Datasets:

---

DanielPFlorian

/

Transformers-Github-Issues

like 0

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

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/3113?src=pr&el=h1) Report\n> Merging [#3113](https://codecov.io/gh/huggingface/transformers/pull/3113?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/e9e6efdc452b74947d40a5a2e8af2fc444c63b5b?src=pr&el=desc) will **decrease** coverage by `0.51%`.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/3113?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #3113 +/- ##\n==========================================\n- Coverage 78.35% 77.84% -0.52% \n==========================================\n Files 98 98 \n Lines 16422 16422 \n==========================================\n- Hits 12868 12784 -84 \n- Misses 3554 3638 +84\n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/3113?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/data/processors/squad.py](https://codecov.io/gh/huggingface/transformers/pull/3113/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9kYXRhL3Byb2Nlc3NvcnMvc3F1YWQucHk=) | `28.89% <0%> (-27.6%)` | :arrow_down: |\n| [src/transformers/modeling\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/3113/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ191dGlscy5weQ==) | `94.61% <0%> (+0.15%)` | :arrow_up: |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/3113?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/3113?src=pr&el=footer). Last update [e9e6efd...07a4c8b](https://codecov.io/gh/huggingface/transformers/pull/3113?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 }, { "id": 1834056635, "node_id": "MDU6TGFiZWwxODM0MDU2NjM1", "url": "https://api.github.com/repos/huggingface/transformers/labels/Core:%20Tokenization", "name": "Core: Tokenization", "color": "FF4446", "default": false, "description": "Internals of the library; Tokenization." }, { "id": 1834088753, "node_id": "MDU6TGFiZWwxODM0MDg4NzUz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Tests", "name": "Tests", "color": "a6fcca", "default": false, "description": "Related to tests" } ]

[]

[ "I can't assign reviewers, but you asked me in #3058 to ping @LysandreJik and @mfuntowicz when I do this.", "I added a test for the issue from #3088.", "Added another test for #3091.", "@dirkgr thanks for taking the time to include your tests into ours. \r\n\r\nIt will definitively help making sure everything is working as expected on your side 👍 ", "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 can be closed now, right?", "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": 1838412367, "node_id": "MDU6TGFiZWwxODM4NDEyMzY3", "url": "https://api.github.com/repos/huggingface/transformers/labels/model%20card", "name": "model card", "color": "92d5f4", "default": false, "description": "Related to pretrained model cards" } ]

[]

[ "You have to add the thumbnail", "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/3111?src=pr&el=h1) Report\n> Merging [#3111](https://codecov.io/gh/huggingface/transformers/pull/3111?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/e9e6efdc452b74947d40a5a2e8af2fc444c63b5b?src=pr&el=desc) will **decrease** coverage by `0.52%`.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/3111?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #3111 +/- ##\n==========================================\n- Coverage 78.35% 77.83% -0.53% \n==========================================\n Files 98 98 \n Lines 16422 16422 \n==========================================\n- Hits 12868 12782 -86 \n- Misses 3554 3640 +86\n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/3111?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/data/processors/squad.py](https://codecov.io/gh/huggingface/transformers/pull/3111/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9kYXRhL3Byb2Nlc3NvcnMvc3F1YWQucHk=) | `28.89% <0%> (-27.6%)` | :arrow_down: |\n| [src/transformers/modeling\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/3111/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ191dGlscy5weQ==) | `94.29% <0%> (-0.16%)` | :arrow_down: |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/3111?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/3111?src=pr&el=footer). Last update [e9e6efd...3060b8c](https://codecov.io/gh/huggingface/transformers/pull/3111?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n", "Hi, Julien. Can you add the default thumbnail you usually add to models? (name of the model + Huggingface logo)" ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/3110?src=pr&el=h1) Report\n> Merging [#3110](https://codecov.io/gh/huggingface/transformers/pull/3110?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/5c5af879b6d45c879c987154f66d4ea978925fb2?src=pr&el=desc) will **increase** coverage by `<.01%`.\n> The diff coverage is `100%`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/3110?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #3110 +/- ##\n==========================================\n+ Coverage 77.83% 77.84% +<.01% \n==========================================\n Files 98 98 \n Lines 16422 16422 \n==========================================\n+ Hits 12782 12783 +1 \n+ Misses 3640 3639 -1\n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/3110?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/modeling\\_bart.py](https://codecov.io/gh/huggingface/transformers/pull/3110/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19iYXJ0LnB5) | `93.38% <100%> (+0.33%)` | :arrow_up: |\n| [src/transformers/modeling\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/3110/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ191dGlscy5weQ==) | `94.45% <0%> (-0.16%)` | :arrow_down: |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/3110?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/3110?src=pr&el=footer). Last update [5c5af87...0893268](https://codecov.io/gh/huggingface/transformers/pull/3110?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n" ]

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

[]

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

[ { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null }, { "id": 1834056761, "node_id": "MDU6TGFiZWwxODM0MDU2NzYx", "url": "https://api.github.com/repos/huggingface/transformers/labels/Core:%20Modeling", "name": "Core: Modeling", "color": "FF8446", "default": false, "description": "Internals of the library; Models." }, { "id": 1845609017, "node_id": "MDU6TGFiZWwxODQ1NjA5MDE3", "url": "https://api.github.com/repos/huggingface/transformers/labels/seq2seq", "name": "seq2seq", "color": "fef2c0", "default": false, "description": "" } ]

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

[ "if you install from master it seems to work on 'bart-large'. \r\nSeems like it's only an issue on 'bart-large-cnn'\r\n```\r\ntokenizer = BartTokenizer.from_pretrained('bart-large')\r\nmodel = BartForMaskedLM.from_pretrained('bart-large',output_past=True)\r\n\r\nARTICLE_TO_SUMMARIZE = \"My friends are <mask> but they eat too many carbs.\"\r\ninputs = tokenizer.batch_encode_plus([ARTICLE_TO_SUMMARIZE], max_length=1024, return_tensors='pt')\r\ngenerated_ids = model.generate(inputs['input_ids'], attention_mask=inputs['attention_mask'], num_return_sequences=4)\r\nprint([tokenizer.decode(g, skip_special_tokens=True, clean_up_tokenization_spaces=False) for g in generated_ids])\r\n```\r\noutput: \r\n```\r\n['My kids are good, but they eat too many carbs. My friends are good.', 'My kids are good, but they eat too many carbs. My friends are good.', 'My kids are good, but they eat too many carbs. My friends are good.', 'My kids are good, but they eat too many carbs. My friends are good.']\r\n```", "Bart-large-cnn doesn't have a mask_token_id, which is admittedly confusing.\r\n\r\nthis is how I would do mask filling\r\n```python\r\nmodel = BartForMaskedLM.from_pretrained('bart-large')\r\ntokenizer = AutoTokenizer.from_pretrained('bart-large')\r\nARTICLE_TO_SUMMARIZE = \"My friends are <mask> but they eat too many carbs.\"\r\ninputs = tokenizer.batch_encode_plus([ARTICLE_TO_SUMMARIZE], return_tensors='pt')\r\ninput_ids = inputs['input_ids']\r\n#generated_ids = model(, attention_mask=inputs['attention_mask'])[0]\r\nlogits = model(input_ids)[0]\r\nmasked_index = (input_ids[0] == tokenizer.mask_token_id).nonzero().item()\r\nprobs = logits[0, masked_index].softmax(dim=0)\r\nvalues, predictions = probs.topk(10)\r\ntokenizer.decode(predictions).split()\r\n# ['good', 'great', 'all', 'really', 'very', 'healthy', 'also', 'not', 'the', 'doing']\r\n```\r\n", "One liner courtesy of @julien-c\r\n```\r\nfrom transformers import pipeline\r\nnlp = pipeline('fill-mask', 'bart-large')\r\nnlp(\"My friends are <mask> but they eat too many carbs.\")\r\n```", "Thanks @sshleifer, that will do the trick!\r\n\r\nThe following does work:\r\n```\r\ntokenizer = AutoTokenizer.from_pretrained('bart-large-cnn')\r\ntokenizer.mask_token_id\r\n>>> 50264\r\n```\r\n...which is a bit counterintuitive as it implies that `<mask>` _is_ available. It's also not clear from the docs that `bart-large` can be used successfully with `BartForMaskedLM`. " ]

[ { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null }, { "id": 1834056635, "node_id": "MDU6TGFiZWwxODM0MDU2NjM1", "url": "https://api.github.com/repos/huggingface/transformers/labels/Core:%20Tokenization", "name": "Core: Tokenization", "color": "FF4446", "default": false, "description": "Internals of the library; Tokenization." } ]

[]

[ "When I copy/paste your code, I get this in `dumped/model/`:\r\n`vocab.txt`, `tokenizer_config.json` and `special_tokens_map.json`\r\n\r\nThe tokenizer config does save the lowercase, and the output of the code is two `False`.\r\n\r\nBelow the code snippet I had to actually run it:\r\n\r\n```py\r\nimport transformers\r\nimport os\r\n\r\ntokenizer = transformers.BertTokenizer.from_pretrained('bert-base-cased', do_lower_case=False)\r\nprint(tokenizer.basic_tokenizer.do_lower_case)\r\nos.makedirs(\"~/dumped/model\", exist_ok=True)\r\ntokenizer.save_pretrained('~/dumped/model')\r\ntokenizer = transformers.BertTokenizer.from_pretrained('~/dumped/model')\r\nprint(tokenizer.basic_tokenizer.do_lower_case)\r\n\r\n```", "Hi @LysandreJik,\r\n\r\nIt does work when you initialize the model `.from_pretrained('bert-base-cased')`, because the pretrained tokenizer already has a configuration that is saved afterwards.\r\n\r\nI was talking about a case when you do not have a config and load only the local `vocab.txt` file:\r\n```python3\r\nimport os\r\n\r\nfrom transformers import BertTokenizer\r\n\r\n# Download the `bert-base-cased` tokenizer and initialized from pretrained\r\ntokenizer = BertTokenizer.from_pretrained('bert-base-cased')\r\n# A configuration is already there, `do_lower_case` is `False`\r\nprint(tokenizer.basic_tokenizer.do_lower_case)\r\nos.makedirs(\"~/dumped/model\", exist_ok=True)\r\n# Save it locally\r\ntokenizer.save_pretrained('~/dumped/model')\r\n\r\n# We can see that the config file has data\r\nwith open(\"~/dumped/model/tokenizer_config.json\") as f:\r\n print(f.read())\r\n\r\n# Initialize as if we only have a local `vocab.txt` which is my case\r\ntokenizer = BertTokenizer('~/dumped/model/vocab.txt', do_lower_case=False)\r\nprint(tokenizer.basic_tokenizer.do_lower_case)\r\ntokenizer.save_pretrained('~/dumped/model')\r\n\r\n# After saving the config is empty\r\nwith open(\"~/dumped/model/tokenizer_config.json\") as f:\r\n print(f.read())\r\n\r\n# And after initializing from pretrained `do_lower_case` is `True`\r\ntokenizer = BertTokenizer.from_pretrained('~/dumped/model')\r\nprint(tokenizer.basic_tokenizer.do_lower_case)\r\n```", "Hmm, that makes sense. Indeed, that seems problematic. Thanks for opening this issue, we're looking into it!", "@LysandreJik Looked a bit into it quickly, and here is the deal:\r\n\r\n`.save_pretrained` from `BertTokenizer` is inherited from `PreTrainedTokenizer`, and save config based on the `self.init_kwargs` dict.\r\n\r\nHowever, `do_lower_case` [is not passed to the `super().__init__()`](https://github.com/huggingface/transformers/blob/e9e6efdc452b74947d40a5a2e8af2fc444c63b5b/src/transformers/tokenization_bert.py#L177) in the Bert tokenizer.\r\n\r\nAdding it to the `kwargs` passed to `super()` should do the trick. I can send a PR if that helps!\r\n\r\n\r\n", "@RaphaelMeudec, it would not help: the config is empty even though `unt_token`, `sep_token`, etc., are passed to `super().__init__()`.\r\nAlso, I've tried it :slightly_smiling_face: ", "@yoptar Indeed! What I can't explain is that `self.init_kwargs` is initialized [here](https://github.com/huggingface/transformers/blob/e9e6efdc452b74947d40a5a2e8af2fc444c63b5b/src/transformers/tokenization_utils.py#L327) but remains untouched in the rest of the code? \r\n\r\nI've been able to make your code work as expected by initializing `self.init_kwargs = kwargs` (instead of `{}`) and passing `do_lower_case=do_lower_case` in BertTokenizer `super()` resulting in:\r\n\r\n```\r\nFalse\r\n{\"do_lower_case\": false, \"max_len\": 512}\r\nFalse\r\n{\"unk_token\": \"[UNK]\", \"sep_token\": \"[SEP]\", \"pad_token\": \"[PAD]\", \"cls_token\": \"[CLS]\", \"mask_token\": \"[MASK]\", \"do_lower_case\": false}\r\nFalse\r\n```\r\n\r\n@LysandreJik Do you have more insights on how `self.init_kwargs` is modified in the current code?", "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 do_sample=True tested anywhere?\r\n\r\nYes for the randomly initialized models, with dummy input and also for some Integration tests", "Good to merge for me" ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/3105?src=pr&el=h1) Report\n> Merging [#3105](https://codecov.io/gh/huggingface/transformers/pull/3105?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/b31f7150190cdf13950607f8ee1efe11b352c909?src=pr&el=desc) will **increase** coverage by `0.02%`.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/3105?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #3105 +/- ##\n==========================================\n+ Coverage 77.6% 77.62% +0.02% \n==========================================\n Files 98 98 \n Lines 16250 16230 -20 \n==========================================\n- Hits 12611 12599 -12 \n+ Misses 3639 3631 -8\n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/3105?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/pipelines.py](https://codecov.io/gh/huggingface/transformers/pull/3105/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9waXBlbGluZXMucHk=) | `70.95% <ø> (+0.52%)` | :arrow_up: |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/3105?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/3105?src=pr&el=footer). Last update [b31f715...da52b39](https://codecov.io/gh/huggingface/transformers/pull/3105?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n" ]

