url stringlengths 62 66 | repository_url stringclasses 1
value | labels_url stringlengths 76 80 | comments_url stringlengths 71 75 | events_url stringlengths 69 73 | html_url stringlengths 50 56 | id int64 377M 2.15B | node_id stringlengths 18 32 | number int64 1 29.2k | title stringlengths 1 487 | user dict | labels list | state stringclasses 2
values | locked bool 2
classes | assignee dict | assignees list | comments list | created_at int64 1.54k 1.71k | updated_at int64 1.54k 1.71k | closed_at int64 1.54k 1.71k ⌀ | author_association stringclasses 4
values | active_lock_reason stringclasses 2
values | body stringlengths 0 234k ⌀ | reactions dict | timeline_url stringlengths 71 75 | state_reason stringclasses 3
values | draft bool 2
classes | pull_request dict |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
https://api.github.com/repos/huggingface/transformers/issues/909 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/909/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/909/comments | https://api.github.com/repos/huggingface/transformers/issues/909/events | https://github.com/huggingface/transformers/pull/909 | 473,448,735 | MDExOlB1bGxSZXF1ZXN0MzAxNjEzOTM4 | 909 | [develop] Convenience args.{train/dev}_file arguments. | {
"login": "kshitij12345",
"id": 19503980,
"node_id": "MDQ6VXNlcjE5NTAzOTgw",
"avatar_url": "https://avatars.githubusercontent.com/u/19503980?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/kshitij12345",
"html_url": "https://github.com/kshitij12345",
"followers_url": "https://api.github.c... | [] | closed | false | null | [] | [
"Why don't give the full path of the train/dev file instead of giving data_dir?",
"My thought was not the change much of the argument interface. If we support full path, then the `data_dir` will not be required and considered. So I wasn't sure if that change is the way to go. Sure, we can change it like that as... | 1,564 | 1,568 | 1,568 | CONTRIBUTOR | null | Adds Arguments
```
--train_file any_train_file.tsv \
--dev_file any_dev_file.tsv \
```
to use any file for training/dev in the pointed data directory.
Especially handy for evaluation.
Allows for
```
python run_glue.py \
--model_type bert \
--model_name_or_path bert-base-uncased \
--t... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/909/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/909/timeline | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/909",
"html_url": "https://github.com/huggingface/transformers/pull/909",
"diff_url": "https://github.com/huggingface/transformers/pull/909.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/909.patch",
"merged_at": null
... |
https://api.github.com/repos/huggingface/transformers/issues/908 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/908/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/908/comments | https://api.github.com/repos/huggingface/transformers/issues/908/events | https://github.com/huggingface/transformers/issues/908 | 473,442,900 | MDU6SXNzdWU0NzM0NDI5MDA= | 908 | Cannot inherit from BertPretrainedModel anymore after migrating to pytorch-transformers | {
"login": "ereday",
"id": 13196191,
"node_id": "MDQ6VXNlcjEzMTk2MTkx",
"avatar_url": "https://avatars.githubusercontent.com/u/13196191?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ereday",
"html_url": "https://github.com/ereday",
"followers_url": "https://api.github.com/users/ereday/fo... | [] | closed | false | null | [] | [
"You should do `from pytorch_transformers.modeling_bert import BertPreTrainedModel`\r\n\r\nI'll add these to the main `__init__.py`",
"Thank you for the answer @thomwolf . It solved that error but now I'm getting another one (which wasn't there when I was using previous versions of the repository): `TypeError: un... | 1,564 | 1,566 | 1,564 | NONE | null | Hi,
After I updated my environment today, I cannot run my old code anymore. I think I followed all the steps in migration section of README but still the following code gives me the `NameError: name 'BertPreTrainedModel' is not defined` error. To migrate latest version, I cloned the repository and run `pip install ... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/908/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/908/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/907 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/907/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/907/comments | https://api.github.com/repos/huggingface/transformers/issues/907/events | https://github.com/huggingface/transformers/pull/907 | 473,361,809 | MDExOlB1bGxSZXF1ZXN0MzAxNTQzNTcy | 907 | Fix convert to tf | {
"login": "dhpollack",
"id": 368699,
"node_id": "MDQ6VXNlcjM2ODY5OQ==",
"avatar_url": "https://avatars.githubusercontent.com/u/368699?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/dhpollack",
"html_url": "https://github.com/dhpollack",
"followers_url": "https://api.github.com/users/dhpo... | [] | closed | false | null | [] | [
"# [Codecov](https://codecov.io/gh/huggingface/pytorch-transformers/pull/907?src=pr&el=h1) Report\n> Merging [#907](https://codecov.io/gh/huggingface/pytorch-transformers/pull/907?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/pytorch-transformers/commit/46cc9dd2b51a152b2e262ec12e40dddd13235aba?src... | 1,564 | 1,565 | 1,565 | CONTRIBUTOR | null | I struggled with this same problem for a long time. The naive `assign` op way, puts all of the weights into both the checkpoint file (`.ckpt.data-XXXXX-of-YYYYY`) and the meta file (`.ckpt.meta`). This is because assign adds an operation to the graph. So basically, you have two instructions in your meta file, one th... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/907/reactions",
"total_count": 2,
"+1": 1,
"-1": 0,
"laugh": 0,
"hooray": 1,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/907/timeline | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/907",
"html_url": "https://github.com/huggingface/transformers/pull/907",
"diff_url": "https://github.com/huggingface/transformers/pull/907.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/907.patch",
"merged_at": 15650... |
https://api.github.com/repos/huggingface/transformers/issues/906 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/906/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/906/comments | https://api.github.com/repos/huggingface/transformers/issues/906/events | https://github.com/huggingface/transformers/issues/906 | 473,233,676 | MDU6SXNzdWU0NzMyMzM2NzY= | 906 | cuda out of memory | {
"login": "Ravikiran2611",
"id": 40524495,
"node_id": "MDQ6VXNlcjQwNTI0NDk1",
"avatar_url": "https://avatars.githubusercontent.com/u/40524495?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Ravikiran2611",
"html_url": "https://github.com/Ravikiran2611",
"followers_url": "https://api.githu... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"Try to implement gradient accumulation during training, instead of updating parameters in each iteration. Please check this nice and easy-to-follow tutorial by @thomwolf [here](https://medium.com/huggingface/training-larger-batches-practical-tips-on-1-gpu-multi-gpu-distributed-setups-ec88c3e51255) . I used this te... | 1,564 | 1,598 | 1,581 | NONE | null | `import torch
from pytorch_transformers import BertTokenizer, BertForSequenceClassification
tokenizer = BertTokenizer.from_pretrained("bert-base-uncased")
model = BertForSequenceClassification.from_pretrained("bert-base-uncased")
import csv
data = []
label = []
with open('Training.csv','r') as file:
reader ... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/906/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/906/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/905 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/905/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/905/comments | https://api.github.com/repos/huggingface/transformers/issues/905/events | https://github.com/huggingface/transformers/pull/905 | 473,064,372 | MDExOlB1bGxSZXF1ZXN0MzAxMzA1NjEw | 905 | Bugfix for encoding error during GPT2Tokenizer.from_pretrained('local… | {
"login": "DrStoop",
"id": 19177740,
"node_id": "MDQ6VXNlcjE5MTc3NzQw",
"avatar_url": "https://avatars.githubusercontent.com/u/19177740?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/DrStoop",
"html_url": "https://github.com/DrStoop",
"followers_url": "https://api.github.com/users/DrStoo... | [] | closed | false | null | [] | [
"# [Codecov](https://codecov.io/gh/huggingface/pytorch-transformers/pull/905?src=pr&el=h1) Report\n> Merging [#905](https://codecov.io/gh/huggingface/pytorch-transformers/pull/905?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/pytorch-transformers/commit/46cc9dd2b51a152b2e262ec12e40dddd13235aba?src... | 1,564 | 1,565 | 1,565 | NONE | null | …/path/to/mode')
BUG DESCRIPTION: Loading GPT2-tokenizer from local path with
GPT2Tokenizer.from_pretrained(pretrained_model_name_or_path='local/path/to/model')
returns following error due to encoding error for json.load():
Traceback (most recent call last):
File "/opt/pycharm-2019.1.3/helpers/pydev/pydevd.py"... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/905/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/905/timeline | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/905",
"html_url": "https://github.com/huggingface/transformers/pull/905",
"diff_url": "https://github.com/huggingface/transformers/pull/905.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/905.patch",
"merged_at": null
... |
https://api.github.com/repos/huggingface/transformers/issues/904 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/904/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/904/comments | https://api.github.com/repos/huggingface/transformers/issues/904/events | https://github.com/huggingface/transformers/issues/904 | 473,037,070 | MDU6SXNzdWU0NzMwMzcwNzA= | 904 | AssertionError while using DataParallelModel | {
"login": "sajidrahman",
"id": 4258481,
"node_id": "MDQ6VXNlcjQyNTg0ODE=",
"avatar_url": "https://avatars.githubusercontent.com/u/4258481?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/sajidrahman",
"html_url": "https://github.com/sajidrahman",
"followers_url": "https://api.github.com/us... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"You don't need to use this method here because the models have built-in losses computation.\r\nJust feed the labels and you will get the loss back (see the doc/docstrings of the models).",
"Hi @thomwolf, thanks for the suggestion. After following your advice, I'm not getting the error anymore, but now I'm a bit ... | 1,564 | 1,617 | 1,570 | NONE | null | Hi,
I'm trying to use _Load Balancing during multi-GPU_ environment. I'm following the tutorial by @thomwolf published at [medium](https://medium.com/huggingface/training-larger-batches-practical-tips-on-1-gpu-multi-gpu-distributed-setups-ec88c3e51255). I'm fine-tuning GPT-2 small for a classification task. Here're ... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/904/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/904/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/903 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/903/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/903/comments | https://api.github.com/repos/huggingface/transformers/issues/903/events | https://github.com/huggingface/transformers/issues/903 | 472,817,449 | MDU6SXNzdWU0NzI4MTc0NDk= | 903 | why the acc of chinese model(bert) is just 0.438 | {
"login": "zsk423200",
"id": 18025765,
"node_id": "MDQ6VXNlcjE4MDI1NzY1",
"avatar_url": "https://avatars.githubusercontent.com/u/18025765?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/zsk423200",
"html_url": "https://github.com/zsk423200",
"followers_url": "https://api.github.com/users/... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"i have the same issue as you. Do you have a good solution for it？",
"I met the same problem in multi-labels classification task, have no idea about the problem!",
"@zsk423200 Maybe you can try the \"bert-base-multilingual-cased-pytorch_model\" , it's performance seems better a lot than the pure Chinese ver ... | 1,564 | 1,571 | 1,571 | NONE | null | dataset: XNLI-1.0
i run the dataset of xnli-1.0, but the result is `acc = 0.43855421686746987`, and i run on google bert in tf, the result is `eval_accuracy = 0.7674699`. i use the same epochs and lr, i really don't know why.
i add the dataprocess of xnli ,the same with the version of tf bert:
```
class XnliProce... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/903/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/903/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/902 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/902/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/902/comments | https://api.github.com/repos/huggingface/transformers/issues/902/events | https://github.com/huggingface/transformers/issues/902 | 472,804,028 | MDU6SXNzdWU0NzI4MDQwMjg= | 902 | Torchscript Trace slower with C++ runtime environment. | {
"login": "sukuya",
"id": 4861350,
"node_id": "MDQ6VXNlcjQ4NjEzNTA=",
"avatar_url": "https://avatars.githubusercontent.com/u/4861350?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/sukuya",
"html_url": "https://github.com/sukuya",
"followers_url": "https://api.github.com/users/sukuya/foll... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"2 possible reasons:\r\n1. the first time you run `forward` will do some preheating work, maybe you should exclude the first run.\r\n2. try exclude `toTuple`\r\n\r\nAccording to my experience, jit with python or c++ will cost almost the same time.",
"@Meteorix Forward is called once before the loop, are you talki... | 1,564 | 1,571 | 1,571 | CONTRIBUTOR | null | I traced the BERT model from PyTorchTransformers library and getting the following results for 10 iterations.
a) Using Python runtime for running the forward: 979,292 µs
```
import time
model = torch.jit.load('models_backup/2_2.pt')
x = torch.randint(2000, (1, 14), dtype=torch.long, device='cpu')
start = time.t... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/902/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/902/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/901 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/901/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/901/comments | https://api.github.com/repos/huggingface/transformers/issues/901/events | https://github.com/huggingface/transformers/issues/901 | 472,768,061 | MDU6SXNzdWU0NzI3NjgwNjE= | 901 | bug: it is broken to use tokenizer path | {
"login": "zsk423200",
"id": 18025765,
"node_id": "MDQ6VXNlcjE4MDI1NzY1",
"avatar_url": "https://avatars.githubusercontent.com/u/18025765?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/zsk423200",
"html_url": "https://github.com/zsk423200",
"followers_url": "https://api.github.com/users/... | [] | closed | false | null | [] | [
"Had the same issue when passing the exact path of the vocabulary file. Fixed it by just passing the name of the directory that contains the vocabulary file (in my case it was `vocab.txt`).",
"Good catch.\r\n\r\nFor non-BPE models with a single vocabulary file (Bert, XLNet, Transformer-XL) we can fix this workflo... | 1,564 | 1,564 | 1,564 | NONE | null | run run_glue.py with the parameter of tokenizer_name:
`--tokenizer_name=/path/bert-base-chinese-vocab.txt`
but get the error:
```
Traceback (most recent call last):
File "run_glue.py", line 485, in <module>
main()
File "run_glue.py", line 418, in main
tokenizer = tokenizer_class.from_pretrained(args... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/901/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/901/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/900 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/900/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/900/comments | https://api.github.com/repos/huggingface/transformers/issues/900/events | https://github.com/huggingface/transformers/issues/900 | 472,748,401 | MDU6SXNzdWU0NzI3NDg0MDE= | 900 | SpanBERT support | {
"login": "stefan-it",
"id": 20651387,
"node_id": "MDQ6VXNlcjIwNjUxMzg3",
"avatar_url": "https://avatars.githubusercontent.com/u/20651387?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/stefan-it",
"html_url": "https://github.com/stefan-it",
"followers_url": "https://api.github.com/users/... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"are we going to get this? :) thanks :)",
"Fyi https://github.com/mandarjoshi90/coref#pretrained-coreference-models describes how to obtain the coreference models that should contain SpanBERT.\r\n",
"@ArneBinder Thanks for that hint!\r\n\r\nI downloaded the *SpanBERT* (base) model. Unfortunately, the TF checkpo... | 1,564 | 1,574 | 1,574 | COLLABORATOR | null | Hi,
I think the new *SpanBERT* model should also be supported in `pytorch-transformers` 😅
> We present SpanBERT, a pre-training method that is designed to better represent and predict spans of text.
Paper can be found [here](https://arxiv.org/abs/1907.10529).
