/users/thomwolf/gists{/gist_id}", "starred_url": "https://api.github.com/users/thomwolf/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/thomwolf/subscriptions", "organizations_url": "https://api.github.com/users/thomwolf/orgs", "repos_url": "https://api.github.com/users/thomwolf/repos", "events_url": "https://api.github.com/users/thomwolf/events{/privacy}", "received_events_url": "https://api.github.com/users/thomwolf/received_events", "type": "User", "site_admin": false }, { "login": "VictorSanh", "id": 16107619, "node_id": "MDQ6VXNlcjE2MTA3NjE5", "avatar_url": "https://avatars.githubusercontent.com/u/16107619?v=4", "gravatar_id": "", "url": "https://api.github.com/users/VictorSanh", "html_url": "https://github.com/VictorSanh", "followers_url": "https://api.github.com/users/VictorSanh/followers", "following_url": "https://api.github.com/users/VictorSanh/following{/other_user}", "gists_url": "https://api.github.com/users/VictorSanh/gists{/gist_id}", "starred_url": "https://api.github.com/users/VictorSanh/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/VictorSanh/subscriptions", "organizations_url": "https://api.github.com/users/VictorSanh/orgs", "repos_url": "https://api.github.com/users/VictorSanh/repos", "events_url": "https://api.github.com/users/VictorSanh/events{/privacy}", "received_events_url": "https://api.github.com/users/VictorSanh/received_events", "type": "User", "site_admin": false }, { "login": "LysandreJik", "id": 30755778, "node_id": "MDQ6VXNlcjMwNzU1Nzc4", "avatar_url": "https://avatars.githubusercontent.com/u/30755778?v=4", "gravatar_id": "", "url": "https://api.github.com/users/LysandreJik", "html_url": "https://github.com/LysandreJik", "followers_url": "https://api.github.com/users/LysandreJik/followers", "following_url": "https://api.github.com/users/LysandreJik/following{/other_user}", "gists_url": "https://api.github.com/users/LysandreJik/gists{/gist_id}", "starred_url": "https://api.github.com/users/LysandreJik/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/LysandreJik/subscriptions", "organizations_url": "https://api.github.com/users/LysandreJik/orgs", "repos_url": "https://api.github.com/users/LysandreJik/repos", "events_url": "https://api.github.com/users/LysandreJik/events{/privacy}", "received_events_url": "https://api.github.com/users/LysandreJik/received_events", "type": "User", "site_admin": false } ]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/1448?src=pr&el=h1) Report\n> Merging [#1448](https://codecov.io/gh/huggingface/transformers/pull/1448?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/8fcc6507ce9d0922ddb60f4a31d4b9a839de1270?src=pr&el=desc) will **not change** coverage.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/1448?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #1448 +/- ##\n=======================================\n Coverage 84.72% 84.72% \n=======================================\n Files 84 84 \n Lines 12591 12591 \n=======================================\n Hits 10668 10668 \n Misses 1923 1923\n```\n\n\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/1448?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/1448?src=pr&el=footer). Last update [8fcc650...45de313](https://codecov.io/gh/huggingface/transformers/pull/1448?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/1447?src=pr&el=h1) Report\n> Merging [#1447](https://codecov.io/gh/huggingface/transformers/pull/1447?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/1615360c71f75da7b8aefd14c5d8a461486f865b?src=pr&el=desc) will **not change** coverage.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/1447?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #1447 +/- ##\n=======================================\n Coverage 84.72% 84.72% \n=======================================\n Files 84 84 \n Lines 12591 12591 \n=======================================\n Hits 10668 10668 \n Misses 1923 1923\n```\n\n\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/1447?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/1447?src=pr&el=footer). Last update [1615360...7afd00a](https://codecov.io/gh/huggingface/transformers/pull/1447?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n" ]

[]

[]

[ "Hello! You should specify `add_prefix_space=True` in your encode method to obtain that behavior.", "Thank you! That works!" ]

[]

[]

[ "I'm so sorry, I haven't replaced all occurrences of `pytorch-tranformers` to `tranformers`. That was the source of the problem.", "Glad to hear that!" ]