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

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

[ "Can you reproduce with the code on master?\r\n\r\n```\r\ngit clone https://github.com/huggingface/transformers\r\ncd transformers\r\npip install .\r\n```\r\n(duplicate of https://github.com/huggingface/transformers/issues/3079)\r\n", "Closing for now, reopen if this is broken on the latest code. Otherwise, it will be in the next pip release. Thanks!" ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/3103?src=pr&el=h1) Report\n> Merging [#3103](https://codecov.io/gh/huggingface/transformers/pull/3103?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/a088d75e510d5641808ccd72f5dca4df36d95b8e?src=pr&el=desc) will **increase** coverage by `0.05%`.\n> The diff coverage is `100%`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/3103?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #3103 +/- ##\n==========================================\n+ Coverage 77.82% 77.88% +0.05% \n==========================================\n Files 98 98 \n Lines 16422 16461 +39 \n==========================================\n+ Hits 12781 12821 +40 \n+ Misses 3641 3640 -1\n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/3103?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/modeling\\_tf\\_albert.py](https://codecov.io/gh/huggingface/transformers/pull/3103/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9hbGJlcnQucHk=) | `85.53% <100%> (+0.08%)` | :arrow_up: |\n| [src/transformers/modeling\\_tf\\_gpt2.py](https://codecov.io/gh/huggingface/transformers/pull/3103/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9ncHQyLnB5) | `96.16% <100%> (+0.02%)` | :arrow_up: |\n| [src/transformers/modeling\\_tf\\_openai.py](https://codecov.io/gh/huggingface/transformers/pull/3103/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9vcGVuYWkucHk=) | `95.84% <100%> (ø)` | :arrow_up: |\n| [src/transformers/modeling\\_tf\\_xlnet.py](https://codecov.io/gh/huggingface/transformers/pull/3103/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl94bG5ldC5weQ==) | `89.03% <100%> (+0.04%)` | :arrow_up: |\n| [src/transformers/modeling\\_tf\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/3103/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl91dGlscy5weQ==) | `92.2% <100%> (+0.63%)` | :arrow_up: |\n| [src/transformers/modeling\\_tf\\_bert.py](https://codecov.io/gh/huggingface/transformers/pull/3103/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9iZXJ0LnB5) | `96.07% <100%> (+0.01%)` | :arrow_up: |\n| [src/transformers/modeling\\_tf\\_transfo\\_xl.py](https://codecov.io/gh/huggingface/transformers/pull/3103/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl90cmFuc2ZvX3hsLnB5) | `91.18% <100%> (+0.04%)` | :arrow_up: |\n| [src/transformers/modeling\\_tf\\_ctrl.py](https://codecov.io/gh/huggingface/transformers/pull/3103/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9jdHJsLnB5) | `99.57% <100%> (ø)` | :arrow_up: |\n| [src/transformers/modeling\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/3103/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ191dGlscy5weQ==) | `94.61% <0%> (+0.15%)` | :arrow_up: |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/3103?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/3103?src=pr&el=footer). Last update [a088d75...4c91a3a](https://codecov.io/gh/huggingface/transformers/pull/3103?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n", "Crud, as the test shows, of course load doesn't work in this first-stab implementation, because the layer classes get instantiated with a `dict` instead of a `PretrainedConfig` instance. So needs more work.", "With this change, seven of the 11 `TF*MainLayer` classes pass the test (saving, loading and producing same output for same input after the Keras save/load roundtrip). So I'm just not marking the other four `@keras_serializable` as I haven't gotten it working for those. Specifically:\r\n\r\n* TFT5MainLayer does not accept the same model inputs directly, as produced by `self.model_tester.prepare_config_and_inputs_for_common()`, so calling it fails\r\n* `TFXLMMainLayer`, `TFOpenAIGPTMainLayer`, and `TFDistilBertMainLayer` all fail the test (if I add the `@keras_serializable` attribute to them) in the same way: by outputting a tensor of shape `(7, 32)` after the save/load round-trip, which is just the first row of the `(13, 7, 32)` tensor that's output before. I haven't figured out the cause of this.", "Ok this looks good to me. \r\n\r\nDo you want to make `make style` and `make quality` to pass the code quality checks (see our [contributor guide](https://github.com/huggingface/transformers/blob/master/CONTRIBUTING.md) if needed), and I guess we can merge it.", "Fixed quality thing, added @functools.wraps (really shouldn't wrap without that as it causes confusing function metadata), added a docstring to the `keras_serializable` decorator, and changed to use a name specific to this library rather than the general name `config`, to be clearer where `transformers` is used along with other things in a Keras model. @thomwolf OK with all that?", "Ok this is good to me.\r\nThanks a lot for the awesome work on this.\r\n\r\nMerging @LysandreJik @julien-c ", "This landed in 2.6.0 but is missing in the release notes there https://github.com/huggingface/transformers/releases/tag/v2.6.0", "Indeed, that is my fault, I'm sorry for missing it.\r\n\r\nI've added it to the v2.7.0 release notes of this morning: https://github.com/huggingface/transformers/releases/tag/v2.7.0\r\n\r\nThanks again for your contribution @gthb !" ]

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

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/3102?src=pr&el=h1) Report\n> Merging [#3102](https://codecov.io/gh/huggingface/transformers/pull/3102?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/eec5ec807135ae61fa2266f3c7ad947cc207abda?src=pr&el=desc) will **decrease** coverage by `<.01%`.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/3102?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #3102 +/- ##\n==========================================\n- Coverage 77.59% 77.59% -0.01% \n==========================================\n Files 98 98 \n Lines 16250 16250 \n==========================================\n- Hits 12610 12609 -1 \n- Misses 3640 3641 +1\n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/3102?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/modeling\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/3102/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ191dGlscy5weQ==) | `92.86% <0%> (-0.16%)` | :arrow_down: |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/3102?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/3102?src=pr&el=footer). Last update [eec5ec8...2538e35](https://codecov.io/gh/huggingface/transformers/pull/3102?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n", "Thanks for sharing this is awesome!" ]

[]

[]

[ "Thanks a lot for investigating this and submitting a fix, it's awesome.\r\n\r\nI respond in the PR it-self" ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/3100?src=pr&el=h1) Report\n> Merging [#3100](https://codecov.io/gh/huggingface/transformers/pull/3100?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/eec5ec807135ae61fa2266f3c7ad947cc207abda?src=pr&el=desc) will **decrease** coverage by `<.01%`.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/3100?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #3100 +/- ##\n==========================================\n- Coverage 77.59% 77.59% -0.01% \n==========================================\n Files 98 98 \n Lines 16250 16250 \n==========================================\n- Hits 12610 12609 -1 \n- Misses 3640 3641 +1\n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/3100?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/modeling\\_auto.py](https://codecov.io/gh/huggingface/transformers/pull/3100/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19hdXRvLnB5) | `75.47% <ø> (ø)` | :arrow_up: |\n| [src/transformers/modeling\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/3100/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ191dGlscy5weQ==) | `92.86% <0%> (-0.16%)` | :arrow_down: |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/3100?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/3100?src=pr&el=footer). Last update [eec5ec8...44215a4](https://codecov.io/gh/huggingface/transformers/pull/3100?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/3099?src=pr&el=h1) Report\n> Merging [#3099](https://codecov.io/gh/huggingface/transformers/pull/3099?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/eec5ec807135ae61fa2266f3c7ad947cc207abda?src=pr&el=desc) will **decrease** coverage by `<.01%`.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/3099?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #3099 +/- ##\n==========================================\n- Coverage 77.59% 77.59% -0.01% \n==========================================\n Files 98 98 \n Lines 16250 16250 \n==========================================\n- Hits 12610 12609 -1 \n- Misses 3640 3641 +1\n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/3099?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/modeling\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/3099/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ191dGlscy5weQ==) | `92.86% <0%> (-0.16%)` | :arrow_down: |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/3099?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/3099?src=pr&el=footer). Last update [eec5ec8...de4dc15](https://codecov.io/gh/huggingface/transformers/pull/3099?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n" ]

[]

[]

[ "Could you try with `BertForPreTraining` instead of `BertModel` ?", "> Could you try with `BertForPreTraining` instead of `BertModel` ?\r\n\r\n@LysandreJik the error not occurs,thank you very much, but in the previous,when I use \r\n\r\n```py\r\nmodel = BertForNextSentencePrediction.from_pretrained('bert-base-cased')\r\n```\r\nI can get seq_relationship_logits like \r\n\r\n```\r\ntensor([[ 4.6285, -4.9732]], grad_fn=<AddmmBackward>)\r\n```\r\n\r\nafter softmax,I can get probs between seq_A and seq_B,whether seq_B is a continuation of seq_A\r\nlike \r\n```\r\ntensor([[9.9993e-01, 6.7607e-05]], grad_fn=<SoftmaxBackward>)\r\n```\r\nbut when I use BertForPreTraining ,\r\n\r\n```py\r\nmodel = BertForPreTraining.from_pretrained(.....)\r\n```\r\nI get seq_relationship_logits like :\r\n\r\n```\r\ntensor([[[ -7.3790, -7.2666, -7.4841, ..., -6.1682, -5.8256, -6.2910],\r\n [ -7.9165, -8.1490, -7.9572, ..., -6.5870, -6.3568, -6.8383],\r\n [-14.1834, -13.2084, -13.7673, ..., -9.0377, -9.7575, -9.4470],\r\n ...,\r\n [-12.9208, -12.8706, -13.0834, ..., -10.2187, -8.6429, -11.6360],\r\n [-13.2808, -13.3348, -13.2491, ..., -10.7655, -8.8089, -11.0420],\r\n [-15.5444, -15.2074, -15.8938, ..., -11.9712, -12.5488, -14.7295]]],\r\n grad_fn=<AddBackward0>)\r\n```\r\n\r\nhow can I get probs between seq_A and seq_B ? thanks.\r\n\r\n\r\n", "Hi, `BertForNextSentencePrediction` is a model that can only perform the NSP objective. `BertForPreTraining` is a model that can perform both NSP and traditional MLM. \r\n\r\nIt, therefore, outputs tensors for both those tasks, the NSP result is the second value in the output tuple of `BertForPreTraining`:\r\n\r\n```py\r\nfrom transformers import BertForNextSentencePrediction, BertForPreTraining, BertTokenizer\r\n\r\nnsp = BertForNextSentencePrediction.from_pretrained(\"bert-base-cased\")\r\nbpt = BertForPreTraining.from_pretrained(\"bert-base-cased\")\r\ntokenizer = BertTokenizer.from_pretrained(\"bert-base-cased\")\r\n\r\ninputs = tokenizer.encode_plus(\"I like cats.\", \"I like dogs too.\", return_tensors=\"pt\")\r\nnsp_output = nsp(**inputs)\r\nbpt_outupt = bpt(**inputs)[1]\r\n\r\nprint(nsp_output)\r\nprint(bpt_outupt)\r\n```\r\n\r\nreturns\r\n\r\n```\r\n(tensor([[ 5.1543, -5.8147]], grad_fn=<AddmmBackward>),)\r\ntensor([[ 5.1543, -5.8147]], grad_fn=<AddmmBackward>)\r\n```\r\n", "> Hi, `BertForNextSentencePrediction` is a model that can only perform the NSP objective. `BertForPreTraining` is a model that can perform both NSP and traditional MLM.\r\n> \r\n> It, therefore, outputs tensors for both those tasks, the NSP result is the second value in the output tuple of `BertForPreTraining`:\r\n> \r\n> ```python\r\n> from transformers import BertForNextSentencePrediction, BertForPreTraining, BertTokenizer\r\n> \r\n> nsp = BertForNextSentencePrediction.from_pretrained(\"bert-base-cased\")\r\n> bpt = BertForPreTraining.from_pretrained(\"bert-base-cased\")\r\n> tokenizer = BertTokenizer.from_pretrained(\"bert-base-cased\")\r\n> \r\n> inputs = tokenizer.encode_plus(\"I like cats.\", \"I like dogs too.\", return_tensors=\"pt\")\r\n> nsp_output = nsp(**inputs)\r\n> bpt_outupt = bpt(**inputs)[1]\r\n> \r\n> print(nsp_output)\r\n> print(bpt_outupt)\r\n> ```\r\n> \r\n> returns\r\n> \r\n> ```\r\n> (tensor([[ 5.1543, -5.8147]], grad_fn=<AddmmBackward>),)\r\n> tensor([[ 5.1543, -5.8147]], grad_fn=<AddmmBackward>)\r\n> ```\r\n\r\nOK， I got it ,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 } ]