Model is currently not released yet, I'll updat... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/900/reactions",
"total_count": 14,
"+1": 14,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/900/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/899 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/899/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/899/comments | https://api.github.com/repos/huggingface/transformers/issues/899/events | https://github.com/huggingface/transformers/pull/899 | 472,745,736 | MDExOlB1bGxSZXF1ZXN0MzAxMDQ1OTQ2 | 899 | Fixed import to use torchscript flag. | {
"login": "sukuya",
"id": 4861350,
"node_id": "MDQ6VXNlcjQ4NjEzNTA=",
"avatar_url": "https://avatars.githubusercontent.com/u/4861350?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/sukuya",
"html_url": "https://github.com/sukuya",
"followers_url": "https://api.github.com/users/sukuya/foll... | [] | closed | false | null | [] | [
"# [Codecov](https://codecov.io/gh/huggingface/pytorch-transformers/pull/899?src=pr&el=h1) Report\n> Merging [#899](https://codecov.io/gh/huggingface/pytorch-transformers/pull/899?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/pytorch-transformers/commit/067923d3267325f525f4e46f357360c191ba562e?src... | 1,564 | 1,564 | 1,564 | CONTRIBUTOR | null | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/899/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/899/timeline | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/899",
"html_url": "https://github.com/huggingface/transformers/pull/899",
"diff_url": "https://github.com/huggingface/transformers/pull/899.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/899.patch",
"merged_at": 15640... | |
https://api.github.com/repos/huggingface/transformers/issues/898 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/898/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/898/comments | https://api.github.com/repos/huggingface/transformers/issues/898/events | https://github.com/huggingface/transformers/issues/898 | 472,699,140 | MDU6SXNzdWU0NzI2OTkxNDA= | 898 | fp16 is still has the problem | {
"login": "zsk423200",
"id": 18025765,
"node_id": "MDQ6VXNlcjE4MDI1NzY1",
"avatar_url": "https://avatars.githubusercontent.com/u/18025765?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/zsk423200",
"html_url": "https://github.com/zsk423200",
"followers_url": "https://api.github.com/users/... | [] | closed | false | null | [] | [
"yes, I fixed it in #896 and waiting for author to merge..\r\nbut I still can't figure out why fp16 didn't save memory and didn't speed up....",
"Merged",
"close"
] | 1,564 | 1,564 | 1,564 | NONE | null | hello, as mentioned in #868 and #871 ,fp16 is broken, and you have fixed in the master once, but i am afraid it also has problem, DP is need after the amp.initialize() too. i review the code of apex, found that amp do not support the model of parallel type:
```
def check_models(models):
for model in models:
... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/898/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/898/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/897 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/897/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/897/comments | https://api.github.com/repos/huggingface/transformers/issues/897/events | https://github.com/huggingface/transformers/pull/897 | 472,682,991 | MDExOlB1bGxSZXF1ZXN0MzAwOTk2NDM3 | 897 | Fix FileNotFoundError when running on SQuAD-v1.1 | {
"login": "bzantium",
"id": 19511788,
"node_id": "MDQ6VXNlcjE5NTExNzg4",
"avatar_url": "https://avatars.githubusercontent.com/u/19511788?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/bzantium",
"html_url": "https://github.com/bzantium",
"followers_url": "https://api.github.com/users/bza... | [] | closed | false | null | [] | [
"Duplicate to #882 "
] | 1,564 | 1,564 | 1,564 | CONTRIBUTOR | null | At "utils_squad_evaluate.py" line 291, no matter version_2_with_negative is True or False, it tries to load "output_null_log_odds_file" which is not saved when version_2_with_negative is False. | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/897/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/897/timeline | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/897",
"html_url": "https://github.com/huggingface/transformers/pull/897",
"diff_url": "https://github.com/huggingface/transformers/pull/897.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/897.patch",
"merged_at": null
... |
https://api.github.com/repos/huggingface/transformers/issues/896 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/896/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/896/comments | https://api.github.com/repos/huggingface/transformers/issues/896/events | https://github.com/huggingface/transformers/pull/896 | 472,668,372 | MDExOlB1bGxSZXF1ZXN0MzAwOTg0Nzk0 | 896 | fix multi-gpu training bug when using fp16 | {
"login": "zijunsun",
"id": 20966464,
"node_id": "MDQ6VXNlcjIwOTY2NDY0",
"avatar_url": "https://avatars.githubusercontent.com/u/20966464?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/zijunsun",
"html_url": "https://github.com/zijunsun",
"followers_url": "https://api.github.com/users/zij... | [] | closed | false | null | [] | [
"Thanks, can you update `run_squad` similarly?",
"> Thanks, can you update `run_squad` similarly?\r\n\r\nupdated already.",
"# [Codecov](https://codecov.io/gh/huggingface/pytorch-transformers/pull/896?src=pr&el=h1) Report\n> Merging [#896](https://codecov.io/gh/huggingface/pytorch-transformers/pull/896?src=pr&e... | 1,564 | 1,564 | 1,564 | CONTRIBUTOR | null | multi-gpu training (orch.nn.DataParallel) should also be after apex fp16 initialization. | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/896/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/896/timeline | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/896",
"html_url": "https://github.com/huggingface/transformers/pull/896",
"diff_url": "https://github.com/huggingface/transformers/pull/896.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/896.patch",
"merged_at": 15641... |
https://api.github.com/repos/huggingface/transformers/issues/895 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/895/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/895/comments | https://api.github.com/repos/huggingface/transformers/issues/895/events | https://github.com/huggingface/transformers/pull/895 | 472,662,788 | MDExOlB1bGxSZXF1ZXN0MzAwOTgwMzg2 | 895 | fix a bug of saving added tokens | {
"login": "askerlee",
"id": 1575461,
"node_id": "MDQ6VXNlcjE1NzU0NjE=",
"avatar_url": "https://avatars.githubusercontent.com/u/1575461?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/askerlee",
"html_url": "https://github.com/askerlee",
"followers_url": "https://api.github.com/users/asker... | [] | closed | false | null | [] | [
"# [Codecov](https://codecov.io/gh/huggingface/pytorch-transformers/pull/895?src=pr&el=h1) Report\n> Merging [#895](https://codecov.io/gh/huggingface/pytorch-transformers/pull/895?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/pytorch-transformers/commit/067923d3267325f525f4e46f357360c191ba562e?src... | 1,564 | 1,564 | 1,564 | NONE | null | Refer to the code that loads `added_tokens.json`:
`added_tok_encoder = json.load(open(added_tokens_file, encoding="utf-8"))`
We can see that `added_tokens_encoder` should be saved in `added_tokens.json`. But the original code saved `added_tokens_decoder`.
| {
"url": "https://api.github.com/repos/huggingface/transformers/issues/895/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/895/timeline | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/895",
"html_url": "https://github.com/huggingface/transformers/pull/895",
"diff_url": "https://github.com/huggingface/transformers/pull/895.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/895.patch",
"merged_at": null
... |
https://api.github.com/repos/huggingface/transformers/issues/894 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/894/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/894/comments | https://api.github.com/repos/huggingface/transformers/issues/894/events | https://github.com/huggingface/transformers/issues/894 | 472,642,820 | MDU6SXNzdWU0NzI2NDI4MjA= | 894 | Sequence length more than 512 | {
"login": "nayakt",
"id": 15123057,
"node_id": "MDQ6VXNlcjE1MTIzMDU3",
"avatar_url": "https://avatars.githubusercontent.com/u/15123057?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/nayakt",
"html_url": "https://github.com/nayakt",
"followers_url": "https://api.github.com/users/nayakt/fo... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"https://github.com/google-research/bert/issues/27#issuecomment-435265194",
"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"
] | 1,564 | 1,569 | 1,569 | NONE | null | Hi,
My dataset has sequence with more than 512 words and when use wordpieces sequence length goes beyond 512. How to handle this issue with BERT ?
Regards
Tapas | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/894/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/894/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/893 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/893/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/893/comments | https://api.github.com/repos/huggingface/transformers/issues/893/events | https://github.com/huggingface/transformers/pull/893 | 472,602,474 | MDExOlB1bGxSZXF1ZXN0MzAwOTM0NDgy | 893 | make save_pretrained do the right thing with added tokens | {
"login": "joelgrus",
"id": 1308313,
"node_id": "MDQ6VXNlcjEzMDgzMTM=",
"avatar_url": "https://avatars.githubusercontent.com/u/1308313?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/joelgrus",
"html_url": "https://github.com/joelgrus",
"followers_url": "https://api.github.com/users/joelg... | [] | closed | false | null | [] | [
"# [Codecov](https://codecov.io/gh/huggingface/pytorch-transformers/pull/893?src=pr&el=h1) Report\n> Merging [#893](https://codecov.io/gh/huggingface/pytorch-transformers/pull/893?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/pytorch-transformers/commit/067923d3267325f525f4e46f357360c191ba562e?src... | 1,564 | 1,564 | 1,564 | CONTRIBUTOR | null | right now it's dumping the *decoder* when it should be dumping the *encoder*. and then (for obvious reasons) you get an error when you try to load "from_pretrained" using that dump.
this PR fixes that. | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/893/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/893/timeline | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/893",
"html_url": "https://github.com/huggingface/transformers/pull/893",
"diff_url": "https://github.com/huggingface/transformers/pull/893.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/893.patch",
"merged_at": 15640... |
https://api.github.com/repos/huggingface/transformers/issues/892 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/892/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/892/comments | https://api.github.com/repos/huggingface/transformers/issues/892/events | https://github.com/huggingface/transformers/issues/892 | 472,594,468 | MDU6SXNzdWU0NzI1OTQ0Njg= | 892 | How to add new special token | {
"login": "dchang56",
"id": 24575558,
"node_id": "MDQ6VXNlcjI0NTc1NTU4",
"avatar_url": "https://avatars.githubusercontent.com/u/24575558?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/dchang56",
"html_url": "https://github.com/dchang56",
"followers_url": "https://api.github.com/users/dch... | [] | closed | false | null | [] | [
"How did you add a new vocab.txt file? ",
"I actually figured it out. I manually replaced one of the unused tokens in the vocab file with [NEW] and added \"additiona_special_tokens\": \"[NEW]\" to the special_tokens.json file in the same directory as the vocab.txt file. It works, but I realized that adding new to... | 1,564 | 1,564 | 1,564 | NONE | null | I noticed the never_split functionality is no longer used to keep track of special tokens to never split on. If I wanted to add a new special token like '[NEW]' so the tokenizer never splits it, how should I go about doing that? (I've already manually added it to vocab.txt by replacing an unused token with [NEW]. Now I... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/892/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/892/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/891 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/891/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/891/comments | https://api.github.com/repos/huggingface/transformers/issues/891/events | https://github.com/huggingface/transformers/issues/891 | 472,594,256 | MDU6SXNzdWU0NzI1OTQyNTY= | 891 | BERT: run_squad.py falling over after eval | {
"login": "jforbes14",
"id": 29598836,
"node_id": "MDQ6VXNlcjI5NTk4ODM2",
"avatar_url": "https://avatars.githubusercontent.com/u/29598836?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/jforbes14",
"html_url": "https://github.com/jforbes14",
"followers_url": "https://api.github.com/users/... | [] | closed | false | null | [] | [
"Fixed in #882"
] | 1,564 | 1,564 | 1,564 | NONE | null | I'm having an issue fine-tuning BERT with run.squad.py, as it falls over at the end of the evaluation stage. I'm fine tuning on SQuAD v1.1. Has anyone else encountered the same issue, or is able to point out where I'm going wrong?
`python run_squad.py \
--model_type bert \
--model_name_or_path bert-base-un... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/891/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/891/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/890 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/890/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/890/comments | https://api.github.com/repos/huggingface/transformers/issues/890/events | https://github.com/huggingface/transformers/issues/890 | 472,522,763 | MDU6SXNzdWU0NzI1MjI3NjM= | 890 | PreTrainedTokenizer.from_pretrained should be more general | {
"login": "matt-gardner",
"id": 3291951,
"node_id": "MDQ6VXNlcjMyOTE5NTE=",
"avatar_url": "https://avatars.githubusercontent.com/u/3291951?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/matt-gardner",
"html_url": "https://github.com/matt-gardner",
"followers_url": "https://api.github.com... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"Yes, this is a nice idea, I was thinking about implementing something like this for another reason (simplifying the task of maintaining `torch.hub` configuration files).\r\n\r\nRegarding the library architecture, I think it's better to make a new (very simple) class, something like `AutoTokenizer` in a new file `t... | 1,564 | 1,569 | 1,569 | NONE | null | I'm trying to implement a general interface to any of these transformer models in AllenNLP. I would love to be able to do something like `PreTrainedTokenizer.from_pretrained(model_name)`, and have this work for any model name across any of your implemented models. It looks like what needs to happen for this is to det... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/890/reactions",
"total_count": 5,
"+1": 4,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 1,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/890/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/889 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/889/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/889/comments | https://api.github.com/repos/huggingface/transformers/issues/889/events | https://github.com/huggingface/transformers/issues/889 | 472,485,595 | MDU6SXNzdWU0NzI0ODU1OTU= | 889 | Increased number of hidden states returned from transformers in latest release | {
"login": "anlsh",
"id": 2720400,
"node_id": "MDQ6VXNlcjI3MjA0MDA=",
"avatar_url": "https://avatars.githubusercontent.com/u/2720400?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/anlsh",
"html_url": "https://github.com/anlsh",
"followers_url": "https://api.github.com/users/anlsh/follower... | [] | closed | false | null | [] | [
"Yes, this is the initial embedding layer (i.e. this is layers \"0\" through 12 or 24). There are a number of small changes that haven't been migrated into the documentation yet. "
] | 1,563 | 1,564 | 1,564 | NONE | null | I noticed an (undocumented?) change in the latest release: namely that transformers now include the pre-encoder input vector in the list returned when `output_hidden_states` is True.
For example, the `hidden_states` output from `BertEncoder` now returns a length-13 list of tensors, whereas it used to return a length... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/889/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/889/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/888 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/888/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/888/comments | https://api.github.com/repos/huggingface/transformers/issues/888/events | https://github.com/huggingface/transformers/pull/888 | 472,431,708 | MDExOlB1bGxSZXF1ZXN0MzAwODQzMDk0 | 888 | Update docs for parameter rename | {
"login": "rococode",
"id": 32279130,
"node_id": "MDQ6VXNlcjMyMjc5MTMw",
"avatar_url": "https://avatars.githubusercontent.com/u/32279130?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/rococode",
"html_url": "https://github.com/rococode",
"followers_url": "https://api.github.com/users/roc... | [] | closed | false | null | [] | [
"# [Codecov](https://codecov.io/gh/huggingface/pytorch-transformers/pull/888?src=pr&el=h1) Report\n> Merging [#888](https://codecov.io/gh/huggingface/pytorch-transformers/pull/888?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/pytorch-transformers/commit/067923d3267325f525f4e46f357360c191ba562e?src... | 1,563 | 1,564 | 1,564 | CONTRIBUTOR | null | small fix: OpenAIGPTLMHeadModel now accepts `labels` instead of `lm_labels`
| {
"url": "https://api.github.com/repos/huggingface/transformers/issues/888/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/888/timeline | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/888",
"html_url": "https://github.com/huggingface/transformers/pull/888",
"diff_url": "https://github.com/huggingface/transformers/pull/888.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/888.patch",
"merged_at": 15640... |
https://api.github.com/repos/huggingface/transformers/issues/887 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/887/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/887/comments | https://api.github.com/repos/huggingface/transformers/issues/887/events | https://github.com/huggingface/transformers/issues/887 | 472,368,073 | MDU6SXNzdWU0NzIzNjgwNzM= | 887 | No gradient clipping in AdamW | {
"login": "OlegPlatonov",
"id": 32016523,
"node_id": "MDQ6VXNlcjMyMDE2NTIz",
"avatar_url": "https://avatars.githubusercontent.com/u/32016523?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/OlegPlatonov",
"html_url": "https://github.com/OlegPlatonov",
"followers_url": "https://api.github.c... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"Yes, the LM fine-tuning example will be refactored.\r\n\r\nAdding the removal of gradient clipping to the list of breaking changes, thanks.",
"This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contribution... | 1,563 | 1,570 | 1,570 | NONE | null | Hi!
After moving from pretrained-bert to transformers I've noticed that the new AdamW optimizer does not perform gradient clipping, even though both BertAdam and OpenAIAdam used to do it.
Also, in finetune_on_pregenerated example bias correction is turned off only for FusedAdam, but not for AdamW. | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/887/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/887/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/886 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/886/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/886/comments | https://api.github.com/repos/huggingface/transformers/issues/886/events | https://github.com/huggingface/transformers/issues/886 | 472,326,469 | MDU6SXNzdWU0NzIzMjY0Njk= | 886 | BERT uncased model outputs a tuple instead of a normal pytorch tensor | {
"login": "rajaswa",
"id": 34607601,
"node_id": "MDQ6VXNlcjM0NjA3NjAx",
"avatar_url": "https://avatars.githubusercontent.com/u/34607601?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/rajaswa",
"html_url": "https://github.com/rajaswa",
"followers_url": "https://api.github.com/users/rajasw... | [] | closed | false | null | [] | [
"Hi, I was wondering how you managed to resolve this issue? I'm running into a similar problem. :) ",
"Hi, the model outputs are well documented, they're *always* tuples, even if there's a single return value. You can check the documentation [here](https://huggingface.co/transformers/main_classes/output.html).",
... | 1,563 | 1,606 | 1,563 | NONE | null | While finetuning the BERT uncased model for sequence classification as follows:
```
config = BertConfig.from_pretrained('bert-base-uncased')
tokenizer = BertTokenizer.from_pretrained('bert-base-uncased')
model = BertForSequenceClassification(config)
for layer, child in model.named_children():
if layer not... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/886/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/886/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/885 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/885/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/885/comments | https://api.github.com/repos/huggingface/transformers/issues/885/events | https://github.com/huggingface/transformers/issues/885 | 472,307,117 | MDU6SXNzdWU0NzIzMDcxMTc= | 885 | Can lm_finetuning be used with non-english data ? | {
"login": "ereday",
"id": 13196191,
"node_id": "MDQ6VXNlcjEzMTk2MTkx",
"avatar_url": "https://avatars.githubusercontent.com/u/13196191?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ereday",
"html_url": "https://github.com/ereday",
"followers_url": "https://api.github.com/users/ereday/fo... | [] | closed | false | null | [] | [
"@ereday Take the `simple_lm_finetuning.py` script for example. It has a `--bert_model` argument. When your target domain is German, then you should use the recently introduced [BERT model for german](https://github.com/huggingface/pytorch-transformers/pull/688) via passing `bert-base-german-cased`. \r\n\r\nThis sh... | 1,563 | 1,564 | 1,564 | NONE | null | Hi,
My target domain is in German. Can I still use the scripts &codes under `lm_finetuning` folder to finetune pre-trained Bert models or are those only for English target domains? | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/885/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/885/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/884 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/884/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/884/comments | https://api.github.com/repos/huggingface/transformers/issues/884/events | https://github.com/huggingface/transformers/issues/884 | 472,224,934 | MDU6SXNzdWU0NzIyMjQ5MzQ= | 884 | Customized BertForTokenClassification Model | {
"login": "lixin4ever",
"id": 18526640,
"node_id": "MDQ6VXNlcjE4NTI2NjQw",
"avatar_url": "https://avatars.githubusercontent.com/u/18526640?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/lixin4ever",
"html_url": "https://github.com/lixin4ever",
"followers_url": "https://api.github.com/use... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"@lixin4ever I have the same question. Do you solve it?",
"I have solve it. Thank you!",
"@searchlink How do you solve this problem?",
"For reference, the updated resource link mentioned in the original post can be now found [here](https://huggingface.co/transformers/_modules/transformers/modeling_bert.html#... | 1,563 | 1,587 | 1,587 | NONE | null | I try to customize BertForTokenClassification model by myself to perform sequence tagging and strictly follow the [original implementation](https://huggingface.co/pytorch-transformers/_modules/pytorch_transformers/modeling_bert.html#BertForTokenClassification). However, I cannot obtain the same results (lower scores) w... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/884/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/884/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/883 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/883/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/883/comments | https://api.github.com/repos/huggingface/transformers/issues/883/events | https://github.com/huggingface/transformers/issues/883 | 472,197,750 | MDU6SXNzdWU0NzIxOTc3NTA= | 883 | Upgrade to new FP16 | {
"login": "bhavsarpratik",
"id": 23080576,
"node_id": "MDQ6VXNlcjIzMDgwNTc2",
"avatar_url": "https://avatars.githubusercontent.com/u/23080576?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/bhavsarpratik",
"html_url": "https://github.com/bhavsarpratik",
"followers_url": "https://api.githu... | [] | closed | false | null | [] | [
"Just saw run_glue has the new one."
] | 1,563 | 1,564 | 1,564 | NONE | null | The original FP16_Optimizer and the old “Amp” API are deprecated and subject to removal at any time. Should we consider moving to the new one?
https://nvidia.github.io/apex/amp.html#for-users-of-the-old-fp16-optimizer | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/883/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/883/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/882 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/882/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/882/comments | https://api.github.com/repos/huggingface/transformers/issues/882/events | https://github.com/huggingface/transformers/pull/882 | 472,140,936 | MDExOlB1bGxSZXF1ZXN0MzAwNjA5MDg5 | 882 | fix squad v1 error (na_prob_file should be None) | {
"login": "Liangtaiwan",
"id": 20909894,
"node_id": "MDQ6VXNlcjIwOTA5ODk0",
"avatar_url": "https://avatars.githubusercontent.com/u/20909894?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Liangtaiwan",
"html_url": "https://github.com/Liangtaiwan",
"followers_url": "https://api.github.com/... | [] | closed | false | null | [] | [
"# [Codecov](https://codecov.io/gh/huggingface/pytorch-transformers/pull/882?src=pr&el=h1) Report\n> Merging [#882](https://codecov.io/gh/huggingface/pytorch-transformers/pull/882?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/pytorch-transformers/commit/067923d3267325f525f4e46f357360c191ba562e?src... | 1,563 | 1,564 | 1,564 | CONTRIBUTOR | null | When running squad v1, na_prob_file should be None.
Or there will be an error when evaluate on testing data. | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/882/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/882/timeline | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/882",
"html_url": "https://github.com/huggingface/transformers/pull/882",
"diff_url": "https://github.com/huggingface/transformers/pull/882.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/882.patch",
"merged_at": 15640... |
https://api.github.com/repos/huggingface/transformers/issues/881 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/881/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/881/comments | https://api.github.com/repos/huggingface/transformers/issues/881/events | https://github.com/huggingface/transformers/issues/881 | 472,126,020 | MDU6SXNzdWU0NzIxMjYwMjA= | 881 | can not convert_tf_checkpoint_to_pytorch | {
"login": "Zhangxuri",
"id": 19551110,
"node_id": "MDQ6VXNlcjE5NTUxMTEw",
"avatar_url": "https://avatars.githubusercontent.com/u/19551110?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Zhangxuri",
"html_url": "https://github.com/Zhangxuri",
"followers_url": "https://api.github.com/users/... | [] | closed | false | null | [] | [
"```\r\npython convert.py --tf_checkpoint_path=./uncased_L-12_H-768_A-12/bert_model.ckpt --bert_config_file=./uncased_L-12_H-768_A-12/bert_config.json --pytorch_dump_path=./uncased_L-12_H-768_A-12/bert_model.bin\r\n```"
] | 1,563 | 1,563 | 1,563 | NONE | null | ```
.
├── convert_tf_checkpoint_to_pytorch.py
├── uncased_L-12_H-768_A-12
│ ├── bert_config.json
│ ├── bert_model.ckpt.data-00000-of-00001
│ ├── bert_model.ckpt.index
│ ├── bert_model.ckpt.meta
│ └── vocab.txt
├── uncased_L-12_H-768_A-12.zip
└── Untitled.ipynb
```
```
(base) ➜ ckpt_to_bin git:... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/881/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/881/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/880 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/880/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/880/comments | https://api.github.com/repos/huggingface/transformers/issues/880/events | https://github.com/huggingface/transformers/issues/880 | 472,115,663 | MDU6SXNzdWU0NzIxMTU2NjM= | 880 | Printing Iteration every example problem | {
"login": "AhmedBahaaElDinMohammed",
"id": 51789113,
"node_id": "MDQ6VXNlcjUxNzg5MTEz",
"avatar_url": "https://avatars.githubusercontent.com/u/51789113?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/AhmedBahaaElDinMohammed",
"html_url": "https://github.com/AhmedBahaaElDinMohammed",
"foll... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"Are you running in jupyter? This might be an artifact of how `tqdm` is interacting with whatever shell you're running it in. If you don't want to print anything, you could simply drop the `tdqm` wrapper and just iterate over `train_dataloader`. ",
"This issue has been automatically marked as stale because it has... | 1,563 | 1,570 | 1,570 | NONE | null | ```
Iteration: 0%| | 1/250 [00:00<03:28, 1.19it/s]
Iteration: 1%| | 2/250 [00:01<03:21, 1.23it/s]
Iteration: 1%| | 3/250 [00:02<03:17, 1.25it/s]
Iteration: 2%|▏ | 4/250 [00:03<03:14, 1.27it/s]
Iteration: 2%|▏ | 5/250 [00:03<03... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/880/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/880/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/879 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/879/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/879/comments | https://api.github.com/repos/huggingface/transformers/issues/879/events | https://github.com/huggingface/transformers/pull/879 | 472,114,745 | MDExOlB1bGxSZXF1ZXN0MzAwNTg4MjM0 | 879 | fix #878 | {
"login": "xijiz",
"id": 12234085,
"node_id": "MDQ6VXNlcjEyMjM0MDg1",
"avatar_url": "https://avatars.githubusercontent.com/u/12234085?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/xijiz",
"html_url": "https://github.com/xijiz",
"followers_url": "https://api.github.com/users/xijiz/follow... | [] | closed | false | null | [] | [
"# [Codecov](https://codecov.io/gh/huggingface/pytorch-transformers/pull/879?src=pr&el=h1) Report\n> Merging [#879](https://codecov.io/gh/huggingface/pytorch-transformers/pull/879?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/pytorch-transformers/commit/067923d3267325f525f4e46f357360c191ba562e?src... | 1,563 | 1,566 | 1,566 | NONE | null | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/879/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/879/timeline | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/879",
"html_url": "https://github.com/huggingface/transformers/pull/879",
"diff_url": "https://github.com/huggingface/transformers/pull/879.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/879.patch",
"merged_at": null
... | |
https://api.github.com/repos/huggingface/transformers/issues/878 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/878/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/878/comments | https://api.github.com/repos/huggingface/transformers/issues/878/events | https://github.com/huggingface/transformers/issues/878 | 472,113,887 | MDU6SXNzdWU0NzIxMTM4ODc= | 878 | Fail to load pre-trained tokens. | {
"login": "xijiz",
"id": 12234085,
"node_id": "MDQ6VXNlcjEyMjM0MDg1",
"avatar_url": "https://avatars.githubusercontent.com/u/12234085?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/xijiz",
"html_url": "https://github.com/xijiz",
"followers_url": "https://api.github.com/users/xijiz/follow... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"Hi, how do you solve this problem? If we set `pretrained_model_name_or_path` as a path to vocab.txt, it still need the two files: added_tokens.json, special_tokens_map.json. Where can we get these files? ",
"> Hi, how do you solve this problem? If we set `pretrained_model_name_or_path` as a path to vocab.txt, it... | 1,563 | 1,571 | 1,571 | NONE | null | The PreTrainedTokenizer fails to load tokenizer files when I load tokenizer files from local tokenizer files.
The error is caused by code line 174 - 182 in [tokenization_utils.py](https://github.com/huggingface/pytorch-transformers/blob/master/pytorch_transformers/tokenization_utils.py). The code assumes that there ... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/878/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/878/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/877 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/877/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/877/comments | https://api.github.com/repos/huggingface/transformers/issues/877/events | https://github.com/huggingface/transformers/issues/877 | 471,990,319 | MDU6SXNzdWU0NzE5OTAzMTk= | 877 | error when tried to migrate from pretrained-bert to transformers. | {
"login": "y8miao",
"id": 47309353,
"node_id": "MDQ6VXNlcjQ3MzA5MzUz",
"avatar_url": "https://avatars.githubusercontent.com/u/47309353?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/y8miao",
"html_url": "https://github.com/y8miao",
"followers_url": "https://api.github.com/users/y8miao/fo... | [] | closed | false | null | [] | [
"The `labels` input for the model is not the number of labels but the tensor of labels (see the docstrings and doc).",
"> The `labels` input for the model is not the number of labels but the tensor of labels (see the docstrings and doc).\r\n\r\nThank you for the answer. I'm trying to train the model to do polarit... | 1,563 | 1,568 | 1,563 | NONE | null | The code used to be:
`logits = model(input_ids, segment_ids, input_mask, labels=None)
if OUTPUT_MODE == "classification":
loss_fct = CrossEntropyLoss()
loss = loss_fct(logits.view(-1, num_labels), label_ids.view(-1))
elif OUTPUT_MODE == "regression":
loss_fct ... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/877/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/877/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/876 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/876/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/876/comments | https://api.github.com/repos/huggingface/transformers/issues/876/events | https://github.com/huggingface/transformers/issues/876 | 471,988,735 | MDU6SXNzdWU0NzE5ODg3MzU= | 876 | How to use BERT for finding similar sentences or similar news? | {
"login": "Raghavendra15",
"id": 7957331,
"node_id": "MDQ6VXNlcjc5NTczMzE=",
"avatar_url": "https://avatars.githubusercontent.com/u/7957331?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Raghavendra15",
"html_url": "https://github.com/Raghavendra15",
"followers_url": "https://api.github.... | [] | closed | false | null | [] | [
"Hi,\r\nBERT out-of-the-box is not the best option for this task, as the run-time in your setup scales with the number of sentences in your corpus. I.e., if you have 10,000 sentences/articles in your corpus, you need to classify 10k pairs with BERT, which is rather slow.\r\n\r\nA better option is to generate senten... | 1,563 | 1,698 | 1,619 | NONE | null | I have used BERT NextSentencePredictor to find similar sentences or similar news, However, It's super slow. Even on Tesla V100 which is the fastest GPU till now. It takes around 10secs for a query title with around 3,000 articles. Is there a way to use BERT better for finding similar sentences or similar news given a c... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/876/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/876/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/875 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/875/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/875/comments | https://api.github.com/repos/huggingface/transformers/issues/875/events | https://github.com/huggingface/transformers/issues/875 | 471,850,052 | MDU6SXNzdWU0NzE4NTAwNTI= | 875 | XLNet bidirectional input pipeline requires batch size at least 2 | {
"login": "langfield",
"id": 35980963,
"node_id": "MDQ6VXNlcjM1OTgwOTYz",
"avatar_url": "https://avatars.githubusercontent.com/u/35980963?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/langfield",
"html_url": "https://github.com/langfield",
"followers_url": "https://api.github.com/users/... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | 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"
] | 1,563 | 1,569 | 1,569 | NONE | null | This may not be a true bug since it's mentioned in the paper that
> each of the forward and backward directions takes half of the batch size
but when using the bidirectional input pipeline, any call to `XLNetModel.forward()` will raise an error of the form
```
RuntimeError: shape '[x, y, z]' is invalid for ... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/875/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/875/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/874 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/874/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/874/comments | https://api.github.com/repos/huggingface/transformers/issues/874/events | https://github.com/huggingface/transformers/issues/874 | 471,835,517 | MDU6SXNzdWU0NzE4MzU1MTc= | 874 | Fine-tuning model and Generation | {
"login": "antmarakis",
"id": 17463361,
"node_id": "MDQ6VXNlcjE3NDYzMzYx",
"avatar_url": "https://avatars.githubusercontent.com/u/17463361?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/antmarakis",
"html_url": "https://github.com/antmarakis",
"followers_url": "https://api.github.com/use... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"Having the same question - how to use bert for generation? ",
"the same problem， how to train my own data for text generation?",
"We'll add an example for fine-tuning this month.",
"@thomwolf as I read in other issues, BERT model cannot be used to generate text directly (your reply https://github.com/hugging... | 1,563 | 1,616 | 1,573 | CONTRIBUTOR | null | Hello!