[]

[ "You can't really change the vocabulary without re-training the whole model. Is there some overlap between the BERT vocabulary and your custom vocabulary? If so, you can add the 20k+ tokens using `add_tokens` (which will probably slow down things, as that's a lot of added tokens).", "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", "you can now do this. Keras (tf-nightly version) has added a new util `keras.utils.warmstart_embedding_matrix`. Using this you can continuously train your model with changing vocabulary. https://www.tensorflow.org/api_docs/python/tf/keras/utils/warmstart_embedding_matrix\r\n" ]

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

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

[ "Thanks for reporting this!" ]

[]

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

[ "Hi @Damiox, \r\n\r\nFirst, you have to be careful if you do LM inference with batches of different lengths. Besides using an attention mask, you also have to change the position_ids when using GPT2 (see #3021 ). Considering batch language generation with GPT2 I highly recommend making sure that the input_ids batch is does not need padding (which is actually the case in your example above). If you do have to pad certain batch_idxs then make sure you change the input_position_ids as well and use an attention_mask, but even then it's not promised that GPT2 will generate good results. \r\n\r\nSecond, if your prefix input_ids are padded, let's say:\r\n`[ [ I like cats ], [ I like ] ] -> encode_plus -> [ [0, 1, 2], [0, 1, <PAD> ] ]` and you then append something `-> [ [ that are ], [ dogs] ]`, you essentially put in `[ [0, 1, 2, 4, 5], [0, 1, <PAD>, 3, <PAD>] ]` into the model, which is different from what would happen if you `encode_plus` everything directly (`[ [ 0, 1, 2, 4, 5], [0, 1, 3, <PAD>, <PAD> ] ]`). That's why you a) have to make sure `input_postion_ids` are correct and also b) that you sample from the last not padded token for the next word, which is not that simple (again see #3021 ). \r\nAgain, I recommend to only work with `input_ids` that are not padded or just do `batch_size=1` LM inference. \r\n\r\nThree, when appending words of lists make sure that you have a space `\" \"` between them. In your example above printing the list `[d + n for d, n in zip(docs, docs_next)]` actually gives: \r\n\r\n`['I like tosoda and ', 'Please help mewith this']`,\r\n\r\nwhich is probably not what you want. If you would change `docs_next = [\"soda and \", \"with this\"]` to `docs_next = [\" soda and \", \" with this\"] `both outputs actually produce the same results, but this is only because `docs_tensors['input_ids']` is not padded (same lengths for all batch_idxs)\r\n", "Hey @patrickvonplaten thanks for all your help. Indeed, there was an issue in the code snippet I shared. Thanks!\r\nI see now that pads add complexity because it needs the position_ids to be accommodated, but if it's correctly implemented from my side... shouldn't it just work? I'm confused about your phrase in `it's not promised that GPT2 will generate good results`. ", "GPT2 was never trained on padding tokens so I'm not 100% sure whether you will get good results. But it's for sure worth trying out, if you absolutely have to use batch_size > 1", "Alright. I think it should be fine by manipulating the tensors in the \"past\" as long as I keep consistency across the 24 items in the \"past\". Right?\r\nFor instance... I see that if I generate a past for 1 input with 256 tokens, then I'm getting back a \"past\" which is a list of 24 tensors of 5 dimensions (e.g. (2, 1, 16, 256, 64)). On the other hand, if I generate a past for 2 inputs with 540 tokens, I'm getting back a \"past\" which is a list of 24 tensors of 5 dimensions too (e.g. (2, 2, 16, 540, 64)). So I think that if I wanted to exclude the last sentence from the last \"past\" I can simply manipulate the 2nd dimension in all the 24 items and drop the corresponding value from it. I guess... ? ", "> So I think that if I wanted to exclude the last sentence from the last \"past\" I can simply manipulate the 2nd dimension in all the 24 items and drop the corresponding value from it. I guess... ?\r\n\r\nI don't really understand this sentence. If you are concerned about sampling from a padded past, you don't have to manipulate the past variable I think.\r\n\r\nLet's say we forwarded:` ['I', 'live', 'in', <PAD>, <PAD>]` then the past variable will be of dimension `(2, 5, 16, num_tokens, 64)`. If you now want to sample from \r\n`past = key, values of ['I', 'live', 'in', <PAD>, <PAD>]` and \r\n`input_id = ['London'] `\r\nthen you just have to sample from the outputted token because the operation is the same as \r\nsampling from the last token of `['I', 'live', 'in', <PAD>, <PAD>, 'London']` and the last token is not a `<PAD>` token from which output logits it can be sampled from. ", "@patrickvonplaten I think from your above example, the past variable will be `(2, 1, 16, num_tokens, 64)` instead of `(2, 5, 16, num_tokens, 64`. Right?\r\n\r\n> So I think that if I wanted to exclude the last sentence from the last \"past\" I can simply manipulate the 2nd dimension in all the 24 items and drop the corresponding value from it. I guess... ?\r\n\r\nWhat I meant here is to manipulate the past tensors in case the tensor dimension I want to predict for the subsequent batches should exclude some sentence. I found removing information from the past tensors may be complicated. So please see my example above with another approach to use \"past\" I am experimenting now.\r\n\r\nFor now to simplify this a bit, I'm not focusing on padding the past. Thus, it'll be generated from a batch of N docs. For instance below N=3:\r\n```\r\n['I', 'live', 'in', 'NYC']\r\n['I', 'live', 'in', 'Barcelona']\r\n['I', 'live', 'in', 'Moscow']\r\n```\r\nNote: I'm going to make sure to group docs by tokens length. I'm going to get a `past` from that. Past here will be a list of 24 elements with dimensions `(2, 3, 16, 4, 64)`\r\n\r\nThen I'm planning to use that past variable along with M suffix phrases. Those suffix phrases may have different lengths and below to different sentences that were calculated above, so I'm planning to add padding here first. Also another characteristic is that M will be always equal to or greater than N. For example, M=6 (it's a coincide that num_tokens is also 6 here):\r\n```\r\n['and', 'I', 'speak', 'english', 'fluently', '<PAD>']\r\n['and', 'I', 'live', 'in', 'North', 'Manhattan']\r\n['and', 'I', 'like', 'football', '<PAD>', '<PAD>']\r\n['and', 'I', 'don\\'t', 'speak', 'catalan', '<PAD>']\r\n['and', 'I', 'take', 'the', 'bus', '<PAD>']\r\n['and', 'I', 'drink', 'vodka', 'sometimes', 'you?']\r\n```\r\nNote: I'm also building an attention mask properly as you indicated by concatenating a tensor full with 1s with length = 4 to make this work. For example:\r\n```\r\n['1', '1', '1', '1', '1', '1', '1', '1', '1', '0']\r\n['1', '1', '1', '1', '1', '1', '1', '1', '1', '1']\r\n['1', '1', '1', '1', '1', '1', '1', '1', '0', '0']\r\n['1', '1', '1', '1', '1', '1', '1', '1', '1', '0']\r\n['1', '1', '1', '1', '1', '1', '1', '1', '1', '0'] \r\n['1', '1', '1', '1', '1', '1', '1', '1', '1', '1']\r\n```\r\n\r\nTo make \"past\" fit here, I'm planning to expand the past variable. So for each tensor in the past array, expand the tensor results as if I had sent to gpt2 initially the following first batch to get the \"past\" for:\r\n```\r\n['I', 'live', 'in', 'NYC']\r\n['I', 'live', 'in', 'NYC']\r\n['I', 'live', 'in', 'NYC']\r\n['I', 'live', 'in', 'Barcelona']\r\n['I', 'live', 'in', 'Barcelona']\r\n['I', 'live', 'in', 'Moscow']\r\n```\r\n\r\n\r\nIn this case, I'll build a past with dimensions: `(2, 6, 16, 4, 64)` instead of the original dimensions: `(2, 3, 16, 4, 64)`. I found https://pytorch.org/docs/stable/torch.html#torch.repeat_interleave very useful for this...\r\n\r\nDo you think this make sense? Any warning about this? Thanks", "@patrickvonplaten can I not manipulate the `past` variable as explained above? Do the other dimensions contain some kind of aggregated data that makes the past to be immutable?", "@patrickvonplaten just to confirm I'm on the right track...: Can I manipulate the dimensions for the layer tensors in the past array? It's working for me with latest release (2.6.0), but just wanted to make sure I'm on the right track if I go ahead with this. So basically I want to make sure I can expand the dimension 1 of every tensor in the past array from N to M. So that I can re-use that past with different predictions that will reuse that past. Hope it makes sense my question. " ]

[]

[]

[ "@julien-c any idea why this would cause \r\n```\r\nFAILED tests/test_pipelines.py::MonoColumnInputTestCase::test_tf_feature_extraction \r\n```", "I think this one might sometimes fail randomly" ]

[]

[]

[]

[ { "id": 1260952223, "node_id": "MDU6TGFiZWwxMjYwOTUyMjIz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Discussion", "name": "Discussion", "color": "22870e", "default": false, "description": "Discussion on a topic (keep it focused or open a new issue though)" }, { "id": 1834056635, "node_id": "MDU6TGFiZWwxODM0MDU2NjM1", "url": "https://api.github.com/repos/huggingface/transformers/labels/Core:%20Tokenization", "name": "Core: Tokenization", "color": "FF4446", "default": false, "description": "Internals of the library; Tokenization." } ]

[]

[ "The proper way to decode a value is using the `decode` method:\r\n\r\n```py\r\nfrom transformers import GPT2Tokenizer\r\ntokenizer = GPT2Tokenizer.from_pretrained(\"gpt2\")\r\n\r\ntokenizer.decode([198]) # '\\n'\r\n```\r\n\r\nSome byte indices are shifted in the GPT-2 vocabulary, especially the control characters and characters that resemble whitespace. This is an example, and you can see the method that does it in `~transformers.tokenization_gpt2.bytes_to_unicode`." ]

[ { "id": 1834056635, "node_id": "MDU6TGFiZWwxODM0MDU2NjM1", "url": "https://api.github.com/repos/huggingface/transformers/labels/Core:%20Tokenization", "name": "Core: Tokenization", "color": "FF4446", "default": false, "description": "Internals of the library; Tokenization." } ]

[]

[ "This is now fixed on `master`:\r\n```\r\n>>> t = transformers.AutoTokenizer.from_pretrained(\"bert-base-cased\", use_fast=True)\r\n>>> t.encode_plus(\" \", add_special_tokens=False)\r\n{'input_ids': [], 'token_type_ids': [], 'attention_mask': []}\r\n```\r\nAlso, `add_special_tokens` now works exactly the same for both slow and fast tokenizers. We give it with `encode`, `tokenize`, ... and not during initialization anymore." ]

[]

[]

[ "Tried debugging with CPU (an aside - this has an issue in itself apparently when --no_cuda flag is used --> run_language_modeling.py needs to set args.n_gpu to 0)\r\n\r\nFound the fix -> Needed to call model.resize_token_embeddings(len(tokenizer)) after adding tokens in the eval mode as well. " ]

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

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/3089?src=pr&el=h1) Report\n> Merging [#3089](https://codecov.io/gh/huggingface/transformers/pull/3089?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/f169957d0cf17b110f27cacc1b1fb43efaa01218?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/3089?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #3089 +/- ##\n==========================================\n+ Coverage 77.59% 77.59% +<.01% \n==========================================\n Files 98 98 \n Lines 16250 16250 \n==========================================\n+ Hits 12609 12610 +1 \n+ Misses 3641 3640 -1\n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/3089?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/modeling\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/3089/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ191dGlscy5weQ==) | `93.02% <0%> (+0.15%)` | :arrow_up: |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/3089?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/3089?src=pr&el=footer). Last update [f169957...ac7be7c](https://codecov.io/gh/huggingface/transformers/pull/3089?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n", "Thanks for sharing!" ]

[ { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null }, { "id": 1834056635, "node_id": "MDU6TGFiZWwxODM0MDU2NjM1", "url": "https://api.github.com/repos/huggingface/transformers/labels/Core:%20Tokenization", "name": "Core: Tokenization", "color": "FF4446", "default": false, "description": "Internals of the library; Tokenization." } ]

[]

[ "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" ]

[]

[]

[ "You should not have any issues with this as `GradientAccumulator` is correctly imported in the __init__.py file:\r\n\r\nhttps://github.com/huggingface/transformers/blob/298bed16a841fae3608d334441ccae4d9043611f/src/transformers/__init__.py#L426-L427\r\n\r\nIs `transformers` correctly installed in your pip environment, or did you simply clone the repository?", "Ok. Thank you. I cannot check it now because Colab PRO accounts are having problems. I will let you know ASAP.", "I met analogous problem，but it was solved after rebooting computer. And another issue is it can not find the installed module，this issue can be solved by reinstall the module that can not be found.", "In my case I was not using tf 2.x\n\nEl mar., 10 mar. 2020 9:50, over-shine <[email protected]> escribió:\n\n> I met analogous problem，but it was solved after rebooting computer. And\n> another issue is it can not find the installed module，this issue can be\n> solved by reinstall the module that can not be found.\n>\n> —\n> You are receiving this because you modified the open/close state.\n> Reply to this email directly, view it on GitHub\n> <https://github.com/huggingface/transformers/issues/3087?email_source=notifications&email_token=AA34BHNOLTNBPN37WKT4UELRGX5MHA5CNFSM4K73DTO2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEOKQYOY#issuecomment-596970555>,\n> or unsubscribe\n> <https://github.com/notifications/unsubscribe-auth/AA34BHIJAGHRKSZE5TNILB3RGX5MHANCNFSM4K73DTOQ>\n> .\n>\n" ]

[ { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null }, { "id": 1834054694, "node_id": "MDU6TGFiZWwxODM0MDU0Njk0", "url": "https://api.github.com/repos/huggingface/transformers/labels/TensorFlow", "name": "TensorFlow", "color": "FF6F00", "default": false, "description": "Anything TensorFlow" }, { "id": 1834056761, "node_id": "MDU6TGFiZWwxODM0MDU2NzYx", "url": "https://api.github.com/repos/huggingface/transformers/labels/Core:%20Modeling", "name": "Core: Modeling", "color": "FF8446", "default": false, "description": "Internals of the library; Models." } ]

[]

[ "TFBertModel can auto use gpu? i find model(np.array(input_ids)) is slow,cost 100+ms\r\nIn [14]: %timeit model_outputs = model(np.array(input_ids))\r\n133 ms ± 3.4 ms per loop (mean ± std. dev. of 7 runs, 10 loops each)\r\n\r\nIn [15]: %time model_outputs = model(np.array(input_ids))\r\nCPU times: user 330 ms, sys: 14.4 ms, total: 344 ms\r\nWall time: 158 ms\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", "I have a similar issue. Did you manage to find a fix for this @msandrewhoang03 ?" ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/3085?src=pr&el=h1) Report\n> Merging [#3085](https://codecov.io/gh/huggingface/transformers/pull/3085?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/0e56b37e805279ecb61670159fa8c71487214e0a?src=pr&el=desc) will **decrease** coverage by `1.02%`.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/3085?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #3085 +/- ##\n=========================================\n- Coverage 77.62% 76.6% -1.03% \n=========================================\n Files 98 98 \n Lines 16230 16230 \n=========================================\n- Hits 12599 12433 -166 \n- Misses 3631 3797 +166\n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/3085?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/modeling\\_tf\\_pytorch\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/3085/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9weXRvcmNoX3V0aWxzLnB5) | `8.72% <0%> (-81.21%)` | :arrow_down: |\n| [src/transformers/modeling\\_roberta.py](https://codecov.io/gh/huggingface/transformers/pull/3085/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19yb2JlcnRhLnB5) | `85.71% <0%> (-10%)` | :arrow_down: |\n| [src/transformers/modeling\\_xlnet.py](https://codecov.io/gh/huggingface/transformers/pull/3085/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ194bG5ldC5weQ==) | `73.48% <0%> (-2.3%)` | :arrow_down: |\n| [src/transformers/modeling\\_ctrl.py](https://codecov.io/gh/huggingface/transformers/pull/3085/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19jdHJsLnB5) | `96.03% <0%> (-2.21%)` | :arrow_down: |\n| [src/transformers/modeling\\_openai.py](https://codecov.io/gh/huggingface/transformers/pull/3085/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19vcGVuYWkucHk=) | `80.2% <0%> (-1.35%)` | :arrow_down: |\n| [src/transformers/modeling\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/3085/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ191dGlscy5weQ==) | `93.02% <0%> (-0.16%)` | :arrow_down: |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/3085?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/3085?src=pr&el=footer). Last update [0e56b37...83f65ff](https://codecov.io/gh/huggingface/transformers/pull/3085?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n" ]

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

[]

[]

[ { "id": 1260952223, "node_id": "MDU6TGFiZWwxMjYwOTUyMjIz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Discussion", "name": "Discussion", "color": "22870e", "default": false, "description": "Discussion on a topic (keep it focused or open a new issue though)" }, { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null }, { "id": 1834052129, "node_id": "MDU6TGFiZWwxODM0MDUyMTI5", "url": "https://api.github.com/repos/huggingface/transformers/labels/High-Level%20feature", "name": "High-Level feature", "color": "f7c9a3", "default": false, "description": "" }, { "id": 1834056635, "node_id": "MDU6TGFiZWwxODM0MDU2NjM1", "url": "https://api.github.com/repos/huggingface/transformers/labels/Core:%20Tokenization", "name": "Core: Tokenization", "color": "FF4446", "default": false, "description": "Internals of the library; Tokenization." } ]

[]

[ "I agree that an on-the-fly tokenisation would be neat as an alternative to pre-processing the whole input file and saving the tensors in memory. ", "Hi @BramVanroy , this is mentioned in the blog post bout training models from scratch, as something that could be done (https://huggingface.co/blog/how-to-train). Is it possible?\r\nThanks!", "I am fairly new to how contributing to HuggingFace works but gave this a little thought today.\r\nAt first I thought we could maybe solve it like this:\r\n\r\nIf we consider this code:\r\n`tokenized_text = tokenizer.convert_tokens_to_ids(tokenizer.tokenize(text))`\r\n\r\nMy first intuition was that the space needed to save the result of this code is significantly less than the space needed to store f.read(). \r\nSo if we could get the result of this code by just reading a line (or reading parts of the texts with overlaps to the previous part as big as max_len of tokens), we might solve it...\r\nHowever I ran a litte experiment and it turns out that `tokenized_text` would still take up around 133GB with an input txt file of 200GB.\r\nSo not a solution.\r\nDo you guys have any idea how to approach this differently? Because of the RandomSampler we also can't store parts of the file trivially.", "The solution would be to work with a dataset that on every call, fetches the lines from the file that are required, rather than reading the whole file in memory. This is bound to be slower, but it is very lenient on memory. A good candidate is linecache, which does smart caching in the process.\r\n\r\n```python\r\nimport linecache\r\nfrom pathlib import Path\r\n\r\nfrom torch.utils.data import Dataset\r\n\r\n\r\nclass LazyTextDataset(Dataset):\r\n def __init__(self, fin):\r\n self.fin = fin\r\n self.num_entries = self._get_n_lines(self.fin)\r\n\r\n @staticmethod\r\n def _get_n_lines(fin):\r\n with Path(fin).resolve().open(encoding='utf-8') as fhin:\r\n for line_idx, _ in enumerate(fhin, 1):\r\n pass\r\n\r\n return line_idx\r\n\r\n def __getitem__(self, idx):\r\n # linecache starts counting from one, not zero, +1 the given index\r\n idx += 1\r\n return linecache.getline(self.fin, idx)\r\n\r\n def __len__(self):\r\n return self.num_entries\r\n```\r\n\r\nThen you can add a custom `collate_fn` to your data loader that will automatically tokenize the whole batch.", "I thought about this but figured it would definitely be too slow for such big files. I didn't know about linecache though, cool!", "I've been using a similar solution to @BramVanroy for a couple of weeks, though I too was not aware of `linecache`, so assume that my solution can be improved by using that tool. \r\n\r\nI implemented this because the up-front loading was taking hours and hours. I did some rough comparisons on smaller data files and found that I was getting the same iters/second using this method as the existing methods.\r\n\r\n\r\n```python\r\nclass LazyUnSupervisedTextExamples:\r\n \"\"\"\r\n Deals with file i/o for lazy retrieval of specific lines of text in file.\r\n\r\n \"\"\"\r\n def __init__(self, path):\r\n \"\"\"\r\n :args:\r\n path: str\r\n : The path to the data file to be loaded.\r\n \"\"\"\r\n self.data_path = path\r\n self.data_stream = open(self.data_path, 'r')\r\n self.offsets = [0]\r\n for line in self.data_stream:\r\n self.offsets.append(self.offsets[-1] + len(line.encode('utf-8')))\r\n self.offsets = self.offsets[1:-1]\r\n self.data_stream.seek(0)\r\n self.current_offset = 0\r\n\r\n def __len__(self):\r\n return len(self.offsets)\r\n\r\n def __getitem__(self, _id):\r\n \"\"\"\r\n :returns:\r\n str; the line of text given by _id if no errors.\r\n None if errors occur.\r\n\r\n PEP8 note: we really do want a bare exception here because an uncaught exception in here has the potential\r\n to bring down a large training run with an error in a single line of the data file.\r\n \"\"\"\r\n\r\n offset = self.offsets[_id]\r\n try:\r\n self.data_stream.seek(offset)\r\n line = self.data_stream.readline()\r\n example = line.strip()\r\n self.data_stream.seek(self.current_offset)\r\n except:\r\n example = None\r\n return example\r\n\r\n def __next__(self):\r\n line = self.data_stream.readline()\r\n self.current_offset = self.data_stream.tell()\r\n return line.strip()\r\n\r\n def close(self):\r\n self.data_stream.close()\r\n\r\n\r\nclass LazyUnSupervisedTextDataset(Dataset):\r\n \"\"\"\r\n Works with datasets of simple lines of text. Lines are loaded and tokenized\r\n lazily rather than being pulled into memory up-front. This reduces the memory\r\n footprint when using large datasets, and also remedies a problem seen when using\r\n the other Datasets (above) whereby they take too long to load all\r\n of the data and tokenize it before doing any training.\r\n\r\n The file i/o work is handled within self.examples. This class just indexes\r\n into that object and applies the tokenization.\r\n \"\"\"\r\n def __init__(self, tokenizer, file_path, block_size=512):\r\n \"\"\"\r\n :args:\r\n tokenizer: tokenizer.implementations.BaseTokenizer object (instantiated)\r\n : This tokenizer will be directly applied to the text data\r\n to prepare the data for passing through the model.\r\n file_path: str\r\n : Path to the data file to be used.\r\n block_size: int\r\n : The maximum length of a sequence (truancated beyond this length).\r\n :returns: None.\r\n \"\"\"\r\n self.examples = LazyUnSupervisedTextExamples(file_path)\r\n self.tokenizer = tokenizer\r\n self.max_len = block_size\r\n\r\n def __len__(self):\r\n return len(self.examples)\r\n\r\n def _text_to_tensor(self, item):\r\n \"\"\"\r\n Defines the logic for transforming a single raw text item to a tokenized\r\n tensor ready to be passed into a model.\r\n\r\n :args:\r\n item: str\r\n : The text item as a string to be passed to the tokenizer.\r\n \"\"\"\r\n return torch.tensor(self.tokenizer.encode(item, max_length=self.max_len))\r\n\r\n def _text_to_item(self, text):\r\n \"\"\"\r\n Convenience functino to encapsulate re-used logic for converting raw\r\n text to the output of __getitem__ of __next__.\r\n\r\n :returns:\r\n torch.Tensor of tokenized text if no errors.\r\n None if any errors encountered.\r\n \"\"\"\r\n try:\r\n if (text is not None):\r\n return self._text_to_tensor(text)\r\n else:\r\n return None\r\n except:\r\n return None\r\n\r\n def __getitem__(self, _id):\r\n \"\"\"\r\n :returns:\r\n torch.Tensor of tokenized text if no errors.\r\n None if any errors encountered.\r\n \"\"\"\r\n text = self.examples[_id]\r\n return self._text_to_item(text)\r\n\r\n def __next__(self):\r\n text = next(self.examples)\r\n return self._text_to_item(text)\r\n\r\n def close(self):\r\n \"\"\"\r\n Since the LazyUnSupervisedTextExamples object self.examples contains a\r\n file handle, this method provides access to its close function to safely\r\n close the open data file when finished. This should be run when the\r\n dataset object is finished with.\r\n \"\"\"\r\n self.examples.close()\r\n```\r\n\r\nThe only change I found necessary to make to the `collate_fn` was a line to filter out lines that failed to load. I'm currently tokenising one item at a time, but prefer @BramVanroy's suggestion of batch tokenisation in the `collate_fn`. \r\n\r\n```python\r\ndef collate(examples: List[torch.Tensor]):\r\n examples = list(filter(lambda ex: ex is not None, examples))\r\n if tokenizer._pad_token is None:\r\n return pad_sequence(examples, batch_first=True)\r\n return pad_sequence(examples, batch_first=True, padding_value=tokenizer.pad_token_id)\r\n```\r\n\r\n\r\nHappy to make above-mentioned sensible changes and contribute.\r\n\r\nDoes anyone have any advice about more sophisticated performance testing to shore-up my above claim that lazy loading isn't any slower per iteration?", "I would recommend to, indeed, run the tokenisation in collate_fn. You can use `batch_encode_plus` there. Concerning your collate function: the filter function can be simplified to `filter(None, examples)` but in fact I'd go with a list comprehension right away: `[ex for ex in examples if ex is not None]`. \r\n\r\nFor timing you can use the timeit module.", "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/3082?src=pr&el=h1) Report\n> Merging [#3082](https://codecov.io/gh/huggingface/transformers/pull/3082?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/b74c9b106b97b0722ff8f98e77e2e2210c688b23?src=pr&el=desc) will **increase** coverage by `0.4%`.\n> The diff coverage is `87.27%`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/3082?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #3082 +/- ##\n========================================\n+ Coverage 77.19% 77.6% +0.4% \n========================================\n Files 98 98 \n Lines 16063 16219 +156 \n========================================\n+ Hits 12400 12586 +186 \n+ Misses 3663 3633 -30\n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/3082?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/tokenization\\_bart.py](https://codecov.io/gh/huggingface/transformers/pull/3082/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy90b2tlbml6YXRpb25fYmFydC5weQ==) | `100% <100%> (ø)` | :arrow_up: |\n| [src/transformers/configuration\\_bart.py](https://codecov.io/gh/huggingface/transformers/pull/3082/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9jb25maWd1cmF0aW9uX2JhcnQucHk=) | `100% <100%> (ø)` | :arrow_up: |\n| [src/transformers/modeling\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/3082/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ191dGlscy5weQ==) | `92.76% <76.66%> (+0.38%)` | :arrow_up: |\n| [src/transformers/modeling\\_bart.py](https://codecov.io/gh/huggingface/transformers/pull/3082/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19iYXJ0LnB5) | `93.05% <88.82%> (+8.47%)` | :arrow_up: |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/3082?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/3082?src=pr&el=footer). Last update [b74c9b1...5656b5e](https://codecov.io/gh/huggingface/transformers/pull/3082?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n", "\r\nNew organization @LysandreJik ", "I like that organisation" ]