I am beginner and I just wanted to run some experiments, but I've hit a road block. I am trying to generate text using `run_generator.py` after I fine-tune a model on my data using `simple_lm_finetuning.py`. I've looked around a bit, and I'm not sure how to go about this, or if this is possible at all. I don'... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/874/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/874/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/873 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/873/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/873/comments | https://api.github.com/repos/huggingface/transformers/issues/873/events | https://github.com/huggingface/transformers/pull/873 | 471,779,098 | MDExOlB1bGxSZXF1ZXN0MzAwMzYzMjcw | 873 | Add nn.Identity replacement for old PyTorch | {
"login": "thomwolf",
"id": 7353373,
"node_id": "MDQ6VXNlcjczNTMzNzM=",
"avatar_url": "https://avatars.githubusercontent.com/u/7353373?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/thomwolf",
"html_url": "https://github.com/thomwolf",
"followers_url": "https://api.github.com/users/thomw... | [] | closed | false | null | [] | [] | 1,563 | 1,578 | 1,563 | MEMBER | null | Fix #869 to keep at least PyTorch 1.0.0 compatiblity. | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/873/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/873/timeline | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/873",
"html_url": "https://github.com/huggingface/transformers/pull/873",
"diff_url": "https://github.com/huggingface/transformers/pull/873.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/873.patch",
"merged_at": 15638... |
https://api.github.com/repos/huggingface/transformers/issues/872 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/872/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/872/comments | https://api.github.com/repos/huggingface/transformers/issues/872/events | https://github.com/huggingface/transformers/pull/872 | 471,714,832 | MDExOlB1bGxSZXF1ZXN0MzAwMzEwOTc5 | 872 | Updating schedules for state_dict saving/loading | {
"login": "thomwolf",
"id": 7353373,
"node_id": "MDQ6VXNlcjczNTMzNzM=",
"avatar_url": "https://avatars.githubusercontent.com/u/7353373?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/thomwolf",
"html_url": "https://github.com/thomwolf",
"followers_url": "https://api.github.com/users/thomw... | [] | closed | false | null | [] | [
"# [Codecov](https://codecov.io/gh/huggingface/pytorch-transformers/pull/872?src=pr&el=h1) Report\n> Merging [#872](https://codecov.io/gh/huggingface/pytorch-transformers/pull/872?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/pytorch-transformers/commit/268c6cc160ba046d6a91747c5f281f82bd88a4d8?src... | 1,563 | 1,578 | 1,563 | MEMBER | null | This PR updates the schedules so that they can be saved/reloaded using the standard `state_dict()` and `load_state_dict()` methods of PyTorch [`LambdaLR` model](https://pytorch.org/docs/stable/optim.html#torch.optim.lr_scheduler.LambdaLR.load_state_dict).
Useful for continuing stopped training as mentioned in #839 | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/872/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/872/timeline | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/872",
"html_url": "https://github.com/huggingface/transformers/pull/872",
"diff_url": "https://github.com/huggingface/transformers/pull/872.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/872.patch",
"merged_at": 15638... |
https://api.github.com/repos/huggingface/transformers/issues/871 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/871/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/871/comments | https://api.github.com/repos/huggingface/transformers/issues/871/events | https://github.com/huggingface/transformers/issues/871 | 471,652,639 | MDU6SXNzdWU0NzE2NTI2Mzk= | 871 | fp16 is not work | {
"login": "zsk423200",
"id": 18025765,
"node_id": "MDQ6VXNlcjE4MDI1NzY1",
"avatar_url": "https://avatars.githubusercontent.com/u/18025765?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/zsk423200",
"html_url": "https://github.com/zsk423200",
"followers_url": "https://api.github.com/users/... | [] | closed | false | null | [] | [
"Duplicate of #868"
] | 1,563 | 1,563 | 1,563 | NONE | null | GPU:v100
run run_glue.py with the command in the README:
```
python ./examples/run_glue.py \
--model_type bert \
--model_name_or_path bert-base-uncased \
--task_name $TASK_NAME \
--do_train \
--do_eval \
--do_lower_case \
--data_dir $GLUE_DIR/$TASK_NAME \
--max_seq_length 12... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/871/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/871/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/870 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/870/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/870/comments | https://api.github.com/repos/huggingface/transformers/issues/870/events | https://github.com/huggingface/transformers/issues/870 | 471,572,906 | MDU6SXNzdWU0NzE1NzI5MDY= | 870 | How to load a fine-tuned model pytorch_model.bin produced by run_bert_swag.py | {
"login": "xiami2019",
"id": 37145051,
"node_id": "MDQ6VXNlcjM3MTQ1MDUx",
"avatar_url": "https://avatars.githubusercontent.com/u/37145051?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/xiami2019",
"html_url": "https://github.com/xiami2019",
"followers_url": "https://api.github.com/users/... | [] | closed | false | null | [] | [
"That's a strange error, what are the exact process you are using and full error log?",
"> That's a strange error, what are the exact process you are using and full error log?\r\n\r\nHi, thanks for the reply.\r\nI used distributed training in one node with 2GPUs and my command is:\r\nexport SWAG_DIR=SWAG; export ... | 1,563 | 1,584 | 1,563 | NONE | null | Hi, guys! I have a little question about how to load a fine-tuned model 'pytorch_model.bin' produced by run_bert_swag.py.
When I load a fine-tuned model pytorch_model.bin with .from_pretrained methods, runtime error occurs as follow:
RuntimeError: storage has wrong size: expected 4357671300540823961 got 589824.
I f... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/870/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/870/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/869 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/869/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/869/comments | https://api.github.com/repos/huggingface/transformers/issues/869/events | https://github.com/huggingface/transformers/issues/869 | 471,551,615 | MDU6SXNzdWU0NzE1NTE2MTU= | 869 | module 'torch.nn' has no attribute 'Identity' | {
"login": "ahlawatankit",
"id": 24656446,
"node_id": "MDQ6VXNlcjI0NjU2NDQ2",
"avatar_url": "https://avatars.githubusercontent.com/u/24656446?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ahlawatankit",
"html_url": "https://github.com/ahlawatankit",
"followers_url": "https://api.github.c... | [] | closed | false | null | [] | [
"This was added in PyTorch 1.1.0 (see [changelog here](https://github.com/pytorch/pytorch/tree/v1.1.0) :)\r\n\r\nSo I guess you just have to update your PyTorch version!",
"Oh yes, I guess we can add a replacement to keep older PyTorch compatibility.\r\nWould be sad to lose backward compatibility just for this.",... | 1,563 | 1,605 | 1,563 | NONE | null | Traceback (most recent call last):
File "trainer.py", line 17, in <module>
model = XLMForSequenceClassification(config)
File "/home/ankit/anaconda3/lib/python3.6/site-packages/pytorch_transformers/modeling_xlm.py", line 823, in __init__
self.sequence_summary = SequenceSummary(config)
File "/home/anki... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/869/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/869/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/868 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/868/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/868/comments | https://api.github.com/repos/huggingface/transformers/issues/868/events | https://github.com/huggingface/transformers/issues/868 | 471,550,517 | MDU6SXNzdWU0NzE1NTA1MTc= | 868 | fp16 is broken | {
"login": "zsk423200",
"id": 18025765,
"node_id": "MDQ6VXNlcjE4MDI1NzY1",
"avatar_url": "https://avatars.githubusercontent.com/u/18025765?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/zsk423200",
"html_url": "https://github.com/zsk423200",
"followers_url": "https://api.github.com/users/... | [] | closed | false | null | [] | [
"Indeed, thanks! Fixed in master"
] | 1,563 | 1,563 | 1,563 | NONE | null | run run_glue.py with the parameter of --fp16, and return error:
```
RuntimeError: Incoming model is an instance of torch.nn.parallel.DistributedDataParallel. Parallel wrappers should only be applied to the model(s) AFTER
the model(s) have been returned from amp.initialize.
```
i find the reason is the wrong orde... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/868/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/868/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/867 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/867/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/867/comments | https://api.github.com/repos/huggingface/transformers/issues/867/events | https://github.com/huggingface/transformers/issues/867 | 471,451,123 | MDU6SXNzdWU0NzE0NTExMjM= | 867 | XLnet sentence vector | {
"login": "luv4me",
"id": 38608151,
"node_id": "MDQ6VXNlcjM4NjA4MTUx",
"avatar_url": "https://avatars.githubusercontent.com/u/38608151?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/luv4me",
"html_url": "https://github.com/luv4me",
"followers_url": "https://api.github.com/users/luv4me/fo... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"You can train the model on a downstream task to get a sentence vector related to your task or you can get a sentence vector by averaging or max-pooling the output sequence of token hidden-states.",
"Try doing:\r\n\r\n```Python\r\nmodel = model_class.from_pretrained(pretrained_weights,\r\n ... | 1,563 | 1,570 | 1,570 | NONE | null | how can i get the XLnet sentence vector by pytorch-transformers. I use the sample but I only get the word vector. it drives me crazy | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/867/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/867/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/866 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/866/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/866/comments | https://api.github.com/repos/huggingface/transformers/issues/866/events | https://github.com/huggingface/transformers/pull/866 | 471,320,188 | MDExOlB1bGxSZXF1ZXN0MzAwMDMxMjQ3 | 866 | Rework how PreTrainedModel.from_pretrained handles its arguments | {
"login": "anlsh",
"id": 2720400,
"node_id": "MDQ6VXNlcjI3MjA0MDA=",
"avatar_url": "https://avatars.githubusercontent.com/u/2720400?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/anlsh",
"html_url": "https://github.com/anlsh",
"followers_url": "https://api.github.com/users/anlsh/follower... | [] | closed | false | null | [] | [
"Hmm, well that's embarrassing. I'll inspect the failing tests some more to see what's up",
"Regarding python 2, yes we want to keep supporting it and thanks for taking care of it.\r\n\r\nGoogle (which is still using python 2) is a major supplier of pretrained model and architectures and having python 2 support i... | 1,563 | 1,563 | 1,563 | NONE | null | Unification of the `from_pretrained` functions belonging to various modules (GPT2PreTrainedModel, OpenAIGPTPreTrainedModel, BertPreTrainedModel) brought changes to the function's argument handling which don't cause any issues within the repository itself (afaik), but have the potential to break a variety of downstream ... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/866/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/866/timeline | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/866",
"html_url": "https://github.com/huggingface/transformers/pull/866",
"diff_url": "https://github.com/huggingface/transformers/pull/866.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/866.patch",
"merged_at": 15638... |
https://api.github.com/repos/huggingface/transformers/issues/865 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/865/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/865/comments | https://api.github.com/repos/huggingface/transformers/issues/865/events | https://github.com/huggingface/transformers/issues/865 | 471,293,096 | MDU6SXNzdWU0NzEyOTMwOTY= | 865 | Using Fp16 half precision makes Bert prediction slower. | {
"login": "Oxi84",
"id": 25420033,
"node_id": "MDQ6VXNlcjI1NDIwMDMz",
"avatar_url": "https://avatars.githubusercontent.com/u/25420033?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Oxi84",
"html_url": "https://github.com/Oxi84",
"followers_url": "https://api.github.com/users/Oxi84/follow... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"gtx 1080. Is there any other way to make the predictions faster?",
"Hi, you need at least a Volta GPU to get benefits from fp16 unfortunately.",
"@thomwolf Does P100 applicable?",
"I don't think so",
"This issue has been automatically marked as stale because it has not had recent activity. It will be clos... | 1,563 | 1,570 | 1,570 | NONE | null | When I use:
model = BertForMaskedLM.from_pretrained('bert-large-cased')
model = model.half()
model.eval()
model.to('cuda')
by adding Fp16:
model = model.half()
It runs around 50% slower. Why is that?
I run it on ubuntu 18.04, cuda 9, pytorch 1.1 and python 3.6.8
| {
"url": "https://api.github.com/repos/huggingface/transformers/issues/865/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/865/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/864 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/864/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/864/comments | https://api.github.com/repos/huggingface/transformers/issues/864/events | https://github.com/huggingface/transformers/pull/864 | 471,267,386 | MDExOlB1bGxSZXF1ZXN0MzAwMDAyODY3 | 864 | Fixed PreTrainedModel.from_pretrained(...) not passing cache_dir to PretrainedConfig.from_pretrained(...) | {
"login": "mbugert",
"id": 23331603,
"node_id": "MDQ6VXNlcjIzMzMxNjAz",
"avatar_url": "https://avatars.githubusercontent.com/u/23331603?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/mbugert",
"html_url": "https://github.com/mbugert",
"followers_url": "https://api.github.com/users/mbuger... | [] | closed | false | null | [] | [
"Indeed thanks. We'll subsume this PR with #866 which add a few other stuff.\r\n\r\nI agree with you on the `pop` pattern. We'll move away from this when the first one of these two events happens: (i) google stop open-sourcing interesting new models or (ii) google stop using python 2 internally ;)",
"Okay! :+1: "... | 1,563 | 1,563 | 1,563 | NONE | null | See #863
It's not a beautiful solution, but neither is the practice of modifying incoming parameters via pop. 🤷♂ | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/864/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/864/timeline | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/864",
"html_url": "https://github.com/huggingface/transformers/pull/864",
"diff_url": "https://github.com/huggingface/transformers/pull/864.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/864.patch",
"merged_at": null
... |
https://api.github.com/repos/huggingface/transformers/issues/863 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/863/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/863/comments | https://api.github.com/repos/huggingface/transformers/issues/863/events | https://github.com/huggingface/transformers/issues/863 | 471,266,449 | MDU6SXNzdWU0NzEyNjY0NDk= | 863 | PreTrainedModel.from_pretrained(...) doesn't pass cache_dir to PretrainedConfig.from_pretrained(...) | {
"login": "mbugert",
"id": 23331603,
"node_id": "MDQ6VXNlcjIzMzMxNjAz",
"avatar_url": "https://avatars.githubusercontent.com/u/23331603?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/mbugert",
"html_url": "https://github.com/mbugert",
"followers_url": "https://api.github.com/users/mbuger... | [] | closed | false | null | [] | [
"Fix with #866"
] | 1,563 | 1,563 | 1,563 | NONE | null | The cache_dir key-value parameter does not work as intended in `PreTrainedModel.from_pretrained(...)`. It is popped from the kwargs, then `PretrainedConfig.from_pretrained(...)` is called which expects this parameter in the kwargs, but it's obviously not there anymore. A default location is used as a fallback, but this... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/863/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/863/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/862 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/862/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/862/comments | https://api.github.com/repos/huggingface/transformers/issues/862/events | https://github.com/huggingface/transformers/issues/862 | 471,219,824 | MDU6SXNzdWU0NzEyMTk4MjQ= | 862 | Bert encodings | {
"login": "shubhamagarwal92",
"id": 7984532,
"node_id": "MDQ6VXNlcjc5ODQ1MzI=",
"avatar_url": "https://avatars.githubusercontent.com/u/7984532?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/shubhamagarwal92",
"html_url": "https://github.com/shubhamagarwal92",
"followers_url": "https://ap... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"I have borrowed most of the ideas from [utils](https://github.com/huggingface/pytorch-transformers/blob/master/examples/utils_glue.py#L391) in one of the examples to create a [script to extract embeddings](https://gist.github.com/shubhamagarwal92/37ccb747f7130a35a8e76aa66d60e014). \r\n\r\nHowever, I am still curio... | 1,563 | 1,572 | 1,570 | CONTRIBUTOR | null | Hi,
Really interesting work!
I want to use BERT embeddings for a downstream task. I have been following the [steps](https://github.com/huggingface/pytorch-transformers#quick-tour) here as:
```
import torch
from pytorch_transformers import BertModel, BertTokenizer
pretrained_weights = 'bert-base-uncased'
... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/862/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/862/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/861 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/861/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/861/comments | https://api.github.com/repos/huggingface/transformers/issues/861/events | https://github.com/huggingface/transformers/issues/861 | 471,178,942 | MDU6SXNzdWU0NzExNzg5NDI= | 861 | Deleting models | {
"login": "RuiPChaves",
"id": 33401801,
"node_id": "MDQ6VXNlcjMzNDAxODAx",
"avatar_url": "https://avatars.githubusercontent.com/u/33401801?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/RuiPChaves",
"html_url": "https://github.com/RuiPChaves",
"followers_url": "https://api.github.com/use... | [] | closed | false | null | [] | [
"Models are usually located under `~/.cache/torch/pytorch_pretrained_bert` (older version of this library) or `~/.cache/torch/pytorch_transformers` (now) :)",
"Thank you! I found it elsewhere, actually. What worked for me (quaintly) was:\r\n find . -type f -size +1G -print 2>/dev/null\r\n",
"hey, do you guys k... | 1,563 | 1,605 | 1,563 | NONE | null | I would like to delete the 'bert-base-uncased' and 'bert-large-uncased' models and the tokenizer from my hardrive (working under Ubuntu 18.04). I assumed that uninstalling pytorch-pretrained-bert would do it, but it did not. Where are these models located at?