[ { "id": 1260952223, "node_id": "MDU6TGFiZWwxMjYwOTUyMjIz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Discussion", "name": "Discussion", "color": "22870e", "default": false, "description": "Discussion on a topic (keep it focused or open a new issue though)" }, { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null }, { "id": 1834056761, "node_id": "MDU6TGFiZWwxODM0MDU2NzYx", "url": "https://api.github.com/repos/huggingface/transformers/labels/Core:%20Modeling", "name": "Core: Modeling", "color": "FF8446", "default": false, "description": "Internals of the library; Models." } ]

[]

[ "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": 1834088753, "node_id": "MDU6TGFiZWwxODM0MDg4NzUz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Tests", "name": "Tests", "color": "a6fcca", "default": false, "description": "Related to tests" }, { "id": 1862634478, "node_id": "MDU6TGFiZWwxODYyNjM0NDc4", "url": "https://api.github.com/repos/huggingface/transformers/labels/Should%20Fix", "name": "Should Fix", "color": "FF0000", "default": false, "description": "This has been identified as a bug and should be fixed." } ]

[ { "login": "mfuntowicz", "id": 2241520, "node_id": "MDQ6VXNlcjIyNDE1MjA=", "avatar_url": "https://avatars.githubusercontent.com/u/2241520?v=4", "gravatar_id": "", "url": "https://api.github.com/users/mfuntowicz", "html_url": "https://github.com/mfuntowicz", "followers_url": "https://api.github.com/users/mfuntowicz/followers", "following_url": "https://api.github.com/users/mfuntowicz/following{/other_user}", "gists_url": "https://api.github.com/users/mfuntowicz/gists{/gist_id}", "starred_url": "https://api.github.com/users/mfuntowicz/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/mfuntowicz/subscriptions", "organizations_url": "https://api.github.com/users/mfuntowicz/orgs", "repos_url": "https://api.github.com/users/mfuntowicz/repos", "events_url": "https://api.github.com/users/mfuntowicz/events{/privacy}", "received_events_url": "https://api.github.com/users/mfuntowicz/received_events", "type": "User", "site_admin": false } ]

[ "Should be fixed, this tests were running because I asked to build on every master's commit. \r\n\r\nNow it should only build on new tags." ]

[]

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

[ "I just merged a PR and your example works for me. Would you mind seeing if it is still broken in your system? Thanks for posting! \r\n(note that the fix won't be in v 2.5.1 if you pip installed). \r\n", "Reopen if still broken!" ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/3078?src=pr&el=h1) Report\n> Merging [#3078](https://codecov.io/gh/huggingface/transformers/pull/3078?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/c0135194ebc5de4b1bbef98b31f9c457a0bf746a?src=pr&el=desc) will **decrease** coverage by `0.99%`.\n> The diff coverage is `100%`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/3078?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #3078 +/- ##\n=======================================\n- Coverage 77.6% 76.6% -1% \n=======================================\n Files 98 98 \n Lines 16221 16230 +9 \n=======================================\n- Hits 12588 12433 -155 \n- Misses 3633 3797 +164\n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/3078?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/modeling\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/3078/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ191dGlscy5weQ==) | `93.02% <100%> (+0.26%)` | :arrow_up: |\n| [src/transformers/modeling\\_tf\\_pytorch\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/3078/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9weXRvcmNoX3V0aWxzLnB5) | `8.72% <0%> (-81.21%)` | :arrow_down: |\n| [src/transformers/modeling\\_roberta.py](https://codecov.io/gh/huggingface/transformers/pull/3078/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19yb2JlcnRhLnB5) | `85.71% <0%> (-10%)` | :arrow_down: |\n| [src/transformers/modeling\\_xlnet.py](https://codecov.io/gh/huggingface/transformers/pull/3078/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ194bG5ldC5weQ==) | `73.48% <0%> (-2.3%)` | :arrow_down: |\n| [src/transformers/modeling\\_ctrl.py](https://codecov.io/gh/huggingface/transformers/pull/3078/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19jdHJsLnB5) | `96.03% <0%> (-2.21%)` | :arrow_down: |\n| [src/transformers/modeling\\_openai.py](https://codecov.io/gh/huggingface/transformers/pull/3078/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19vcGVuYWkucHk=) | `80.2% <0%> (-1.35%)` | :arrow_down: |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/3078?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/3078?src=pr&el=footer). Last update [c013519...5b9164a](https://codecov.io/gh/huggingface/transformers/pull/3078?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n", "The self-hosted runner tests that fail are:\r\n\r\nFAILED tests/test_pipelines.py::MonoColumnInputTestCase::test_tf_feature_extraction\r\nFAILED tests/test_pipelines.py::MonoColumnInputTestCase::test_tf_fill_mask\r\nFAILED tests/test_pipelines.py::MonoColumnInputTestCase::test_tf_ner\r\nFAILED tests/test_pipelines.py::MonoColumnInputTestCase::test_tf_sentiment_analysis\r\n\r\nand are all related to memory exhaustion (`Resource exhausted: OOM when allocating tensor ... `). \r\nThe tests are not related to the PR. Not sure what to do @julien-c \r\n", "LGTM for merging", "Discussed with @thomwolf as well and also agreed that generate() function is not too complex and good as it is now. I will take a closer look at the issue with beam search decoding when `do_sample=True` (PR #2317 ) in a separate PR. Good to merge for me! " ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/3077?src=pr&el=h1) Report\n> Merging [#3077](https://codecov.io/gh/huggingface/transformers/pull/3077?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/298bed16a841fae3608d334441ccae4d9043611f?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/3077?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #3077 +/- ##\n==========================================\n+ Coverage 77.18% 77.19% +<.01% \n==========================================\n Files 98 98 \n Lines 16063 16063 \n==========================================\n+ Hits 12399 12400 +1 \n+ Misses 3664 3663 -1\n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/3077?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/modeling\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/3077/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ191dGlscy5weQ==) | `92.38% <0%> (+0.16%)` | :arrow_up: |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/3077?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/3077?src=pr&el=footer). Last update [298bed1...69041c1](https://codecov.io/gh/huggingface/transformers/pull/3077?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n" ]

[ { "id": 1108649053, "node_id": "MDU6TGFiZWwxMTA4NjQ5MDUz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Help%20wanted", "name": "Help wanted", "color": "008672", "default": false, "description": "Extra attention is needed, help appreciated" }, { "id": 1260952223, "node_id": "MDU6TGFiZWwxMjYwOTUyMjIz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Discussion", "name": "Discussion", "color": "22870e", "default": false, "description": "Discussion on a topic (keep it focused or open a new issue though)" }, { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null }, { "id": 1834056761, "node_id": "MDU6TGFiZWwxODM0MDU2NzYx", "url": "https://api.github.com/repos/huggingface/transformers/labels/Core:%20Modeling", "name": "Core: Modeling", "color": "FF8446", "default": false, "description": "Internals of the library; Models." } ]

[]

[ "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 }, { "id": 1834081910, "node_id": "MDU6TGFiZWwxODM0MDgxOTEw", "url": "https://api.github.com/repos/huggingface/transformers/labels/Usage", "name": "Usage", "color": "e28436", "default": false, "description": "General questions about the library" } ]

[]

[ "Maybe this is a little late but you could take a look in both `examples/run_tf_glue.py` and [this function](https://github.com/huggingface/transformers/blob/master/src/transformers/data/processors/glue.py#L31-L168) from`src/transformers/data/processors/glue.py` and write a custom training script based from those.", "To make things a little more concrete, I've written and annotated [an end-to-end example](https://gist.github.com/papapabi/124c6ac406e6bbd1f28df732e953ac6d) of how to fine-tune a `bert-base-cased` model from your `DataFrame`'s spec. Do comment if it helps you out!", "@papapabi Thank you for your inputs. I will check this out.\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 }, { "id": 1834081910, "node_id": "MDU6TGFiZWwxODM0MDgxOTEw", "url": "https://api.github.com/repos/huggingface/transformers/labels/Usage", "name": "Usage", "color": "e28436", "default": false, "description": "General questions about the library" } ]

[]

[ "I'm not sure I understand your question. The O tag is here in order to identify a token which is not an entity in the IOB1 tagging scheme. If you don't have such a tag, every token will have to be classified as an entity, which does not make sense?\r\n\r\nIf you would like to define a custom tagging scheme and train a model to predict on that tagging scheme, you would have to create a dataset for that and train your model on that dataset.", "Thanks @LysandreJik \r\nLet me provide some additional details. Given a sentence, I use some external resources to find what are the candidates for tagging. For the candidates, I need to classify between 2 different labels (binary classification). That's the reason why I wrote that I am not interested in predicting the O tag, since I use external resources for it. \r\nI have data for train/test. ", "Hey @jmamou , we might be able to help. Mind sending me an email at clement [at] huggingface [dot] co?", "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 } ]

[]

[ "Please post your code using [code blocks](https://help.github.com/en/github/writing-on-github/creating-and-highlighting-code-blocks). Don't post screenshots.", "FYI: \r\nphoto of **format input data**; https://i.stack.imgur.com/t472b.png\r\nphoto of **tag2name** ; https://i.stack.imgur.com/RO7dp.png\r\n\r\n\r\nAssuming the data goes in in the right format, here is the model initialization and evaluation loop.\r\n```\r\ntokenizer = AutoTokenizer.from_pretrained(\"bert-base-cased-finetuned-conll03-english\")\r\nmodel = BertForTokenClassification.from_pretrained('bert-base-cased-finetuned-conll03-english')\r\n\r\n\r\n#eval LOOP\r\n\r\nmodel.eval();\r\neval_loss, eval_accuracy = 0, 0\r\nnb_eval_steps, nb_eval_examples = 0, 0\r\ny_true = []\r\ny_pred = []\r\nvaldataset = []\r\n\r\nprint(\"***** Running evaluation *****\")\r\nprint(\" Num examples ={}\".format(len(val_inputs)))\r\nprint(\" Batch size = {}\".format(batch_num))\r\n\r\nfor step, batch in enumerate(valid_dataloader):\r\n batch = tuple(t.to(device) for t in batch) # set every example of batch to device\r\n input_ids, input_mask, label_ids = batch #same as we did in training loop but only 1 epoch now\r\n \r\n \r\n with torch.no_grad(): #means we don't care about gradients and updating tensors \r\n outputs = model(input_ids, token_type_ids=None,\r\n attention_mask=input_mask)\r\n # For eval mode, the first result of outputs is logits (for training mode this was loss)\r\n logits = outputs[0] # In context of deep learning the logits layer means the layer that feeds in to softmax (or other such normalization).\r\n \r\n # Get NER predict result\r\n logits = torch.argmax(F.log_softmax(logits,dim=2),dim=2)#feed logits into softmax and take the prediction that is maximal \r\n logits = logits.detach().cpu().numpy()\r\n \r\n if step==1:\r\n print(logits[0][0:15])\r\n print(logits[1][0:15])\r\n print(logits[3][0:15])\r\n print(logits[4][0:15])\r\n print(logits[5][0:15])\r\n\r\n print(label_ids[0][0:15])\r\n print(label_ids[1][0:15])\r\n print(label_ids[2][0:15])\r\n print(label_ids[3][0:15])\r\n \r\n \r\n # Get NER true result\r\n label_ids = label_ids.to('cpu').numpy()\r\n \r\n \r\n # Only predict the real word, mark=0, will not calculate\r\n input_mask = input_mask.to('cpu').numpy()\r\n \r\n # Compare the valuable predict result\r\n for i,mask in enumerate(input_mask):\r\n # Real one\r\n temp_1 = []\r\n # Predicted one\r\n temp_2 = []\r\n \r\n valtemp = []\r\n\r\n for j, m in enumerate(mask):\r\n # Mark=0, meaning its a pad word, dont compare\r\n if m:\r\n if tag2name[label_ids[i][j]] != \"X\" and tag2name[label_ids[i][j]] != \"[CLS]\" and tag2name[label_ids[i][j]] != \"[SEP]\" : # Exclude the X label\r\n temp_1.append(tag2name[label_ids[i][j]])\r\n temp_2.append(tag2name[logits[i][j]])\r\n \r\n if tag2name[label_ids[i][j]] != \"[CLS]\" and tag2name[label_ids[i][j]] != \"[SEP]\" :\r\n valtemp.append(input_ids[i][j].item())\r\n \r\n else:\r\n break\r\n \r\n #here are the two lists that contain true and pred labels. \r\n y_true.append(temp_1)\r\n y_pred.append(temp_2)\r\n\r\n \r\n valdataset.append(valtemp)\r\n\r\ntokenized_text_con = [tokenizer.decode(val) for val in valdataset]\r\n\r\n\r\n\r\n ```\r\nprint output: https://i.stack.imgur.com/qS62L.png\r\n\r\n", "Hi! From my experience using the community-contributed `dbmdz/bert-large-cased-finetuned-conll03-english` (which is the same checkpoint as) `bert-large-cased-finetuned-conll03-english`, using the `bert-base-cased` tokenizer instead of the tokenizer loaded from that checkpoint works better.\r\n\r\nYou can see an example of this in the [usage](https://huggingface.co/transformers/usage.html#named-entity-recognition), let me know if it helps.\r\n\r\nI suspect the difference between the tokenizers is due to a lowercasing of all inputs. I'm looking into it now.\r\n\r\nPS: the file `bert-large-cased-finetuned-conll03-english` is deprecated in favor of the aforementionned `dbmdz/bert-large-cased-finetuned-conll03-english` as they are duplicates. @julien-c is currently deleting it from the S3, please use the `dbmdz` file/folder.", "Also cc'ing @stefan-it for information :)", "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 am not sure which Chinese BERT you are referring to, but the original multilingual BERT has trained Chinese on the character-level. From their [README](https://github.com/google-research/bert/blob/master/multilingual.md):\r\n\r\n> Because Chinese (and Japanese Kanji and Korean Hanja) does not have whitespace characters, we add spaces around every character in the CJK Unicode range before applying WordPiece. This means that Chinese is effectively character-tokenized. Note that the CJK Unicode block only includes Chinese-origin characters and does not include Hangul Korean or Katakana/Hiragana Japanese, which are tokenized with whitespace+WordPiece like all other languages.", "OK, Thanks for your help \r\nBut I mean I want Chinese BERT for word-level not for character-level.", "Yes, I understand that, and as I said the default multilingual BERT does **not** support that. You'll have to find another implementation, perhaps https://arxiv.org/abs/1906.08101" ]

[ { "id": 1834081910, "node_id": "MDU6TGFiZWwxODM0MDgxOTEw", "url": "https://api.github.com/repos/huggingface/transformers/labels/Usage", "name": "Usage", "color": "e28436", "default": false, "description": "General questions about the library" } ]

[]

[ "Example\r\nMy sentence: Мой кот ... (My cat ...)\r\n3 word is ест (eat)\r\nList of possible words: ест (eat), поглощает (absorb), глотает (swallow), кушают (eats) etc.\r\nI need to determine the probabilities of each word from the given list in the context of the phrase and make the most correct sentence.\r\nOutput: Мой кот ест (My cat eat).", "This is a very general question. Please use [Stack Overflow](https://stackoverflow.com/) for this.\r\n\r\nNote that you'll need to use a model that is pretrained on Russian." ]

[ { "id": 1834056761, "node_id": "MDU6TGFiZWwxODM0MDU2NzYx", "url": "https://api.github.com/repos/huggingface/transformers/labels/Core:%20Modeling", "name": "Core: Modeling", "color": "FF8446", "default": false, "description": "Internals of the library; Models." } ]

[]

[ "change \r\n`BertModel.form_pretrained`\r\nto\r\n`BertForPreTraining.from_pretrained`\r\nit seems to work", "Glad you could get it to work! Indeed, `BertForPreTraining` should be used to convert from official BERT models." ]

[]

[]

[ "Hi @alvinchangw, \r\n\r\nGPT2 always uses causal masking no matter what kind of attention_mask you give it. \r\nIt's easy to see when you print out the computed attentions for each layer (by setting `output_attentions=True`) => see for this also #2975. \r\n\r\nIn the code this is done in this line:\r\nhttps://github.com/huggingface/transformers/blob/298bed16a841fae3608d334441ccae4d9043611f/src/transformers/modeling_gpt2.py#L146\r\n\r\nI admit it is very cryptic and probably should have better naming. Essentially what happens here is the following:\r\n`self.bias` is defined as a lower triangular mask (see torch function [here](https://pytorch.org/docs/stable/torch.html?highlight=tril#torch.tril) ). according to the sequence length (params `nd` and `ns`), we derive `b`. `b` is then a lower triangular mask of shape sequence length x sequence length. Using this mask, we substract 10^4 from all values in `w` which should be masked, which sets their attention to 0.\r\n", "Hi @patrickvonplaten,\r\n\r\nThank you for pointing this out and for the detailed explanation! " ]

[ { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null }, { "id": 1834056635, "node_id": "MDU6TGFiZWwxODM0MDU2NjM1", "url": "https://api.github.com/repos/huggingface/transformers/labels/Core:%20Tokenization", "name": "Core: Tokenization", "color": "FF4446", "default": false, "description": "Internals of the library; Tokenization." } ]

[]

[ "Intuitively, I would say that this might not be a bug but a limitation of the size of the vocabulary. In the cased version, all data that the tokenizer is 'trained' on is cased, meaning that there are tokens in the vocabulary that only differs by case (e.g. `Be` and `be`). As a consequence, this may cause the vocabulary to be a lot bigger, leaving less room for other tokens. \r\n\r\nThat is just my intuition and not based on any research.\r\n\r\nYou can try out KoBERT, though: https://github.com/SKTBrain/KoBERT", "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": 1260952223, "node_id": "MDU6TGFiZWwxMjYwOTUyMjIz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Discussion", "name": "Discussion", "color": "22870e", "default": false, "description": "Discussion on a topic (keep it focused or open a new issue though)" }, { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "Hi, there's no reason ALBERT would be faster than BERT. They have the same number of layers, ALBERT just uses repeating layers (its `n` layers are a just a single one) instead of different layers (BERT's `n` layers are `n` different layers).", "Hi @LysandreJik !\r\nIf ALBERT is a lot smaller than BERT (in terms of parameters), wouldn't it be faster? Taking up less memory, allowing bigger batch sizes, etc. Isn't this \"scalability\" one of the advantages of ALBERT over BERT?\r\nThanks!", "If training speeds \"proportionality\" is similar to inferencing, then Section 4.3 and Table 2, page 7 of the latest Albert paper, [https://arxiv.org/pdf/1909.11942.pdf](https://arxiv.org/pdf/1909.11942.pdf) compares \"Speedup\" of BERT & ALBERT models. For example, Albert_xxlarge \"speed of data throughput\" is 0.3x of BERT_large the baseline, so only 30% the throughput.", "> If training speeds \"proportionality\" is similar to inferencing, then Section 4.3 and Table 2, page 7 of the latest Albert paper, https://arxiv.org/pdf/1909.11942.pdf compares \"Speedup\" of BERT & ALBERT models.\r\n\r\nInteresting. Section 4.3 also says:\r\n\r\n>ALBERT models have higher data throughput compared to their corresponding BERT models. If we\r\nuse BERT-large as the baseline, we observe that ALBERT-large is about 1.7 times faster in iterating\r\nthrough the data while ALBERT-xxlarge is about 3 times slower because of the larger structure.\r\n\r\nJudging from this Section and from Table 2, some versions of ALBERT are indeed faster than BERT.", "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": 1834067346, "node_id": "MDU6TGFiZWwxODM0MDY3MzQ2", "url": "https://api.github.com/repos/huggingface/transformers/labels/Documentation", "name": "Documentation", "color": "77cc3b", "default": false, "description": "" } ]

[]

[]

[ { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null }, { "id": 1834056635, "node_id": "MDU6TGFiZWwxODM0MDU2NjM1", "url": "https://api.github.com/repos/huggingface/transformers/labels/Core:%20Tokenization", "name": "Core: Tokenization", "color": "FF4446", "default": false, "description": "Internals of the library; Tokenization." } ]

[]

[ "Can you post a minimal verifiable example that we can just copy-and-paste to try? I guess you don't really try to add empty strings as tokens?", "Hi @BramVanroy, no ofc, probably i've pasted the wrong snippet of code. \r\nIf you try to expand the vocabulary of the tokenizer with the following code: \r\n\r\n```\r\ntokenizer = XLMRobertaTokenizer.from_pretrained('xlm-roberta-large', do_lower_case=True)\r\ntokenizer.add_tokens(['[A1]', '[/A1]', '[A2]', '[/A2]']) \r\n```\r\n\r\nthe size of tokenizer remains the same. \r\nObviously if you try the same identical code with Bert or DistilBert (the ones i'm testing) all works fine. \r\nAll seems connected with the second condition of if-else block: \r\n\r\n```\r\nif (\r\n token != self.unk_token\r\n and self.convert_tokens_to_ids(token) == self.convert_tokens_to_ids(self.unk_token)\r\n and token not in to_add_tokens\r\n):\r\n```\r\nThis condition seems returns the wrong id for self.unk_token.\r\nRemoving this condition let me add the new tokens and extend the tokenizers.\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": 1834056761, "node_id": "MDU6TGFiZWwxODM0MDU2NzYx", "url": "https://api.github.com/repos/huggingface/transformers/labels/Core:%20Modeling", "name": "Core: Modeling", "color": "FF8446", "default": false, "description": "Internals of the library; Models." } ]

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

[]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/3063?src=pr&el=h1) Report\n> Merging [#3063](https://codecov.io/gh/huggingface/transformers/pull/3063?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/eec5ec807135ae61fa2266f3c7ad947cc207abda?src=pr&el=desc) will **increase** coverage by `0.22%`.\n> The diff coverage is `90.1%`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/3063?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #3063 +/- ##\n==========================================\n+ Coverage 77.59% 77.82% +0.22% \n==========================================\n Files 98 98 \n Lines 16250 16422 +172 \n==========================================\n+ Hits 12610 12780 +170 \n- Misses 3640 3642 +2\n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/3063?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/\\_\\_init\\_\\_.py](https://codecov.io/gh/huggingface/transformers/pull/3063/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9fX2luaXRfXy5weQ==) | `98.91% <100%> (ø)` | :arrow_up: |\n| [src/transformers/modeling\\_ctrl.py](https://codecov.io/gh/huggingface/transformers/pull/3063/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19jdHJsLnB5) | `98.22% <100%> (-0.02%)` | :arrow_down: |\n| [src/transformers/modeling\\_xlnet.py](https://codecov.io/gh/huggingface/transformers/pull/3063/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ194bG5ldC5weQ==) | `75.77% <100%> (ø)` | :arrow_up: |\n| [src/transformers/modeling\\_tf\\_gpt2.py](https://codecov.io/gh/huggingface/transformers/pull/3063/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9ncHQyLnB5) | `96.14% <100%> (+1.47%)` | :arrow_up: |\n| [src/transformers/modeling\\_gpt2.py](https://codecov.io/gh/huggingface/transformers/pull/3063/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19ncHQyLnB5) | `86.07% <100%> (-0.09%)` | :arrow_down: |\n| [src/transformers/modeling\\_tf\\_xlnet.py](https://codecov.io/gh/huggingface/transformers/pull/3063/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl94bG5ldC5weQ==) | `88.98% <100%> (+0.6%)` | :arrow_up: |\n| [src/transformers/modeling\\_tf\\_ctrl.py](https://codecov.io/gh/huggingface/transformers/pull/3063/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9jdHJsLnB5) | `99.57% <100%> (+1.74%)` | :arrow_up: |\n| [src/transformers/modeling\\_transfo\\_xl.py](https://codecov.io/gh/huggingface/transformers/pull/3063/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190cmFuc2ZvX3hsLnB5) | `75.63% <100%> (ø)` | :arrow_up: |\n| [src/transformers/modeling\\_tf\\_transfo\\_xl.py](https://codecov.io/gh/huggingface/transformers/pull/3063/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl90cmFuc2ZvX3hsLnB5) | `91.13% <20%> (-1%)` | :arrow_down: |\n| [src/transformers/modeling\\_tf\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/3063/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl91dGlscy5weQ==) | `91.56% <89.85%> (-1.22%)` | :arrow_down: |\n| ... and [5 more](https://codecov.io/gh/huggingface/transformers/pull/3063/diff?src=pr&el=tree-more) | |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/3063?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/3063?src=pr&el=footer). Last update [eec5ec8...b996a97](https://codecov.io/gh/huggingface/transformers/pull/3063?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n", "> This is cool, good job on the TensorFlow implementation! Regarding the use of the `past` and `mems`, I don't think they're actually implemented in the models though?\r\n\r\nI think the tf models also have the `past` and `mems` functionality implemented (when looking into the tf modeling files the `past` and `mems` variables are used in the code.", "Good to merge for me! \r\nChanged all `<tf_tensor>.shape` to `shape_list(<tf_tensor>)` to make function compatible with both eagermode and no eagermode after discussion with @thomwolf and @jplu .\r\n\r\nWill add additional Integration tests (so far only `tf_gpt2`) for other LMHead models and add beam_search once completed in torch version." ]