Thanks!
| {
"url": "https://api.github.com/repos/huggingface/transformers/issues/861/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/861/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/860 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/860/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/860/comments | https://api.github.com/repos/huggingface/transformers/issues/860/events | https://github.com/huggingface/transformers/pull/860 | 471,076,815 | MDExOlB1bGxSZXF1ZXN0Mjk5ODYzNzQ0 | 860 | read().splitlines() -> readlines() | {
"login": "Yiqing-Zhou",
"id": 40547184,
"node_id": "MDQ6VXNlcjQwNTQ3MTg0",
"avatar_url": "https://avatars.githubusercontent.com/u/40547184?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Yiqing-Zhou",
"html_url": "https://github.com/Yiqing-Zhou",
"followers_url": "https://api.github.com/... | [] | closed | false | null | [] | [
"# [Codecov](https://codecov.io/gh/huggingface/pytorch-transformers/pull/860?src=pr&el=h1) Report\n> Merging [#860](https://codecov.io/gh/huggingface/pytorch-transformers/pull/860?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/pytorch-transformers/commit/2f869dc6651f9cf9253f4c5a43279027a0eccfc5?src... | 1,563 | 1,563 | 1,563 | CONTRIBUTOR | null | splitlines() does not work as what we expect here for bert-base-chinese because there is a '\u2028' (unicode line seperator) token in vocab file. Value of '\u2028'.splitlines() is ['', ''].
Perhaps we should use readlines() instead. | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/860/reactions",
"total_count": 1,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 1,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/860/timeline | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/860",
"html_url": "https://github.com/huggingface/transformers/pull/860",
"diff_url": "https://github.com/huggingface/transformers/pull/860.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/860.patch",
"merged_at": 15638... |
https://api.github.com/repos/huggingface/transformers/issues/859 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/859/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/859/comments | https://api.github.com/repos/huggingface/transformers/issues/859/events | https://github.com/huggingface/transformers/issues/859 | 471,065,871 | MDU6SXNzdWU0NzEwNjU4NzE= | 859 | Bug of BertTokenizer | {
"login": "LeeJuly30",
"id": 31088072,
"node_id": "MDQ6VXNlcjMxMDg4MDcy",
"avatar_url": "https://avatars.githubusercontent.com/u/31088072?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/LeeJuly30",
"html_url": "https://github.com/LeeJuly30",
"followers_url": "https://api.github.com/users/... | [] | closed | false | null | [] | [
"If you are loading the chinese model, this is probably related to #860 and #825.\r\nShould be fixed now.",
"thanks!"
] | 1,563 | 1,563 | 1,563 | NONE | null | when load a tokenizer from pretrain
```python
tokenizer = BertTokenizer.from_pretrained(vocab_path)
```
the vocab length is:
```python
len(tokenizer.vocab)
21128
```
but the last token of vocab is:
```python
next(reversed(tokenizer.vocab.items()))
('##😎', 21129)
```
| {
"url": "https://api.github.com/repos/huggingface/transformers/issues/859/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/859/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/858 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/858/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/858/comments | https://api.github.com/repos/huggingface/transformers/issues/858/events | https://github.com/huggingface/transformers/issues/858 | 471,064,929 | MDU6SXNzdWU0NzEwNjQ5Mjk= | 858 | CLS segment_id for BERT | {
"login": "guillaume-be",
"id": 27071604,
"node_id": "MDQ6VXNlcjI3MDcxNjA0",
"avatar_url": "https://avatars.githubusercontent.com/u/27071604?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/guillaume-be",
"html_url": "https://github.com/guillaume-be",
"followers_url": "https://api.github.c... | [] | closed | false | null | [] | [
"Yes, this one was also mentioned in https://github.com/huggingface/pytorch-transformers/issues/810#issuecomment-512991164.\r\n\r\nIt is fixed now."
] | 1,563 | 1,563 | 1,563 | CONTRIBUTOR | null | Hello,
In your example for GLUE you set the CLS segment id token to 1 for BERT: https://github.com/huggingface/pytorch-transformers/blob/2f869dc6651f9cf9253f4c5a43279027a0eccfc5/examples/run_glue.py#L259
Reading the original reference implementation it seems that CLS should have a segment_id=0. This is also align... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/858/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/858/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/857 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/857/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/857/comments | https://api.github.com/repos/huggingface/transformers/issues/857/events | https://github.com/huggingface/transformers/issues/857 | 471,003,087 | MDU6SXNzdWU0NzEwMDMwODc= | 857 | XLMForMaskedLM | {
"login": "Shuailong",
"id": 1918038,
"node_id": "MDQ6VXNlcjE5MTgwMzg=",
"avatar_url": "https://avatars.githubusercontent.com/u/1918038?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Shuailong",
"html_url": "https://github.com/Shuailong",
"followers_url": "https://api.github.com/users/Sh... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"You can do that using `XLNetLMHeadModel` and custom masks as shown in the [`run_generation` example](https://github.com/huggingface/pytorch-transformers/blob/master/examples/run_generation.py#L115-L121).\r\n\r\nBut note that XLNet is rather bad on short text input completions as I discussed in https://github.com/h... | 1,563 | 1,569 | 1,569 | NONE | null | Hi, I am currently training a BERT model using facebook XLM framework. I use the script in this repo to convert XLM format to PyTorch format. Is it possible to implement an `XLMForMaskedLM` which is just like `BertForMaskedLM` but use XLM trained BERT instead? | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/857/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/857/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/856 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/856/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/856/comments | https://api.github.com/repos/huggingface/transformers/issues/856/events | https://github.com/huggingface/transformers/issues/856 | 470,899,951 | MDU6SXNzdWU0NzA4OTk5NTE= | 856 | manually download models | {
"login": "Arvedek",
"id": 18126379,
"node_id": "MDQ6VXNlcjE4MTI2Mzc5",
"avatar_url": "https://avatars.githubusercontent.com/u/18126379?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Arvedek",
"html_url": "https://github.com/Arvedek",
"followers_url": "https://api.github.com/users/Arvede... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"If you don't want/cannot to use the built-in download/caching method, you can download both files manually, save them in a directory and rename them respectively `config.json` and `pytorch_model.bin`\r\n\r\nThen you can load the model using `model = BertModel.from_pretrained('path/to/your/directory')`",
"What if... | 1,563 | 1,638 | 1,570 | NONE | null | ERROR:pytorch_transformers.modeling_utils:Couldn't reach server at 'https://s3.amazonaws.com/models.huggingface.co/bert/bert-base-uncased-config.json' to download pretrained model configuration file.
ERROR:pytorch_transformers.modeling_utils:Couldn't reach server at 'https://s3.amazonaws.com/models.huggingface.co/bert... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/856/reactions",
"total_count": 9,
"+1": 8,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 1
} | https://api.github.com/repos/huggingface/transformers/issues/856/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/855 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/855/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/855/comments | https://api.github.com/repos/huggingface/transformers/issues/855/events | https://github.com/huggingface/transformers/issues/855 | 470,892,081 | MDU6SXNzdWU0NzA4OTIwODE= | 855 | modeling_xlnet.py lines 798 torch.eisum('i,d->id', pos_seq, inv_freq) | {
"login": "lcy081099",
"id": 15140404,
"node_id": "MDQ6VXNlcjE1MTQwNDA0",
"avatar_url": "https://avatars.githubusercontent.com/u/15140404?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/lcy081099",
"html_url": "https://github.com/lcy081099",
"followers_url": "https://api.github.com/users/... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"I don't understand your question.\r\n\r\nCan you give more details and point to the exact code lines you are referring to?\r\n\r\nYou cannot provide specific position indices to XLNet if that's what you are trying to do. You have to use the built-in relative embeddings.",
"This issue has been automatically marke... | 1,563 | 1,575 | 1,575 | NONE | null | hi I run position embedding in modeling_xlnet.py , but it not work , why not torch.eisum('i,d->id', [pos_seq, inv_freq]) ?i use pytorch 0.4.1 | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/855/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/855/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/854 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/854/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/854/comments | https://api.github.com/repos/huggingface/transformers/issues/854/events | https://github.com/huggingface/transformers/issues/854 | 470,886,856 | MDU6SXNzdWU0NzA4ODY4NTY= | 854 | Get the different result at BertModel | {
"login": "jackhsu2005",
"id": 49636419,
"node_id": "MDQ6VXNlcjQ5NjM2NDE5",
"avatar_url": "https://avatars.githubusercontent.com/u/49636419?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/jackhsu2005",
"html_url": "https://github.com/jackhsu2005",
"followers_url": "https://api.github.com/... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"Have you read in detail the [migration guide](https://github.com/huggingface/pytorch-transformers#migrating-from-pytorch-pretrained-bert-to-pytorch-transformers) of the readme?\r\n\r\nThere is also a new `run_glue` example which is an updated version of the previous `run_classifier` and that you can use as a start... | 1,563 | 1,569 | 1,569 | NONE | null | At old version pytorch-pretrained-bert :
I used the BertModel to fine-tuned, loss will decrease.
But I used the New version BertModel to use the same data to finetune, but loss won't decrease.
> optimzer
>I have tried different optimzer AdamW ,BertAdam.
> learning rate
>0.1 0.01 0.001 ... 0.0000000001
... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/854/reactions",
"total_count": 1,
"+1": 1,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/854/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/853 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/853/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/853/comments | https://api.github.com/repos/huggingface/transformers/issues/853/events | https://github.com/huggingface/transformers/issues/853 | 470,881,273 | MDU6SXNzdWU0NzA4ODEyNzM= | 853 | Error loading converted pytorch checkpoint | {
"login": "hguan6",
"id": 19914123,
"node_id": "MDQ6VXNlcjE5OTE0MTIz",
"avatar_url": "https://avatars.githubusercontent.com/u/19914123?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/hguan6",
"html_url": "https://github.com/hguan6",
"followers_url": "https://api.github.com/users/hguan6/fo... | [] | closed | false | null | [] | [
"Do you have a file named `biobert_v1.1_pubmed/config.json` as mentioned in the error?",
"Oh thanks. There is a config file name \"bert_config.json\" in the directory \"biobert_v1.1_pubmed/\". I changed the file name to \"config.json\" and it works! \r\nI am wondering why the pytorch-pretrained-bert can load the ... | 1,563 | 1,563 | 1,563 | NONE | null | I am using BioBert. After converting the tensorflow checkpoint to pytorch checkpoint, I want to load it to Bert model. I found that the old pytorch-pretrained-bert works perfect but the new pytorch-transformer fails.
Here is the successful run using pytorch-pretrained-bert:
> from pytorch_pretrained_bert import Ber... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/853/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/853/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/852 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/852/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/852/comments | https://api.github.com/repos/huggingface/transformers/issues/852/events | https://github.com/huggingface/transformers/issues/852 | 470,868,437 | MDU6SXNzdWU0NzA4Njg0Mzc= | 852 | UserWarning: Was asked to gather along dimension 0, but all input tensors were scalars; will instead unsqueeze and return a vector. | {
"login": "g-jing",
"id": 44223191,
"node_id": "MDQ6VXNlcjQ0MjIzMTkx",
"avatar_url": "https://avatars.githubusercontent.com/u/44223191?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/g-jing",
"html_url": "https://github.com/g-jing",
"followers_url": "https://api.github.com/users/g-jing/fo... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"It should be fine. There are probably your output losses.",
"@thomwolf Thanks for your reply. Could you explain more about why this happens ? I am still confused though",
"Maybe it is caused by calculating the loss in the model's forward function.\r\n",
"This issue has been automatically marked as stale bec... | 1,563 | 1,705 | 1,570 | NONE | null | When I finetune Bert with simple_lm_finetuning.py, there seems an error:
"UserWarning: Was asked to gather along dimension 0, but all input tensors were scalars; will instead unsqueeze and return a vector."
Will it influence the performance of the finetuning process ? Thanks in advance for any suggestion. | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/852/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/852/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/851 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/851/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/851/comments | https://api.github.com/repos/huggingface/transformers/issues/851/events | https://github.com/huggingface/transformers/issues/851 | 470,859,126 | MDU6SXNzdWU0NzA4NTkxMjY= | 851 | problem when calling resize_token_embeddings | {
"login": "LeeJuly30",
"id": 31088072,
"node_id": "MDQ6VXNlcjMxMDg4MDcy",
"avatar_url": "https://avatars.githubusercontent.com/u/31088072?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/LeeJuly30",
"html_url": "https://github.com/LeeJuly30",
"followers_url": "https://api.github.com/users/... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"Which model were you resizing?",
"I'm working on chinese BertForPreTraining model",
"This is strange because Bert's LM head has no bias...\r\n\r\nWould need to have a more complete error message to be able to understand.",
"```python\r\nclass BertLMPredictionHead(nn.Module):\r\n def __init__(self, config)... | 1,563 | 1,570 | 1,570 | NONE | null | When calling resize_token_embeddings, the model actually only modifies its embedding and decoder weight, whlie the decoder bias is unchanged. So whenever the forward function is called, the following error will be raised.
```python
RuntimeError: The size of tensor a (21215) must match the size of tensor b (21128) at ... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/851/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/851/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/850 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/850/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/850/comments | https://api.github.com/repos/huggingface/transformers/issues/850/events | https://github.com/huggingface/transformers/issues/850 | 470,786,969 | MDU6SXNzdWU0NzA3ODY5Njk= | 850 | Confused about the prune heads operation. | {
"login": "sjcfr",
"id": 34537582,
"node_id": "MDQ6VXNlcjM0NTM3NTgy",
"avatar_url": "https://avatars.githubusercontent.com/u/34537582?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/sjcfr",
"html_url": "https://github.com/sjcfr",
"followers_url": "https://api.github.com/users/sjcfr/follow... | [] | closed | false | null | [] | [
"Yes, I'll add a detailed example for this method in the coming weeks (update of the bertology script).\r\n\r\nThis can be used to remove heads in the model following the work of [Michel et al. (Are Sixteen Heads Really Better than One?)](http://arxiv.org/abs/1905.10650) among others.",
"Thanks a lot！",
"Hi @th... | 1,563 | 1,580 | 1,563 | NONE | null | In codes there are a 'prune_heads' method for the 'BertAttention' class, which refers to the 'prune_linear_layer' operation. Not understanding the meaning of such operation. The codes of 'prune_linear_layer' is listed below. Thanks for any help!
def prune_linear_layer(layer, index, dim=0):
""" Prune a linear... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/850/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/850/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/849 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/849/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/849/comments | https://api.github.com/repos/huggingface/transformers/issues/849/events | https://github.com/huggingface/transformers/issues/849 | 470,782,782 | MDU6SXNzdWU0NzA3ODI3ODI= | 849 | can't find utils_glue | {
"login": "YosiMass",
"id": 6850963,
"node_id": "MDQ6VXNlcjY4NTA5NjM=",
"avatar_url": "https://avatars.githubusercontent.com/u/6850963?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/YosiMass",
"html_url": "https://github.com/YosiMass",
"followers_url": "https://api.github.com/users/YosiM... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"It's hard to tell what could be the source of the error without describing with some minimum details what you tried to do/run and which version of the code you are running...\r\n\r\n_If_ you are running `run_glue.py` in the examples, make sure that [`utils_glue.py`](https://github.com/huggingface/pytorch-transform... | 1,563 | 1,569 | 1,569 | NONE | null | import fails | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/849/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/849/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/848 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/848/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/848/comments | https://api.github.com/repos/huggingface/transformers/issues/848/events | https://github.com/huggingface/transformers/issues/848 | 470,781,190 | MDU6SXNzdWU0NzA3ODExOTA= | 848 | adaptive softmax in transformer-xl | {
"login": "lepus2",
"id": 17767202,
"node_id": "MDQ6VXNlcjE3NzY3MjAy",
"avatar_url": "https://avatars.githubusercontent.com/u/17767202?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/lepus2",
"html_url": "https://github.com/lepus2",
"followers_url": "https://api.github.com/users/lepus2/fo... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"Yes. At the moment, this library is designed for loading pretrained models mostly and no one has open-sourced a Transformer-XL pretrained model using something else than adaptive softmax so I have not spent time adding these options.\r\n\r\nHappy to welcome PR though.\r\n\r\nThe main thing of interest here, if you... | 1,563 | 1,569 | 1,569 | NONE | null | I guess there is some incomplete part for adaptive softmax in [modeling_transfo_xl.py](https://github.com/huggingface/pytorch-transformers/blob/master/pytorch_transformers/modeling_transfo_xl.py)
Actually, it is impossible to build model not using adaptive softmax even though `TransfoXLConfig` has `adaptive` parameter... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/848/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/848/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/847 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/847/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/847/comments | https://api.github.com/repos/huggingface/transformers/issues/847/events | https://github.com/huggingface/transformers/pull/847 | 470,777,251 | MDExOlB1bGxSZXF1ZXN0Mjk5NjMzMjkw | 847 | typos | {
"login": "lpq29743",
"id": 12952648,
"node_id": "MDQ6VXNlcjEyOTUyNjQ4",
"avatar_url": "https://avatars.githubusercontent.com/u/12952648?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/lpq29743",
"html_url": "https://github.com/lpq29743",
"followers_url": "https://api.github.com/users/lpq... | [] | closed | false | null | [] | [
"# [Codecov](https://codecov.io/gh/huggingface/pytorch-transformers/pull/847?src=pr&el=h1) Report\n> Merging [#847](https://codecov.io/gh/huggingface/pytorch-transformers/pull/847?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/pytorch-transformers/commit/a615499076a67dceb8907ecdf8eadaff04bb8d6a?src... | 1,563 | 1,563 | 1,563 | CONTRIBUTOR | null | "ouputs" -> "outputs" | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/847/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/847/timeline | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/847",
"html_url": "https://github.com/huggingface/transformers/pull/847",
"diff_url": "https://github.com/huggingface/transformers/pull/847.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/847.patch",
"merged_at": 15638... |
https://api.github.com/repos/huggingface/transformers/issues/846 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/846/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/846/comments | https://api.github.com/repos/huggingface/transformers/issues/846/events | https://github.com/huggingface/transformers/issues/846 | 470,724,022 | MDU6SXNzdWU0NzA3MjQwMjI= | 846 | XLNET completely wrong and random output | {
"login": "Oxi84",
"id": 25420033,
"node_id": "MDQ6VXNlcjI1NDIwMDMz",
"avatar_url": "https://avatars.githubusercontent.com/u/25420033?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Oxi84",
"html_url": "https://github.com/Oxi84",
"followers_url": "https://api.github.com/users/Oxi84/follow... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"I solved one of the problems, using another way to load the model described bellow, but still it works way worse than BERT.\r\n \r\n tokenizer = XLNetTokenizer.from_pretrained(\"xlnet-large-cased\")\r\n model = XLNetLMHeadModel.from_pretrained(\"xlnet-large-cased\")\r\n model.eval()\r\n if torch.cud... | 1,563 | 1,581 | 1,581 | NONE | null | I followed the example here: https://huggingface.co/pytorch-transformers/model_doc/xlnet.html#pytorch_transformers.XLNetModel
I found that I get completelly wrong output, I mean predicted word for the masked sentences are completelly irelevant and they change each run. I guess there is some bug, culd you please take... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/846/reactions",
"total_count": 1,
"+1": 1,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/846/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/845 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/845/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/845/comments | https://api.github.com/repos/huggingface/transformers/issues/845/events | https://github.com/huggingface/transformers/pull/845 | 470,661,645 | MDExOlB1bGxSZXF1ZXN0Mjk5NTU1MDMy | 845 | fixed version issues in run_openai_gpt | {
"login": "rabeehk",
"id": 6278280,
"node_id": "MDQ6VXNlcjYyNzgyODA=",
"avatar_url": "https://avatars.githubusercontent.com/u/6278280?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/rabeehk",
"html_url": "https://github.com/rabeehk",
"followers_url": "https://api.github.com/users/rabeehk/... | [] | closed | false | null | [] | [
"# [Codecov](https://codecov.io/gh/huggingface/pytorch-transformers/pull/845?src=pr&el=h1) Report\n> Merging [#845](https://codecov.io/gh/huggingface/pytorch-transformers/pull/845?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/pytorch-transformers/commit/a615499076a67dceb8907ecdf8eadaff04bb8d6a?src... | 1,563 | 1,563 | 1,563 | NONE | null | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/845/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/845/timeline | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/845",
"html_url": "https://github.com/huggingface/transformers/pull/845",
"diff_url": "https://github.com/huggingface/transformers/pull/845.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/845.patch",
"merged_at": 15638... | |
https://api.github.com/repos/huggingface/transformers/issues/844 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/844/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/844/comments | https://api.github.com/repos/huggingface/transformers/issues/844/events | https://github.com/huggingface/transformers/pull/844 | 470,652,537 | MDExOlB1bGxSZXF1ZXN0Mjk5NTQ4OTA3 | 844 | Fixed typo | {
"login": "rish-16",
"id": 20137995,
"node_id": "MDQ6VXNlcjIwMTM3OTk1",
"avatar_url": "https://avatars.githubusercontent.com/u/20137995?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/rish-16",
"html_url": "https://github.com/rish-16",
"followers_url": "https://api.github.com/users/rish-1... | [] | closed | false | null | [] | [
"# [Codecov](https://codecov.io/gh/huggingface/pytorch-transformers/pull/844?src=pr&el=h1) Report\n> Merging [#844](https://codecov.io/gh/huggingface/pytorch-transformers/pull/844?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/pytorch-transformers/commit/a615499076a67dceb8907ecdf8eadaff04bb8d6a?src... | 1,563 | 1,563 | 1,563 | CONTRIBUTOR | null | Fixed typo in README.md | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/844/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/844/timeline | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/844",
"html_url": "https://github.com/huggingface/transformers/pull/844",
"diff_url": "https://github.com/huggingface/transformers/pull/844.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/844.patch",
"merged_at": 15637... |
https://api.github.com/repos/huggingface/transformers/issues/843 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/843/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/843/comments | https://api.github.com/repos/huggingface/transformers/issues/843/events | https://github.com/huggingface/transformers/issues/843 | 470,631,241 | MDU6SXNzdWU0NzA2MzEyNDE= | 843 | Issue | {
"login": "bmanishreddy",
"id": 15935444,
"node_id": "MDQ6VXNlcjE1OTM1NDQ0",
"avatar_url": "https://avatars.githubusercontent.com/u/15935444?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/bmanishreddy",
"html_url": "https://github.com/bmanishreddy",
"followers_url": "https://api.github.c... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"@bmanishreddy Your issue contains no text, should it be closed?",
"Yeah .. my bad it can be closed ",
"No worries, please close it so it doesn't create clutter. Thanks!",
"This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occur... | 1,563 | 1,568 | 1,563 | NONE | null | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/843/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/843/timeline | completed | null | null | |
https://api.github.com/repos/huggingface/transformers/issues/842 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/842/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/842/comments | https://api.github.com/repos/huggingface/transformers/issues/842/events | https://github.com/huggingface/transformers/issues/842 | 470,620,212 | MDU6SXNzdWU0NzA2MjAyMTI= | 842 | 16 GB dataset for finetuning fail on reduce_memory | {
"login": "g-jing",
"id": 44223191,
"node_id": "MDQ6VXNlcjQ0MjIzMTkx",
"avatar_url": "https://avatars.githubusercontent.com/u/44223191?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/g-jing",
"html_url": "https://github.com/g-jing",
"followers_url": "https://api.github.com/users/g-jing/fo... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | 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"
] | 1,563 | 1,569 | 1,569 | NONE | null | Hi, I am using 16GB dataset to finetune bert Model. When I do not use reduce_memory, which is loading dataset into memory first, it will use all of my 120GB memory and then crush because of out of memory. Now I am using reduce_memory model, with the increase of loading lines, the memory use is still increasing. But it ... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/842/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/842/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/841 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/841/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/841/comments | https://api.github.com/repos/huggingface/transformers/issues/841/events | https://github.com/huggingface/transformers/issues/841 | 470,613,390 | MDU6SXNzdWU0NzA2MTMzOTA= | 841 | Detaching Variables | {
"login": "rishibommasani",
"id": 47439426,
"node_id": "MDQ6VXNlcjQ3NDM5NDI2",
"avatar_url": "https://avatars.githubusercontent.com/u/47439426?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/rishibommasani",
"html_url": "https://github.com/rishibommasani",
"followers_url": "https://api.gi... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"Maybe you are not using `with torch.grad()` when calling the model for inference?\r\n\r\nI've added that in the readme example (it used to be mentioned there indeed).",
"This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Than... | 1,563 | 1,569 | 1,569 | NONE | null | Something I noticed in transitioning from Pretrained-BERT to Transformers is that for the purposes of using BERT as a feature extractor/probing the pretrained representations, I need to detach variables whereas I previously didn't. I am not sure if this is noted somewhere (I didn't see it in the section in the docs abo... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/841/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/841/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/840 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/840/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/840/comments | https://api.github.com/repos/huggingface/transformers/issues/840/events | https://github.com/huggingface/transformers/issues/840 | 470,552,761 | MDU6SXNzdWU0NzA1NTI3NjE= | 840 | AttributeError: 'BertModel' object has no attribute '_load_from_state_dict' | {
"login": "Shandilya21",
"id": 28632968,
"node_id": "MDQ6VXNlcjI4NjMyOTY4",
"avatar_url": "https://avatars.githubusercontent.com/u/28632968?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Shandilya21",
"html_url": "https://github.com/Shandilya21",
"followers_url": "https://api.github.com/... | [] | closed | false | null | [] | [
"You are probably not using the new release of PyTorch-transformers.\r\nTry `pip install pytorch-transformers --upgrade`.\r\nAnd read the full [readme](https://github.com/huggingface/pytorch-transformers), there are several breaking changes.",
"@thomwolf Thanks for your update, it works for me !!"
] | 1,563 | 1,564 | 1,564 | NONE | null | Hi,
I am getting these error, even the i tried the model straight from the repository examples for the test cases.? anyone help me to understand this issues
Thanks for help in advance
| {
"url": "https://api.github.com/repos/huggingface/transformers/issues/840/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/840/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/839 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/839/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/839/comments | https://api.github.com/repos/huggingface/transformers/issues/839/events | https://github.com/huggingface/transformers/issues/839 | 470,515,113 | MDU6SXNzdWU0NzA1MTUxMTM= | 839 | How to restore a training? | {
"login": "bingyupiaoyao",
"id": 1625353,
"node_id": "MDQ6VXNlcjE2MjUzNTM=",
"avatar_url": "https://avatars.githubusercontent.com/u/1625353?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/bingyupiaoyao",
"html_url": "https://github.com/bingyupiaoyao",
"followers_url": "https://api.github.... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"You will have to modify the provided example to save/reload the model, optimizer and scheduler states.\r\n\r\nI've updated the scheduler classes in #872 so that we can save/reload the schedulers with the standard PyTorch serialization practice:\r\n```\r\ntorch.save(schedule.state_dict(), FILE_NAME) # sav... | 1,563 | 1,569 | 1,569 | NONE | null | For example, I use "run_glue.py" to train a model and stop at Epoch 30, and how to restore the training process from Epoch 30? | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/839/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/839/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/838 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/838/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/838/comments | https://api.github.com/repos/huggingface/transformers/issues/838/events | https://github.com/huggingface/transformers/issues/838 | 470,372,340 | MDU6SXNzdWU0NzAzNzIzNDA= | 838 | Standardized head for Question Answering | {
"login": "andrelmfarias",
"id": 43521764,
"node_id": "MDQ6VXNlcjQzNTIxNzY0",
"avatar_url": "https://avatars.githubusercontent.com/u/43521764?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/andrelmfarias",
"html_url": "https://github.com/andrelmfarias",
"followers_url": "https://api.githu... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"Well, the output of the `BertForQuestionAnswering` model hasn't changed, the returned `start_score` and `end_score` are still scores before the softmax.\r\n\r\nCan you point more specifically to the changes you are referring to?",
"I'm sorry. Indeed the implementation of `BertForQuestionAnswering` remains the sa... | 1,563 | 1,574 | 1,574 | NONE | null | Hi,
With some colleagues, we developed a QA system that uses a whole QA pipeline (Retriever, Reader, Ranker). We use your older version of `BertForQuestionAnswering` as Reader and now we wish to update it to be compatible with your new release and to add others models as well (XLNet, XLM).
Our system uses the logit... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/838/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/838/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/837 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/837/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/837/comments | https://api.github.com/repos/huggingface/transformers/issues/837/events | https://github.com/huggingface/transformers/issues/837 | 470,348,073 | MDU6SXNzdWU0NzAzNDgwNzM= | 837 | run_openai_gpt.py issues with Adamw | {
"login": "rabeehk",
"id": 6278280,
"node_id": "MDQ6VXNlcjYyNzgyODA=",
"avatar_url": "https://avatars.githubusercontent.com/u/6278280?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/rabeehk",
"html_url": "https://github.com/rabeehk",
"followers_url": "https://api.github.com/users/rabeehk/... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"Yes, so this should be fixed by (your own :) PR #845 Thanks again!",
"thanks :)\nBest regards,\nRabeeh\n\nOn Tue, Jul 23, 2019 at 3:30 PM Thomas Wolf <notifications@github.com>\nwrote:\n\n> Yes, so this should be fixed by (your own :) PR #845\n> <https://github.com/huggingface/pytorch-transformers/pull/845> Than... | 1,563 | 1,569 | 1,569 | NONE | null | Hi
Adamw in this script has parameters not existing anymore, ...
Thanks for updates in advance. | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/837/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/837/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/836 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/836/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/836/comments | https://api.github.com/repos/huggingface/transformers/issues/836/events | https://github.com/huggingface/transformers/issues/836 | 470,344,188 | MDU6SXNzdWU0NzAzNDQxODg= | 836 | BertForNextSentencePrediction labels | {
"login": "ArthurCamara",
"id": 709027,
"node_id": "MDQ6VXNlcjcwOTAyNw==",
"avatar_url": "https://avatars.githubusercontent.com/u/709027?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ArthurCamara",
"html_url": "https://github.com/ArthurCamara",
"followers_url": "https://api.github.com/u... | [] | closed | false | null | [] | [
"You should supply your own labels when using the `BertForSequenceClassification` class (`labels` input to the forward method). You can choose the labels you like.\r\n\r\nThe `BertForSequenceClassification` class is **not** related to the Next Sentence Classification task used during Bert pretraining. You can use t... | 1,563 | 1,565 | 1,565 | NONE | null | Hi everyone!
I was reading trough the documentation, and, according to https://huggingface.co/pytorch-transformers/model_doc/bert.html#bertforsequenceclassification, it expects that `next_sentence_label` is `1` if B is **not** a next sequence for A, and `0` if B **is** a sequence for B.
That's somewhat counterintuit... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/836/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/836/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/835 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/835/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/835/comments | https://api.github.com/repos/huggingface/transformers/issues/835/events | https://github.com/huggingface/transformers/issues/835 | 470,177,370 | MDU6SXNzdWU0NzAxNzczNzA= | 835 | How to use the pretrain script with only token classification task ? | {
"login": "g-jing",
"id": 44223191,
"node_id": "MDQ6VXNlcjQ0MjIzMTkx",
"avatar_url": "https://avatars.githubusercontent.com/u/44223191?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/g-jing",
"html_url": "https://github.com/g-jing",
"followers_url": "https://api.github.com/users/g-jing/fo... | [] | closed | false | null | [] | [
"No, you probably need to adapt the example script to your exact task."