[ { "id": 1260952223, "node_id": "MDU6TGFiZWwxMjYwOTUyMjIz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Discussion", "name": "Discussion", "color": "22870e", "default": false, "description": "Discussion on a topic (keep it focused or open a new issue though)" }, { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "I am curious: what kind of behaviour do you see when you freeze the whole base model, and only train the classifier? Also, you may want avoid the use of the `cls` variable name because it is a reserved keyword for the class. In general, the classifier is trained quite quickly (often in two or epochs or less), so you are right in saying that in relative terms the weights of the layers that you add matter very little compared to the rest of the model.", "When I freeze the base model, the overall learning pace is drastically slower - 5 epochs is only enough to reach fraction of the quality when base model not frozen.\r\n\r\nHistograms when base model is frozen:\r\n<img width=\"929\" alt=\"Screenshot 2020-03-02 at 13 56 24\" src=\"https://user-images.githubusercontent.com/6958772/75678405-dc09df00-5c8d-11ea-995e-853fbaa15b71.png\">\r\n", "@BramVanroy do you have any further insights on this?", "Something else that might cause this is that your layers are stuck in some local optimum, or that they are nullified by ReLU. What happens if you use, e.g., gelu instead of relu? But that can't explain everything (because the first linear layer also barely changes its weights. So I'm not sure. ", "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": 1834052129, "node_id": "MDU6TGFiZWwxODM0MDUyMTI5", "url": "https://api.github.com/repos/huggingface/transformers/labels/High-Level%20feature", "name": "High-Level feature", "color": "f7c9a3", "default": false, "description": "" } ]

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

[ "I like this idea. Perhaps it can be added as an example, or even as an argument to the generation script." ]

[ { "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/3059?src=pr&el=h1) Report\n> :exclamation: No coverage uploaded for pull request base (`master@271344f`). [Click here to learn what that means](https://docs.codecov.io/docs/error-reference#section-missing-base-commit).\n> The diff coverage is `86.95%`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/3059?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #3059 +/- ##\n=========================================\n Coverage ? 76.59% \n=========================================\n Files ? 98 \n Lines ? 16219 \n Branches ? 0 \n=========================================\n Hits ? 12423 \n Misses ? 3796 \n Partials ? 0\n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/3059?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/modeling\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/3059/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ191dGlscy5weQ==) | `93.08% <ø> (ø)` | |\n| [src/transformers/modeling\\_bart.py](https://codecov.io/gh/huggingface/transformers/pull/3059/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19iYXJ0LnB5) | `93.05% <86.95%> (ø)` | |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/3059?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/3059?src=pr&el=footer). Last update [271344f...6e13b56](https://codecov.io/gh/huggingface/transformers/pull/3059?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n", "> ## Differences with PretrainedModel.generate\r\n> these are all bc thats the way fairseq does it!\r\n> \r\n> * BeamHypotheses(early_stopping=True)\r\nI think we have that option as well\r\n> * assumptions about various token_ids being present\r\n> * force decoder to start with EOS, then predict BOS\r\nThat's weird, no?\r\n> * decoder only considers the most recently generated token bc everything else is cached.\r\n> * prevent predictions of various special tokens at inopportune moments (all the -inf stuff)\r\n> * force eos if you hit max length\r\nWe had this in our code before as well - I deleted it because I think unfinished sentences (sentences that were finished because they hit `max_length`) should not be ended with an EOS.\r\n> * max_length is about how many tokens you want to generate. Doesn't matter how many you have.\r\nThis makes sense since encoder-decoder models always start from 0 `input_ids` for the decoder model and only have `encoder_input_ids` where as the standard \"only-decoder\" models (GPT2) have `decoder_input_ids` and append their output to it\r\n\r\n" ]

[ { "id": 1834056635, "node_id": "MDU6TGFiZWwxODM0MDU2NjM1", "url": "https://api.github.com/repos/huggingface/transformers/labels/Core:%20Tokenization", "name": "Core: Tokenization", "color": "FF4446", "default": false, "description": "Internals of the library; Tokenization." } ]

[ { "login": "mfuntowicz", "id": 2241520, "node_id": "MDQ6VXNlcjIyNDE1MjA=", "avatar_url": "https://avatars.githubusercontent.com/u/2241520?v=4", "gravatar_id": "", "url": "https://api.github.com/users/mfuntowicz", "html_url": "https://github.com/mfuntowicz", "followers_url": "https://api.github.com/users/mfuntowicz/followers", "following_url": "https://api.github.com/users/mfuntowicz/following{/other_user}", "gists_url": "https://api.github.com/users/mfuntowicz/gists{/gist_id}", "starred_url": "https://api.github.com/users/mfuntowicz/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/mfuntowicz/subscriptions", "organizations_url": "https://api.github.com/users/mfuntowicz/orgs", "repos_url": "https://api.github.com/users/mfuntowicz/repos", "events_url": "https://api.github.com/users/mfuntowicz/events{/privacy}", "received_events_url": "https://api.github.com/users/mfuntowicz/received_events", "type": "User", "site_admin": false } ]

[ "@mfuntowicz, would it make sense for me to integrate our tokenizer tests into your code, so you can see these things immediately? I'd be happy to do so.", "Looks like this is a duplicate of #2917.", "On second reading, this is not the same issue as #2917, though they may be related.", "Roberta has another related issue:\r\n```\r\n>>> import transformers\r\n>>> t_fast = transformers.AutoTokenizer.from_pretrained(\"roberta-base\", use_fast=True, add_special_tokens=False)\r\n>>> sentence = \"I went to the zoo yesterday, but they had only one animal.\"\r\n>>> tokenized = t_fast.encode_plus(sentence, add_special_tokens=False, return_offsets_mapping=True)\r\n>>> for start, end in (t for t in tokenized['offset_mapping'] if t is not None):\r\n... print(repr(sentence[start:end]))\r\n'I'\r\n' went'\r\n' to'\r\n' the'\r\n' zoo'\r\n' yesterday'\r\n','\r\n' but'\r\n' they'\r\n' had'\r\n' only'\r\n' one'\r\n' animal'\r\n'.'\r\n```\r\n\r\nThere are two problems here. `add_special_tokens` is being ignored (#2919), but also, it adds those extra spaces at the front of the words.", "Hi @dirkgr, \r\n\r\nThanks for your report.\r\n\r\nRegarding the integration of your tests, it definitively a good idea, if you can put @LysandreJik and myself as reviewers of the PR, we'll have a look 👍.\r\n\r\nRegarding `add_special_tokens`, the behaviour on Rust side is quite different as it's a parameter that needs to be provided a construction time, whereas Python allows at tokenisation time. We should make it clearer in the doc.\r\n\r\n```python\r\n>>> t = transformers.BertTokenizerFast.from_pretrained('bert-base-cased')\r\n>>> tokenized = t.encode_plus(sentence, add_special_tokens=False, return_offsets_mapping=True)\r\n>>> tokenized['input_ids']\r\n[101, 138, 117, 22607, 103, 4522, 20734, 2101, 5650, 119, 102]\r\n\r\n>>> t = transformers.BertTokenizerFast.from_pretrained('bert-base-cased', add_special_tokens=False)\r\n>>> tokenized = t.encode_plus(sentence, add_special_tokens=False, return_offsets_mapping=True)\r\n>>> tokenized['input_ids']\r\n>>> [138, 117, 22607, 103, 4522, 20734, 2101, 5650, 119]\r\n```\r\n\r\nFor Roberta we're aware of this extra space being included, cc'ing @n1t0 for keeping track of this.", "Ok some more context, `GPT2TokenizerFast` has also the same behaviour, so I would extrapolate this occurs more generally on BPE model. \r\n\r\nAdding `<ModelFast>.from_pretrained(..., add_prefix_space=True)` doesn't append the space before but after the token:\r\n\r\n```python\r\n'I '\r\n'went '\r\n'to '\r\n'the '\r\n'zoo '\r\n'yesterday,'\r\n' '\r\n'but '\r\n'they '\r\n'had '\r\n'only '\r\n'one '\r\n'animal.'\r\n```", "Hi @dirkgr!\r\n\r\nSo there are multiple different things here:\r\n - Your first example is due to the space between `naïve` and `[MASK]` not being trimmed out. We are aware of this behavior and are currently working on a fix.\r\n - The fact that `add_prefix_space=True` moves the space at the end is actually a bug too. This happens because we mess with the offsets while adding the prefix. I am working on a fix for this too.\r\n - Now, the second example you gave is actually expected behavior:\r\n```python\r\nimport transformers\r\nt_fast = transformers.AutoTokenizer.from_pretrained(\"roberta-base\", use_fast=True, add_special_tokens=False)\r\nsentence = \"I went to the zoo yesterday, but they had only one animal.\"\r\ntokenized = t_fast.encode_plus(sentence, return_offsets_mapping=True)\r\noffsets = tokenized['offset_mapping']\r\ntokens = t_fast.tokenize(sentence)\r\nfor token, (start, end) in (t for t in zip(tokens, offsets) if t[1] is not None):\r\n print(repr(token))\r\n print(repr(sentence[start:end]))\r\n print(repr(t_fast.decode(t_fast.convert_tokens_to_ids([token]))))\r\n```\r\nwill give the following output:\r\n```\r\n'I'\r\n'I'\r\n'I'\r\n'Ġwent'\r\n' went'\r\n' went'\r\n'Ġto'\r\n' to'\r\n' to'\r\n'Ġthe'\r\n' the'\r\n' the'\r\n'Ġzoo'\r\n' zoo'\r\n' zoo'\r\n'Ġyesterday'\r\n' yesterday'\r\n' yesterday'\r\n','\r\n','\r\n','\r\n'Ġbut'\r\n' but'\r\n' but'\r\n'Ġthey'\r\n' they'\r\n' they'\r\n'Ġhad'\r\n' had'\r\n' had'\r\n'Ġonly'\r\n' only'\r\n' only'\r\n'Ġone'\r\n' one'\r\n' one'\r\n'Ġanimal'\r\n' animal'\r\n' animal'\r\n'.'\r\n'.'\r\n'.'\r\n```\r\nHere you can see that the space is actually part of these tokens. That's just the way the byte-level BPE used by GPT-2 and Roberta works. The `Ġ` is actually an encoded space. Does it make sense?", "We hit a similar issue when we add new tokens. The input ids are correct, but offsets after the new token are off.\r\n\r\nExample: https://colab.research.google.com/drive/1e2a3iyLF9NSMWZR50pnDYRhpizcsRV_6\r\n\r\n```\r\ntext = \"A [test] C\"\r\n\r\nprint(tokenizer.encode(text, add_special_tokens=True))\r\n\r\nresults = tokenizer.encode_plus(text, \r\n return_offsets_mapping=True,\r\n pad_to_max_length=False,\r\n max_length=128,\r\n return_overflowing_tokens=False,\r\n add_special_tokens=True)\r\n\r\nfor se in results['offset_mapping']:\r\n if se:\r\n print(text[se[0]:se[1]], se)\r\n```\r\n\r\n```\r\n[101, 1037, 30522, 1039, 102]\r\nA (0, 1)\r\n [test (1, 7)\r\n (8, 9)\r\n```\r\n\r\nPotentially related issue huggingface/tokenizers#143\r\n", "I think all of the mentioned bugs on this issue should now be fixed on `master`", "```\r\n>>> import transformers\r\n>>> t_fast = transformers.AutoTokenizer.from_pretrained(\"bert-base-cased\", use_fast=True, add_special_tokens=False)\r\n>>> sentence = \"A, naïve [MASK] AllenNLP sentence.\"\r\n>>> tokenized = t_fast.encode_plus(sentence, add_special_tokens=False, return_offsets_mapping=True)\r\n>>> for start, end in tokenized['offset_mapping']:\r\n... print(repr(sentence[start:end]))\r\n'A'\r\n','\r\n'naïve'\r\n'[MASK]'\r\n'Allen'\r\n'NL'\r\n'P'\r\n'sentence'\r\n'.'\r\n```\r\nand\r\n```\r\n>>> import transformers\r\n>>> t_fast = transformers.AutoTokenizer.from_pretrained(\"roberta-base\", use_fast=True, add_special_tokens=False)\r\n>>> sentence = \"I went to the zoo yesterday, but they had only one animal.\"\r\n>>> tokenized = t_fast.encode_plus(sentence, add_special_tokens=False, return_offsets_mapping=True)\r\n>>> for start, end in (t for t in tokenized['offset_mapping'] if t is not None):\r\n... print(repr(sentence[start:end]))\r\n'I'\r\n'went'\r\n'to'\r\n'the'\r\n'zoo'\r\n'yesterday'\r\n','\r\n'but'\r\n'they'\r\n'had'\r\n'only'\r\n'one'\r\n'animal'\r\n'.' \r\n```\r\nand the last one:\r\n```\r\nfrom transformers import BertTokenizerFast\r\ntokenizer = BertTokenizerFast.from_pretrained('bert-base-uncased')\r\ntokenizer.add_tokens(['[test]'])\r\ntext = \"A [test] C\"\r\n\r\nprint(tokenizer.encode(text, add_special_tokens=True))\r\n\r\nresults = tokenizer.encode_plus(text, \r\n return_offsets_mapping=True,\r\n pad_to_max_length=False,\r\n max_length=128,\r\n return_overflowing_tokens=False,\r\n add_special_tokens=True)\r\n\r\nfor se in results['offset_mapping']:\r\n if se:\r\n print(text[se[0]:se[1]], se)\r\n```\r\ngives\r\n```\r\n[101, 1037, 30522, 1039, 102]\r\n (0, 0)\r\nA (0, 1)\r\n[test] (2, 8)\r\nC (9, 10)\r\n (0, 0)\r\n```" ]

[ { "id": 1834056635, "node_id": "MDU6TGFiZWwxODM0MDU2NjM1", "url": "https://api.github.com/repos/huggingface/transformers/labels/Core:%20Tokenization", "name": "Core: Tokenization", "color": "FF4446", "default": false, "description": "Internals of the library; Tokenization." } ]

[ { "login": "mfuntowicz", "id": 2241520, "node_id": "MDQ6VXNlcjIyNDE1MjA=", "avatar_url": "https://avatars.githubusercontent.com/u/2241520?v=4", "gravatar_id": "", "url": "https://api.github.com/users/mfuntowicz", "html_url": "https://github.com/mfuntowicz", "followers_url": "https://api.github.com/users/mfuntowicz/followers", "following_url": "https://api.github.com/users/mfuntowicz/following{/other_user}", "gists_url": "https://api.github.com/users/mfuntowicz/gists{/gist_id}", "starred_url": "https://api.github.com/users/mfuntowicz/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/mfuntowicz/subscriptions", "organizations_url": "https://api.github.com/users/mfuntowicz/orgs", "repos_url": "https://api.github.com/users/mfuntowicz/repos", "events_url": "https://api.github.com/users/mfuntowicz/events{/privacy}", "received_events_url": "https://api.github.com/users/mfuntowicz/received_events", "type": "User", "site_admin": false } ]

[ "I think it's a duplicate of #2919. Btw, I should have told you when I saw you open the PR, [I opened a bunch of issues related to AllenNLP usage](https://github.com/huggingface/transformers/issues?utf8=%E2%9C%93&q=is%3Aissue+author%3Abryant1410). I think one that was closed it's not completely solved, but not sure.", "It's not the same issue. This one is about special tokens in text form in the middle of your string. #2919 is about `[CLS]` and `[SEP]` being added to the beginning and end. Also, #2919 has already been fixed.", "Ohh, :+1: ", "This is now fixed on `master`:\r\n\r\n```\r\n>>> import transformers\r\n>>> t_fast = transformers.AutoTokenizer.from_pretrained(\"roberta-base\", use_fast=True)\r\n>>> t_fast.encode_plus(\"A <mask> sentence.\")\r\n{'input_ids': [0, 83, 50264, 3645, 4, 2], 'attention_mask': [1, 1, 1, 1, 1, 1]}\r\n>>> t_fast.convert_ids_to_tokens([0, 83, 50264, 3645, 4, 2])\r\n['<s>', 'ĠA', '<mask>', 'Ġsentence', '.', '</s>']\r\n```" ]

[]

[]

[]

[]

[]

[ "Hmm, why do we want to be able to spawn a `NerPipeline` or a `QuestionAnsweringPipeline` without a model or a tokenizer?\r\n\r\nIsn't this what `pipeline(\"ner\")` or `pipeline(\"question-answering\")` is for?", "I think the question is rather why can't we spawn a `NerPipeline` or a `QuestionAnsweringPipeline` even though we have defined defaults for them?\r\n\r\nWhat I see in the `pipeline` factory is the ability to simply specify a task and get the appropriate pipeline. I don't see a strong reason for the task-specific pipelines to not be able to load a default, but I may be missing part of the picture.\r\n\r\nIf you think this is unnecessary I can revert the code - we'll just need to make sure that the doc explains what the `pipeline` factory is for and how it handles defaults compared to task-specific pipelines, because I was misled.", "I'll let @mfuntowicz and @thomwolf chime in, but for me, the subclasses of Pipeline are the actual implementations – preferably well-typed – that do not expose too much magic. \r\n\r\nI don't see the point of having two public APIs that do exactly the same thing.\r\n\r\nE.g., the logic behind [get_defaults](https://github.com/huggingface/transformers/pull/3055/files#diff-1e87b75d7b313550a38be1daecd653f7R485-R504) is a duplication of what's already in `pipeline()`\r\n\r\nIn any cases, the subclasses of Pipeline dont accept a model/tokenizer of type `str`, in contradiction to the doc (it crashes), because the spawning of model/tokenizer with `from_pretrained()` is only inside the `pipeline` wrapper ", "Ok, this makes sense. I'll revert that later today. Thanks @julien-c" ]

[]

[]

[ "I believe the `bool` operator was introduced in PyTorch 1.2.0, won't this break compatibility with PyTorch 1.0.0?", "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/3054?src=pr&el=h1) Report\n> Merging [#3054](https://codecov.io/gh/huggingface/transformers/pull/3054?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/8bcb37bfb80d77e06001f989ad982c9961a69c31?src=pr&el=desc) will **decrease** coverage by `0.01%`.\n> The diff coverage is `100%`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/3054?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #3054 +/- ##\n==========================================\n- Coverage 77.2% 77.18% -0.02% \n==========================================\n Files 98 98 \n Lines 16063 16063 \n==========================================\n- Hits 12401 12399 -2 \n- Misses 3662 3664 +2\n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/3054?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/modeling\\_bart.py](https://codecov.io/gh/huggingface/transformers/pull/3054/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19iYXJ0LnB5) | `84.58% <100%> (ø)` | :arrow_up: |\n| [src/transformers/modeling\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/3054/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ191dGlscy5weQ==) | `92.22% <0%> (-0.33%)` | :arrow_down: |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/3054?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/3054?src=pr&el=footer). Last update [8bcb37b...2116c2b](https://codecov.io/gh/huggingface/transformers/pull/3054?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/3053?src=pr&el=h1) Report\n> Merging [#3053](https://codecov.io/gh/huggingface/transformers/pull/3053?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/8bcb37bfb80d77e06001f989ad982c9961a69c31?src=pr&el=desc) will **decrease** coverage by `1.03%`.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/3053?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #3053 +/- ##\n==========================================\n- Coverage 77.2% 76.16% -1.04% \n==========================================\n Files 98 98 \n Lines 16063 16063 \n==========================================\n- Hits 12401 12234 -167 \n- Misses 3662 3829 +167\n```\n\n\n| [Impacted Files](https://codecov.io/gh/huggingface/transformers/pull/3053?src=pr&el=tree) | Coverage Δ | |\n|---|---|---|\n| [src/transformers/modeling\\_tf\\_pytorch\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/3053/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ190Zl9weXRvcmNoX3V0aWxzLnB5) | `8.72% <0%> (-81.21%)` | :arrow_down: |\n| [src/transformers/modeling\\_roberta.py](https://codecov.io/gh/huggingface/transformers/pull/3053/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19yb2JlcnRhLnB5) | `85.71% <0%> (-10%)` | :arrow_down: |\n| [src/transformers/modeling\\_xlnet.py](https://codecov.io/gh/huggingface/transformers/pull/3053/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ194bG5ldC5weQ==) | `73.48% <0%> (-2.3%)` | :arrow_down: |\n| [src/transformers/modeling\\_ctrl.py](https://codecov.io/gh/huggingface/transformers/pull/3053/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19jdHJsLnB5) | `96.03% <0%> (-2.21%)` | :arrow_down: |\n| [src/transformers/modeling\\_openai.py](https://codecov.io/gh/huggingface/transformers/pull/3053/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ19vcGVuYWkucHk=) | `80.2% <0%> (-1.35%)` | :arrow_down: |\n| [src/transformers/modeling\\_utils.py](https://codecov.io/gh/huggingface/transformers/pull/3053/diff?src=pr&el=tree#diff-c3JjL3RyYW5zZm9ybWVycy9tb2RlbGluZ191dGlscy5weQ==) | `92.22% <0%> (-0.33%)` | :arrow_down: |\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/3053?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/3053?src=pr&el=footer). Last update [8bcb37b...9dc0964](https://codecov.io/gh/huggingface/transformers/pull/3053?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n", "@LysandreJik This doesn't touch the main code, so should be fine to merge. " ]

[]

[]

[ "I read the code carefully, and it do share." ]

[]

[]

[ "# [Codecov](https://codecov.io/gh/huggingface/transformers/pull/3051?src=pr&el=h1) Report\n> Merging [#3051](https://codecov.io/gh/huggingface/transformers/pull/3051?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/transformers/commit/53ce3854a16ad2a715bc6ac8af3e30c18b5a1d11?src=pr&el=desc) will **not change** coverage.\n> The diff coverage is `n/a`.\n\n[](https://codecov.io/gh/huggingface/transformers/pull/3051?src=pr&el=tree)\n\n```diff\n@@ Coverage Diff @@\n## master #3051 +/- ##\n======================================\n Coverage 76.1% 76.1% \n======================================\n Files 98 98 \n Lines 15946 15946 \n======================================\n Hits 12136 12136 \n Misses 3810 3810\n```\n\n\n\n------\n\n[Continue to review full report at Codecov](https://codecov.io/gh/huggingface/transformers/pull/3051?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/3051?src=pr&el=footer). Last update [53ce385...ff701d9](https://codecov.io/gh/huggingface/transformers/pull/3051?src=pr&el=lastupdated). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments).\n", "On the Docker files, do we really need conda? (I try to stay away from conda as much as possible)", "Well, conda brings Intel MKL with just `conda install mkl, mkl-devel` which improves PyTorch and TF by a significant factor. \r\n\r\nDepending on what level of performances we want to provide: \r\n\r\n- I totally remove MKL and install some Open BLAS/LAPACK libraries\r\n- I'll build MKL in the images and include into PATH", "MKL is also available on PyPi. Have a looks-y [here](https://software.intel.com/en-us/articles/installing-the-intel-distribution-for-python-and-intel-performance-libraries-with-pip-and) to check whether everything you need is there. This might also be of interest: https://software.intel.com/en-us/distribution-for-python/choose-download/linux\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 } ]

[]

[ "Any progress on this? Has anyone found a way to disable the logging this?\r\n\r\n\r\nThe issue appears to be tqdm. A work-around is to disable it before importing transformers:\r\n```\r\nimport tqdm\r\n\r\ndef nop(it, *a, **k):\r\n return it\r\ntqdm.tqdm = nop\r\nimport transformers\r\n```", "I agree that it's a valid requirement, we'll look into it", "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", "Wish this would stay open.", "Logger could be really annoying when it comes to applications. Should have some way to turn it off.", "Here you go:\r\n\r\n```\r\n# To control logging level for various modules used in the application:\r\nimport logging\r\nimport re\r\ndef set_global_logging_level(level=logging.ERROR, prefices=[\"\"]):\r\n \"\"\"\r\n Override logging levels of different modules based on their name as a prefix.\r\n It needs to be invoked after the modules have been loaded so that their loggers have been initialized.\r\n\r\n Args:\r\n - level: desired level. e.g. logging.INFO. Optional. Default is logging.ERROR\r\n - prefices: list of one or more str prefices to match (e.g. [\"transformers\", \"torch\"]). Optional.\r\n Default is `[\"\"]` to match all active loggers.\r\n The match is a case-sensitive `module_name.startswith(prefix)`\r\n \"\"\"\r\n prefix_re = re.compile(fr'^(?:{ \"|\".join(prefices) })')\r\n for name in logging.root.manager.loggerDict:\r\n if re.match(prefix_re, name):\r\n logging.getLogger(name).setLevel(level)\r\n```\r\nUsage:\r\n\r\n1. override all module-specific loggers to a desired level (except whatever got logged during modules importing)\r\n```\r\nimport everything, you, need\r\nimport logging\r\nset_global_logging_level(logging.ERROR)\r\n```\r\n\r\n2. In case of transformers you most likely need to call it as:\r\n```\r\nimport transformers, torch, ...\r\nimport logging\r\nset_global_logging_level(logging.ERROR, [\"transformers\", \"nlp\", \"torch\", \"tensorflow\", \"tensorboard\", \"wandb\"])\r\n```\r\nadd/remove modules as needed.\r\n\r\n\r\nTo disable logging globally - place at the beginning of the script\r\n```\r\nimport logging\r\nlogging.disable(logging.INFO) # disable INFO and DEBUG logging everywhere\r\n# or \r\n# logging.disable(logging.WARNING) # disable WARNING, INFO and DEBUG logging everywhere\r\n```\r\n\r\nIf desired, `set_global_logging_level` could be expanded to be a scope manager too.", "Will that kill tqdm? I want to keep tqdm!", "> Will that kill tqdm? I want to keep tqdm!\r\n\r\n```\r\nset_global_logging_level(logging.ERROR, [\"transformers\", \"nlp\", \"torch\", \"tensorflow\", \"tensorboard\", \"wandb\"])\r\n\r\nfrom tqdm import tqdm\r\nfor i in tqdm(range(10000)): x = i**i\r\n```\r\nworks just fine\r\n\r\nand so does disable all:\r\n```\r\nset_global_logging_level(logging.ERROR])\r\n\r\nfrom tqdm import tqdm\r\nfor i in tqdm(range(10000)): x = i**i\r\n```\r\n\r\nor in the case of \"total logging silence\" setting:\r\n```\r\nimport logging\r\nlogging.disable(logging.INFO) # disable INFO and DEBUG logging everywhere\r\n\r\nfrom tqdm import tqdm\r\nfor i in tqdm(range(10000)): x = i**i\r\n```\r\n\r\nworks too. \r\n\r\nI don't think it uses `logging`.\r\n", "PR with the proposed code, plus adding the ability to do that during `pytest` https://github.com/huggingface/transformers/pull/6816", "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 resolved by https://github.com/huggingface/transformers/pull/6434" ]

[ { "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 } ]

[]

[ "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" ]