] | 1,563 | 1,565 | 1,565 | NONE | null | Hi, I need to train on my own twitter corpus, but most of the twitter contains only one sentence. Therefore I can not use the sentence prediction task to train the model. Will the script automatically use only token classification task when there is no next sentence ? Thanks in advance. | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/835/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/835/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/834 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/834/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/834/comments | https://api.github.com/repos/huggingface/transformers/issues/834/events | https://github.com/huggingface/transformers/issues/834 | 470,154,199 | MDU6SXNzdWU0NzAxNTQxOTk= | 834 | git pull pytorch-transformers?? | {
"login": "Linohong",
"id": 19821168,
"node_id": "MDQ6VXNlcjE5ODIxMTY4",
"avatar_url": "https://avatars.githubusercontent.com/u/19821168?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Linohong",
"html_url": "https://github.com/Linohong",
"followers_url": "https://api.github.com/users/Lin... | [] | closed | false | null | [] | [
"I suspect this is more of a general `git` question. We'll close this unless there is something specific to the lib.\r\n\r\nGood luck!",
"@julien-c ,\r\nyeah, I had some of code mismatch issues and that falls into general github matters \r\nand I got it done :) \r\n\r\nThank you :)"
] | 1,563 | 1,564 | 1,563 | NONE | null | Hello,
I have git cloned 'pytorch-pretrained-bert' before there is a new release, pytorch-transformers
and I added many of the comments and new example files in the cloned project.
However, when I found there has been a new version released, git pulling didn't work
for conflicting files issues.
Is it because o... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/834/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/834/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/833 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/833/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/833/comments | https://api.github.com/repos/huggingface/transformers/issues/833/events | https://github.com/huggingface/transformers/issues/833 | 470,152,469 | MDU6SXNzdWU0NzAxNTI0Njk= | 833 | missing 1 required positional argument: 'num_classes' in 'from_pretrained' | {
"login": "desireevl",
"id": 17139032,
"node_id": "MDQ6VXNlcjE3MTM5MDMy",
"avatar_url": "https://avatars.githubusercontent.com/u/17139032?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/desireevl",
"html_url": "https://github.com/desireevl",
"followers_url": "https://api.github.com/users/... | [] | closed | false | null | [] | [
"Yes, @xanlsh is working on an update to reduce the effect of this breaking change in #866.\r\n\r\nYou should be able to keep your script unchanged.",
"@desireevl Since #866 has been merged, your code should work now",
"Thanks!"
] | 1,563 | 1,565 | 1,565 | CONTRIBUTOR | null | I am running a [multiclass BERT classification](https://github.com/desireevl/Bert-Multi-Label-Text-Classification/blob/master/train_bert_multi_label.py) model and am receiving the following error:
`
Traceback (most recent call last):
File "train_bert_multi_label.py", line 144, in <module>
main()
File "tr... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/833/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/833/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/832 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/832/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/832/comments | https://api.github.com/repos/huggingface/transformers/issues/832/events | https://github.com/huggingface/transformers/issues/832 | 470,106,701 | MDU6SXNzdWU0NzAxMDY3MDE= | 832 | Training with wrong GPU count | {
"login": "hairzooc",
"id": 13031514,
"node_id": "MDQ6VXNlcjEzMDMxNTE0",
"avatar_url": "https://avatars.githubusercontent.com/u/13031514?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/hairzooc",
"html_url": "https://github.com/hairzooc",
"followers_url": "https://api.github.com/users/hai... | [] | closed | false | null | [] | [
"Yes, this is expected behavior. Each script in distributed training has ownership over one GPU only.\r\n\r\nYou can read this blog post for details on parallel and distributed training: https://medium.com/huggingface/training-larger-batches-practical-tips-on-1-gpu-multi-gpu-distributed-setups-ec88c3e51255"
] | 1,563 | 1,565 | 1,565 | NONE | null | Hi,
Thank you for your repo :)
I'm fine-tuning with 4 GPU (run_squad, bert model)
And I found that gpu count is wrong when to do distributed training.
I've got 1 GPU count and that's caused by source code below
Is there any reason to set n_gpu = 1 when to do distributed training?
if args.local_rank == -1 or ... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/832/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/832/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/831 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/831/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/831/comments | https://api.github.com/repos/huggingface/transformers/issues/831/events | https://github.com/huggingface/transformers/issues/831 | 470,075,427 | MDU6SXNzdWU0NzAwNzU0Mjc= | 831 | finetune_on_pregenerate Loss.backwards() throw an error | {
"login": "sophiapeng0426",
"id": 11810773,
"node_id": "MDQ6VXNlcjExODEwNzcz",
"avatar_url": "https://avatars.githubusercontent.com/u/11810773?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/sophiapeng0426",
"html_url": "https://github.com/sophiapeng0426",
"followers_url": "https://api.gi... | [] | closed | false | null | [] | [
"Yes this example should have been updated now with #797."
] | 1,563 | 1,563 | 1,563 | NONE | null | In finetune_on_pregenerated.py, loss are tuples and thus loss.backward() is not going to work.
Original:
loss = model(input_ids, segment_ids, input_mask, lm_label_ids, is_next)
Update:
loss, _ , _ = model(input_ids, segment_ids, input_mask, lm_label_ids, is_next)
Is this fix correct? | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/831/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/831/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/830 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/830/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/830/comments | https://api.github.com/repos/huggingface/transformers/issues/830/events | https://github.com/huggingface/transformers/issues/830 | 470,074,075 | MDU6SXNzdWU0NzAwNzQwNzU= | 830 | AdamW does not have args warmup and t_total | {
"login": "sophiapeng0426",
"id": 11810773,
"node_id": "MDQ6VXNlcjExODEwNzcz",
"avatar_url": "https://avatars.githubusercontent.com/u/11810773?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/sophiapeng0426",
"html_url": "https://github.com/sophiapeng0426",
"followers_url": "https://api.gi... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"Yes this example should have been updated now by #797.\r\n\r\nRegarding `AdamW` and the schedule, details, and examples for the conversion are indicated in the migration section of the readme: https://github.com/huggingface/pytorch-transformers#Migrating-from-pytorch-pretrained-bert-to-pytorch-transformers",
"Th... | 1,563 | 1,569 | 1,569 | NONE | null | In finetune_on_pregenerated.py, below code throw error cause AdamW does not have those two arguments. This can be fixed by comment out those two columns but not sure if that means warmup will be not effective after that?
optimizer = AdamW(optimizer_grouped_parameters,
lr=args.learning_r... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/830/reactions",
"total_count": 5,
"+1": 5,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/830/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/829 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/829/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/829/comments | https://api.github.com/repos/huggingface/transformers/issues/829/events | https://github.com/huggingface/transformers/issues/829 | 470,012,887 | MDU6SXNzdWU0NzAwMTI4ODc= | 829 | RoBERTa support | {
"login": "sleepinyourhat",
"id": 1284441,
"node_id": "MDQ6VXNlcjEyODQ0NDE=",
"avatar_url": "https://avatars.githubusercontent.com/u/1284441?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/sleepinyourhat",
"html_url": "https://github.com/sleepinyourhat",
"followers_url": "https://api.gith... | [] | closed | false | null | [] | [
"Working on the code/paper release as we speak :) It largely follows the existing masked_lm implementation in fairseq. Happy to help get this integrated here.",
"Hi @myleott great news :) I'm really excited about the release 🤗 I've some questions: do you plan to perform any comparisons between RoBERTa and BERT o... | 1,563 | 1,575 | 1,575 | NONE | null | https://twitter.com/sleepinyourhat/status/1151940994688016384
The code/parameters aren't out yet, but I figure it couldn't hurt to put in an obnoxious feature request now! | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/829/reactions",
"total_count": 29,
"+1": 10,
"-1": 0,
"laugh": 12,
"hooray": 0,
"confused": 0,
"heart": 7,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/829/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/828 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/828/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/828/comments | https://api.github.com/repos/huggingface/transformers/issues/828/events | https://github.com/huggingface/transformers/issues/828 | 469,978,335 | MDU6SXNzdWU0Njk5NzgzMzU= | 828 | CUDA error: invalid configuration argument when not using DataParallel | {
"login": "Phirefly9",
"id": 16687050,
"node_id": "MDQ6VXNlcjE2Njg3MDUw",
"avatar_url": "https://avatars.githubusercontent.com/u/16687050?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Phirefly9",
"html_url": "https://github.com/Phirefly9",
"followers_url": "https://api.github.com/users/... | [] | closed | false | null | [] | [
"Further testing showed this was caused by the batch size being too high and the card running out of memory, and providing a misleading error."
] | 1,563 | 1,563 | 1,563 | NONE | null | Good Evening,
We have a DGX2 system running the latest Nvidia pytorch docker container - 19.06. When attempting to use the gpt2 or gpt2-medium models to extract out embeddings we are getting the following error, but only when not using dataparallel: (note we are using apex here but an optimization level of 0, this ... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/828/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/828/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/827 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/827/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/827/comments | https://api.github.com/repos/huggingface/transformers/issues/827/events | https://github.com/huggingface/transformers/issues/827 | 469,947,330 | MDU6SXNzdWU0Njk5NDczMzA= | 827 | xlnet input_mask and attention_mask type error | {
"login": "Saner3",
"id": 30628796,
"node_id": "MDQ6VXNlcjMwNjI4Nzk2",
"avatar_url": "https://avatars.githubusercontent.com/u/30628796?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Saner3",
"html_url": "https://github.com/Saner3",
"followers_url": "https://api.github.com/users/Saner3/fo... | [] | closed | false | null | [] | [
"Humm you are right, the docstrings are off, it would be more clear if they were all indicated as `torch.FloatTensor` (even though officially torch.Tensor is an alias for the default tensor type (torch.FloatTensor))."
] | 1,563 | 1,563 | 1,563 | NONE | null | when I use:
```input_mask = (input_ids == 0)```
```perm_mask = perm_mask = torch.zeros((1, input_ids.shape[1], input_ids.shape[1]), dtype=torch.float, device=device)```
```perm_mask[:, :, -1] = 1.0 # Previous tokens don't see last token```
File "pytorch-transformers/pytorch_transformers/modeling_xlnet.py", lin... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/827/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/827/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/826 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/826/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/826/comments | https://api.github.com/repos/huggingface/transformers/issues/826/events | https://github.com/huggingface/transformers/issues/826 | 469,887,826 | MDU6SXNzdWU0Njk4ODc4MjY= | 826 | Providing older documentation | {
"login": "Ricocotam",
"id": 9447752,
"node_id": "MDQ6VXNlcjk0NDc3NTI=",
"avatar_url": "https://avatars.githubusercontent.com/u/9447752?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Ricocotam",
"html_url": "https://github.com/Ricocotam",
"followers_url": "https://api.github.com/users/Ri... | [] | closed | false | null | [] | [
"Hi, you may go to https://github.com/huggingface/pytorch-transformers/releases, select the release you are working with and in its \"Assets\" download the repo and navigate the code, together with documentation",
"Hi, here is the older documentation: https://github.com/huggingface/pytorch-transformers/tree/v0.6.... | 1,563 | 1,563 | 1,563 | NONE | null | Hey, would it be possible to release the previous documentation ? I'm working on previous version and can't find proper doc right now.
Thanks if you can help | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/826/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/826/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/825 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/825/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/825/comments | https://api.github.com/repos/huggingface/transformers/issues/825/events | https://github.com/huggingface/transformers/issues/825 | 469,877,928 | MDU6SXNzdWU0Njk4Nzc5Mjg= | 825 | Chinese BERT broken probably after `pytorch-transformer` release | {
"login": "leemengtw",
"id": 3454980,
"node_id": "MDQ6VXNlcjM0NTQ5ODA=",
"avatar_url": "https://avatars.githubusercontent.com/u/3454980?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/leemengtw",
"html_url": "https://github.com/leemengtw",
"followers_url": "https://api.github.com/users/le... | [] | closed | false | null | [] | [
"We did slightly change the way the tokenizer strip the spaces at the end of the words when loading the tokenizer, as discussed in issue #328, in particular here https://github.com/huggingface/pytorch-transformers/issues/328#issuecomment-503630929.\r\nNow, I'm not exactly sure what is the right solution for both ca... | 1,563 | 1,564 | 1,564 | NONE | null | I suspect that there is some recent code change that breaks the Chinese BERT.
I used the following PyTorch hub code to load the Chinese BERT tokenizer and print out some tokens in the vocab perhaps just a few days ago and everything was fine:
```python
import torch
GITHUB_REPO = "huggingface/pytorch-pretraine... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/825/reactions",
"total_count": 2,
"+1": 2,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/825/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/824 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/824/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/824/comments | https://api.github.com/repos/huggingface/transformers/issues/824/events | https://github.com/huggingface/transformers/issues/824 | 469,862,479 | MDU6SXNzdWU0Njk4NjI0Nzk= | 824 | Bertology example is probably broken | {
"login": "davidefiocco",
"id": 4547987,
"node_id": "MDQ6VXNlcjQ1NDc5ODc=",
"avatar_url": "https://avatars.githubusercontent.com/u/4547987?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/davidefiocco",
"html_url": "https://github.com/davidefiocco",
"followers_url": "https://api.github.com... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"Yes, this example is still work in progress. Hopefully, I can finish it before ACL (but 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"
] | 1,563 | 1,569 | 1,569 | CONTRIBUTOR | null | Hello!
I tried to run `run_bertology.py` in the example dir calling it with
```
export TASK_NAME=CoLA
python ./run_bertology.py --data_dir $GLUE_DIR/$TASK_NAME
--model_name bert-base-uncased
--task_name $TASK_NAME
... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/824/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/824/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/823 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/823/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/823/comments | https://api.github.com/repos/huggingface/transformers/issues/823/events | https://github.com/huggingface/transformers/issues/823 | 469,832,638 | MDU6SXNzdWU0Njk4MzI2Mzg= | 823 | Updating simple_lm_finetuning.py for FP16 training | {
"login": "crowegian",
"id": 14296792,
"node_id": "MDQ6VXNlcjE0Mjk2Nzky",
"avatar_url": "https://avatars.githubusercontent.com/u/14296792?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/crowegian",
"html_url": "https://github.com/crowegian",
"followers_url": "https://api.github.com/users/... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"Hi, has this been fixed? I've tried updating my language modeling script to match but still getting errors.",
"Having the same problem at the moment ...",
"I guess the preferred way is to use `apex.amp` like in this example?\r\nhttps://github.com/huggingface/pytorch-transformers/blob/master/examples/run_glue.... | 1,563 | 1,570 | 1,570 | NONE | null | in simple_lm_finetuning the recent updated code doesn't work with the old optimizer specifications.
When not running with --fp16
`
optimizer = BertAdam(optimizer_grouped_parameters,
lr=args.learning_rate,
warmup=args.warmup_proportion,
t_total=num... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/823/reactions",
"total_count": 2,
"+1": 2,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/823/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/822 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/822/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/822/comments | https://api.github.com/repos/huggingface/transformers/issues/822/events | https://github.com/huggingface/transformers/issues/822 | 469,791,932 | MDU6SXNzdWU0Njk3OTE5MzI= | 822 | XLNet-large-cased on Squad 2.0: can't replicate results | {
"login": "avisil",
"id": 43005718,
"node_id": "MDQ6VXNlcjQzMDA1NzE4",
"avatar_url": "https://avatars.githubusercontent.com/u/43005718?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/avisil",
"html_url": "https://github.com/avisil",
"followers_url": "https://api.github.com/users/avisil/fo... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"This is similar to what the authors ran in the paper (except I could fit only this on 3 v100 GPUs):\r\n\r\n`python run_squad.py --do_lower_case --do_train --do_eval --train_file $SQUAD_DIR/train-v2.0.json --predict_file $SQUAD_DIR/dev-v2.0.json --output_dir $SQUAD_DIR/output --version_2_with_negative --model_name... | 1,563 | 1,579 | 1,572 | NONE | null | I've been trying to replicate the numbers in the Squad 2.0 dev set (F1=86) with this script and the XLnet embeddings. So far the results are really off..{Opening a new issue as the previous one seems dedicated to SST-2}
`python run_squad.py --do_lower_case --do_train --do_eval --train_file $SQUAD_DIR/train-v2.0.json... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/822/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/822/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/821 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/821/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/821/comments | https://api.github.com/repos/huggingface/transformers/issues/821/events | https://github.com/huggingface/transformers/issues/821 | 469,731,350 | MDU6SXNzdWU0Njk3MzEzNTA= | 821 | Couldn't reach server | {
"login": "rabeehk",
"id": 6278280,
"node_id": "MDQ6VXNlcjYyNzgyODA=",
"avatar_url": "https://avatars.githubusercontent.com/u/6278280?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/rabeehk",
"html_url": "https://github.com/rabeehk",
"followers_url": "https://api.github.com/users/rabeehk/... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"If I click on the links you provided above, they are currently reachable for me...\r\nIt may be a silly suggestions, but could it be that your internet connection was momentarily down when the code tried download those files or somehow you are not allowed to reach data on s3?",
"I have an idea about it. We can d... | 1,563 | 1,593 | 1,572 | NONE | null | Hi I am running the very first example in readme. I got these errors, thanks for your help
Couldn't reach server to download vocabulary.
Couldn't reach server at 'https://s3.amazonaws.com/models.huggingface.co/bert/bert-base-uncased-config.json' to download pretrained model configuration file.
Couldn't reach serve... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/821/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/821/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/820 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/820/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/820/comments | https://api.github.com/repos/huggingface/transformers/issues/820/events | https://github.com/huggingface/transformers/issues/820 | 469,692,641 | MDU6SXNzdWU0Njk2OTI2NDE= | 820 | RuntimeError: Creating MTGP constants failed | {
"login": "dkarmon",
"id": 669552,
"node_id": "MDQ6VXNlcjY2OTU1Mg==",
"avatar_url": "https://avatars.githubusercontent.com/u/669552?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/dkarmon",
"html_url": "https://github.com/dkarmon",
"followers_url": "https://api.github.com/users/dkarmon/fo... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"Not sure this comes from pytorch-transformers or CUDA, see: https://github.com/pytorch/pytorch/issues/20489",
"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"
] | 1,563 | 1,573 | 1,573 | NONE | null | Hi,
I successfully fine tuned a BertForTokenClassification model based on bert-base-cased in the past. However, I now encounter with an following error: (see full stack below)
```
RuntimeError: **Creating MTGP constants failed.** at /opt/conda/conda-bld/pytorch_1556653099582/work/aten/src/THC/THCTensorRandom.cu:33
... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/820/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/820/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/819 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/819/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/819/comments | https://api.github.com/repos/huggingface/transformers/issues/819/events | https://github.com/huggingface/transformers/issues/819 | 469,607,950 | MDU6SXNzdWU0Njk2MDc5NTA= | 819 | Output of BertModel does not match the last hidden layer from fixed feature vectors | {
"login": "sasaadi",
"id": 7882383,
"node_id": "MDQ6VXNlcjc4ODIzODM=",
"avatar_url": "https://avatars.githubusercontent.com/u/7882383?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/sasaadi",
"html_url": "https://github.com/sasaadi",
"followers_url": "https://api.github.com/users/sasaadi/... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"What is your exact command to _extract the last hidden layer (layer -1)_?\r\nAnd what is your exact command to get _the outputs[0] from pytorch_transformers.BertModel()_ ?",
"To extract the last hidden layer (layer -1) from BERT, I run the `extract_features.py` as follows:\r\n`python extract_features.py --inpu... | 1,563 | 1,572 | 1,572 | NONE | null | Based on BERT documentation (https://github.com/google-research/bert#using-bert-to-extract-fixed-feature-vectors-like-elmo) we can extract the contextualized token embeddings of each hidden layer separately. However, when I extract the last hidden layer (layer -1), it does not match the `outputs[0]` from `pytorch_trans... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/819/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/819/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/818 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/818/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/818/comments | https://api.github.com/repos/huggingface/transformers/issues/818/events | https://github.com/huggingface/transformers/issues/818 | 469,605,157 | MDU6SXNzdWU0Njk2MDUxNTc= | 818 | GPT sentence log loss: average or summed loss? | {
"login": "jhlau",
"id": 4261132,
"node_id": "MDQ6VXNlcjQyNjExMzI=",
"avatar_url": "https://avatars.githubusercontent.com/u/4261132?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/jhlau",
"html_url": "https://github.com/jhlau",
"followers_url": "https://api.github.com/users/jhlau/follower... | [] | closed | false | null | [] | [
"Yes, it's the average",
"Thanks for the prompt reply. Much appreciated."
] | 1,563 | 1,563 | 1,563 | NONE | null | >>> config = GPT2Config.from_pretrained('gpt2')
>>> tokenizer = GPT2Tokenizer.from_pretrained('gpt2')
>>> model = GPT2LMHeadModel(config)
>>> input_ids = torch.tensor(tokenizer.encode("Hello, my dog is cute")).unsqueeze(0) # Batch size 1
>>> outputs = model(input_ids, labels=input_ids)
>>> loss, logits = outp... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/818/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/818/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/817 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/817/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/817/comments | https://api.github.com/repos/huggingface/transformers/issues/817/events | https://github.com/huggingface/transformers/issues/817 | 469,593,272 | MDU6SXNzdWU0Njk1OTMyNzI= | 817 | from pytorch-pretrained-bert to pytorch-transformers，some problem | {
"login": "SStarLib",
"id": 10860531,
"node_id": "MDQ6VXNlcjEwODYwNTMx",
"avatar_url": "https://avatars.githubusercontent.com/u/10860531?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/SStarLib",
"html_url": "https://github.com/SStarLib",
"followers_url": "https://api.github.com/users/SSt... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"now you should use:\r\n```\r\nmodel = BertModel.from_pretrained('bert-base-cased', output_hidden_states=True)\r\noutputs = model(input_ids)\r\nall_hidden_states = outputs[-1]\r\n```\r\nNote that the first element in `all_hidden_states` (`all_hidden_states[0]`) is the output of the embedding layers (hence the fact ... | 1,563 | 1,569 | 1,569 | NONE | null | TypeError: forward() got an unexpected keyword argument 'output_all_encoded_layers'
| {
"url": "https://api.github.com/repos/huggingface/transformers/issues/817/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/817/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/816 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/816/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/816/comments | https://api.github.com/repos/huggingface/transformers/issues/816/events | https://github.com/huggingface/transformers/pull/816 | 469,588,687 | MDExOlB1bGxSZXF1ZXN0Mjk4NzY5Mzkx | 816 | typos | {
"login": "lpq29743",
"id": 12952648,
"node_id": "MDQ6VXNlcjEyOTUyNjQ4",
"avatar_url": "https://avatars.githubusercontent.com/u/12952648?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/lpq29743",
"html_url": "https://github.com/lpq29743",
"followers_url": "https://api.github.com/users/lpq... | [] | closed | false | null | [] | [
"# [Codecov](https://codecov.io/gh/huggingface/pytorch-transformers/pull/816?src=pr&el=h1) Report\n> Merging [#816](https://codecov.io/gh/huggingface/pytorch-transformers/pull/816?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/pytorch-transformers/commit/71d597dad0a28ccc397308146844486e0031d701?src... | 1,563 | 1,563 | 1,563 | CONTRIBUTOR | null | README.md: "formely known as" -> "formerly known as" | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/816/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/816/timeline | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/816",
"html_url": "https://github.com/huggingface/transformers/pull/816",
"diff_url": "https://github.com/huggingface/transformers/pull/816.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/816.patch",
"merged_at": 15634... |
https://api.github.com/repos/huggingface/transformers/issues/815 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/815/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/815/comments | https://api.github.com/repos/huggingface/transformers/issues/815/events | https://github.com/huggingface/transformers/pull/815 | 469,581,781 | MDExOlB1bGxSZXF1ZXN0Mjk4NzY0MzY1 | 815 | Update Readme link for Fine Tune/Usage section | {
"login": "praateekmahajan",
"id": 7589415,
"node_id": "MDQ6VXNlcjc1ODk0MTU=",
"avatar_url": "https://avatars.githubusercontent.com/u/7589415?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/praateekmahajan",
"html_url": "https://github.com/praateekmahajan",
"followers_url": "https://api.g... | [] | closed | false | null | [] | [
"# [Codecov](https://codecov.io/gh/huggingface/pytorch-transformers/pull/815?src=pr&el=h1) Report\n> Merging [#815](https://codecov.io/gh/huggingface/pytorch-transformers/pull/815?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/pytorch-transformers/commit/71d597dad0a28ccc397308146844486e0031d701?src... | 1,563 | 1,563 | 1,563 | CONTRIBUTOR | null | Incorrect link for `Quick tour: Fine-tuning/usage scripts` | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/815/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/815/timeline | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/815",
"html_url": "https://github.com/huggingface/transformers/pull/815",
"diff_url": "https://github.com/huggingface/transformers/pull/815.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/815.patch",
"merged_at": 15634... |
https://api.github.com/repos/huggingface/transformers/issues/814 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/814/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/814/comments | https://api.github.com/repos/huggingface/transformers/issues/814/events | https://github.com/huggingface/transformers/issues/814 | 469,535,376 | MDU6SXNzdWU0Njk1MzUzNzY= | 814 | Is there any plan of developing softmax-weight function for using 12 hidden BERT layer? | {
"login": "izuna385",
"id": 35322641,
"node_id": "MDQ6VXNlcjM1MzIyNjQx",
"avatar_url": "https://avatars.githubusercontent.com/u/35322641?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/izuna385",
"html_url": "https://github.com/izuna385",
"followers_url": "https://api.github.com/users/izu... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"Yes we might add a module for scalar mixture of layers like the one of AllenNLP, for instance (https://github.com/allenai/allennlp/blob/master/allennlp/modules/scalar_mix.py).",
"I'm looking forward to see that in also pytorch-transformer.\r\nAgain, thanks! I'll keep track on this repository.",
"This issue has... | 1,563 | 1,569 | 1,569 | NONE | null | Thanks for developing very nice/useful library.
My question is about using 12/(or in large model, more) hidden layer.
First, Does how to use hidden layers depend on the downstream task?
(Say, concat, average, only final layer, only mean of top 4 layer, etc...)
For using all layer, I think it's good to use softm... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/814/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/814/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/813 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/813/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/813/comments | https://api.github.com/repos/huggingface/transformers/issues/813/events | https://github.com/huggingface/transformers/issues/813 | 469,515,555 | MDU6SXNzdWU0Njk1MTU1NTU= | 813 | How to use BertModel ? | {
"login": "g-jing",
"id": 44223191,
"node_id": "MDQ6VXNlcjQ0MjIzMTkx",
"avatar_url": "https://avatars.githubusercontent.com/u/44223191?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/g-jing",
"html_url": "https://github.com/g-jing",
"followers_url": "https://api.github.com/users/g-jing/fo... | [
{
"id": 1314768611,
"node_id": "MDU6TGFiZWwxMzE0NzY4NjEx",
"url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix",
"name": "wontfix",
"color": "ffffff",
"default": true,
"description": null
}
] | closed | false | null | [] | [
"This issue has been discussed at [#64](https://github.com/huggingface/pytorch-transformers/issues/64#issuecomment-443703063).",
"This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n",
"Hi -... | 1,563 | 1,612 | 1,569 | NONE | null | I want to use bert-crf in my NER task. But this github only provide softmax as the classifier, I decided to write my own crf. But I am not sure how to use it. Here is an example. Please correct me if I am wrong.
sentence: Here is some text to encode
input: torch.tensor([tokenizer.encode("[CLS]" + "Here is some text... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/813/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/813/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/812 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/812/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/812/comments | https://api.github.com/repos/huggingface/transformers/issues/812/events | https://github.com/huggingface/transformers/issues/812 | 469,510,942 | MDU6SXNzdWU0Njk1MTA5NDI= | 812 | do I need to add sep and cls token in each sequence ? | {
"login": "g-jing",
"id": 44223191,
"node_id": "MDQ6VXNlcjQ0MjIzMTkx",
"avatar_url": "https://avatars.githubusercontent.com/u/44223191?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/g-jing",
"html_url": "https://github.com/g-jing",
"followers_url": "https://api.github.com/users/g-jing/fo... | [] | closed | false | null | [] | [
"They are not added automatically.",
"@thomwolf Thanks !"
] | 1,563 | 1,563 | 1,563 | NONE | null | It might be a stupid question, but I just notice the authors did not add "[cls]" and "[sep]" token in the example. I think whether those tokens are added automatically inside the module ? Thanks | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/812/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/812/timeline | completed | null | null |
https://api.github.com/repos/huggingface/transformers/issues/811 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/811/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/811/comments | https://api.github.com/repos/huggingface/transformers/issues/811/events | https://github.com/huggingface/transformers/pull/811 | 469,465,244 | MDExOlB1bGxSZXF1ZXN0Mjk4NjgzOTM4 | 811 | Fix openai-gpt ROCStories example's issues with AdamW optimizer | {
"login": "prrao87",
"id": 35005448,
"node_id": "MDQ6VXNlcjM1MDA1NDQ4",
"avatar_url": "https://avatars.githubusercontent.com/u/35005448?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/prrao87",
"html_url": "https://github.com/prrao87",
"followers_url": "https://api.github.com/users/prrao8... | [] | closed | false | null | [] | [
"# [Codecov](https://codecov.io/gh/huggingface/pytorch-transformers/pull/811?src=pr&el=h1) Report\n> Merging [#811](https://codecov.io/gh/huggingface/pytorch-transformers/pull/811?src=pr&el=desc) into [master](https://codecov.io/gh/huggingface/pytorch-transformers/commit/71d597dad0a28ccc397308146844486e0031d701?src... | 1,563 | 1,563 | 1,563 | NONE | null | Fixes the `AdamW` optimizer instance in the `openai-gpt` ROCStories example as per the new API. The default arguments for it are now set as per the [documentation](https://huggingface.co/pytorch-transformers/model_doc/bert.html?highlight=adamw#pytorch_transformers.AdamW). | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/811/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/811/timeline | null | false | {
"url": "https://api.github.com/repos/huggingface/transformers/pulls/811",
"html_url": "https://github.com/huggingface/transformers/pull/811",
"diff_url": "https://github.com/huggingface/transformers/pull/811.diff",
"patch_url": "https://github.com/huggingface/transformers/pull/811.patch",
"merged_at": null
... |
https://api.github.com/repos/huggingface/transformers/issues/810 | https://api.github.com/repos/huggingface/transformers | https://api.github.com/repos/huggingface/transformers/issues/810/labels{/name} | https://api.github.com/repos/huggingface/transformers/issues/810/comments | https://api.github.com/repos/huggingface/transformers/issues/810/events | https://github.com/huggingface/transformers/issues/810 | 469,436,460 | MDU6SXNzdWU0Njk0MzY0NjA= | 810 | SEG_ID constants for XLNet misleading/off | {
"login": "sleepinyourhat",
"id": 1284441,
"node_id": "MDQ6VXNlcjEyODQ0NDE=",
"avatar_url": "https://avatars.githubusercontent.com/u/1284441?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/sleepinyourhat",
"html_url": "https://github.com/sleepinyourhat",
"followers_url": "https://api.gith... | [] | closed | false | null | [] | [
"Yes I will remove them. They are used in the `run_glue.py` example (like in the original TF repo) but they don't have any reason to be in the library it-self.\r\n\r\nIn XLNet segment ids (what we call `token_type_ids in the repo) don't correspond to embeddings, they are just numbers and the only important thing is... | 1,563 | 1,563 | 1,563 | NONE | null | https://github.com/huggingface/pytorch-transformers/blob/master/pytorch_transformers/tokenization_xlnet.py#L47 shows:
```
# Segments (not really needed)
SEG_ID_A = 0
SEG_ID_B = 1
SEG_ID_CLS = 2
SEG_ID_SEP = 3
SEG_ID_PAD = 4
```
These don't seem to be used anywhere in the repo, but I tried using them as a... | {
"url": "https://api.github.com/repos/huggingface/transformers/issues/810/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
} | https://api.github.com/repos/huggingface/transformers/issues/810/timeline | completed | null | null |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.