{
    "info": {
        "author": "Thomas Wolf, Lysandre Debut, Victor Sanh, Julien Chaumond, Google AI Language Team Authors, Open AI team Authors",
        "author_email": "thomas@huggingface.co",
        "bugtrack_url": null,
        "classifiers": [
            "Intended Audience :: Science/Research",
            "License :: OSI Approved :: Apache Software License",
            "Programming Language :: Python :: 3",
            "Topic :: Scientific/Engineering :: Artificial Intelligence"
        ],
        "description": "# \ud83d\udc7e PyTorch-Transformers\n\n[![CircleCI](https://circleci.com/gh/huggingface/pytorch-transformers.svg?style=svg)](https://circleci.com/gh/huggingface/pytorch-transformers)\n\nPyTorch-Transformers (formerly known as `pytorch-pretrained-bert`) is a library of state-of-the-art pre-trained models for Natural Language Processing (NLP).\n\nThe library currently contains PyTorch implementations, pre-trained model weights, usage scripts and conversion utilities for the following models:\n\n1. **[BERT](https://github.com/google-research/bert)** (from Google) released with the paper [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://arxiv.org/abs/1810.04805) by Jacob Devlin, Ming-Wei Chang, Kenton Lee and Kristina Toutanova.\n2. **[GPT](https://github.com/openai/finetune-transformer-lm)** (from OpenAI) released with the paper [Improving Language Understanding by Generative Pre-Training](https://blog.openai.com/language-unsupervised/) by Alec Radford, Karthik Narasimhan, Tim Salimans and Ilya Sutskever.\n3. **[GPT-2](https://blog.openai.com/better-language-models/)** (from OpenAI) released with the paper [Language Models are Unsupervised Multitask Learners](https://blog.openai.com/better-language-models/) by Alec Radford*, Jeffrey Wu*, Rewon Child, David Luan, Dario Amodei** and Ilya Sutskever**.\n4. **[Transformer-XL](https://github.com/kimiyoung/transformer-xl)** (from Google/CMU) released with the paper [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) by Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov.\n5. **[XLNet](https://github.com/zihangdai/xlnet/)** (from Google/CMU) released with the paper [\u200bXLNet: Generalized Autoregressive Pretraining for Language Understanding](https://arxiv.org/abs/1906.08237) by Zhilin Yang*, Zihang Dai*, Yiming Yang, Jaime Carbonell, Ruslan Salakhutdinov, Quoc V. Le.\n6. **[XLM](https://github.com/facebookresearch/XLM/)** (from Facebook) released together with the paper [Cross-lingual Language Model Pretraining](https://arxiv.org/abs/1901.07291) by Guillaume Lample and Alexis Conneau.\n7. **[RoBERTa](https://github.com/pytorch/fairseq/tree/master/examples/roberta)** (from Facebook), released together with the paper a [Robustly Optimized BERT Pretraining Approach](https://arxiv.org/abs/1907.11692) by Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, Veselin Stoyanov.\n8. **[DistilBERT](https://github.com/huggingface/pytorch-transformers/tree/master/examples/distillation)** (from HuggingFace), released together with the blogpost [Smaller, faster, cheaper, lighter: Introducing DistilBERT, a distilled version of\u00a0BERT](https://medium.com/huggingface/distilbert-8cf3380435b5\n) by Victor Sanh, Lysandre Debut and Thomas Wolf.\n\nThese implementations have been tested on several datasets (see the example scripts) and should match the performances of the original implementations (e.g. ~93 F1 on SQuAD for BERT Whole-Word-Masking, ~88 F1 on RocStories for OpenAI GPT, ~18.3 perplexity on WikiText 103 for Transformer-XL, ~0.916 Peason R coefficient on STS-B for XLNet). You can find more details on the performances in the Examples section of the [documentation](https://huggingface.co/pytorch-transformers/examples.html).\n\n| Section | Description |\n|-|-|\n| [Installation](#installation) | How to install the package |\n| [Quick tour: Usage](#quick-tour) | Tokenizers & models usage: Bert and GPT-2 |\n| [Quick tour: Fine-tuning/usage scripts](#quick-tour-of-the-fine-tuningusage-scripts) | Using provided scripts: GLUE, SQuAD and Text generation |\n| [Migrating from pytorch-pretrained-bert to pytorch-transformers](#Migrating-from-pytorch-pretrained-bert-to-pytorch-transformers) | Migrating your code from pytorch-pretrained-bert to pytorch-transformers |\n| [Documentation](https://huggingface.co/pytorch-transformers/) | Full API documentation and more |\n\n## Installation\n\nThis repo is tested on Python 2.7 and 3.5+ (examples are tested only on python 3.5+) and PyTorch 1.0.0+\n\n### With pip\n\nPyTorch-Transformers can be installed by pip as follows:\n\n```bash\npip install pytorch-transformers\n```\n\n### From source\n\nClone the repository and run:\n\n```bash\npip install [--editable] .\n```\n\n### Tests\n\nA series of tests is included for the library and the example scripts. Library tests can be found in the [tests folder](https://github.com/huggingface/pytorch-transformers/tree/master/pytorch_transformers/tests) and examples tests in the [examples folder](https://github.com/huggingface/pytorch-transformers/tree/master/examples).\n\nThese tests can be run using `pytest` (install pytest if needed with `pip install pytest`).\n\nYou can run the tests from the root of the cloned repository with the commands:\n\n```bash\npython -m pytest -sv ./pytorch_transformers/tests/\npython -m pytest -sv ./examples/\n```\n\n### Do you want to run a Transformer model on a mobile device?\n\nYou should check out our [`swift-coreml-transformers`](https://github.com/huggingface/swift-coreml-transformers) repo.\n\nIt contains an example of a conversion script from a Pytorch trained Transformer model (here, `GPT-2`) to a CoreML model that runs on iOS devices.\n\nAt some point in the future, you'll be able to seamlessly move from pre-training or fine-tuning models in PyTorch to productizing them in CoreML,\nor prototype a model or an app in CoreML then research its hyperparameters or architecture from PyTorch. Super exciting!\n\n\n## Quick tour\n\nLet's do a very quick overview of PyTorch-Transformers. Detailed examples for each model architecture (Bert, GPT, GPT-2, Transformer-XL, XLNet and XLM) can be found in the [full documentation](https://huggingface.co/pytorch-transformers/).\n\n```python\nimport torch\nfrom pytorch_transformers import *\n\n# PyTorch-Transformers has a unified API\n# for 7 transformer architectures and 30 pretrained weights.\n#          Model          | Tokenizer          | Pretrained weights shortcut\nMODELS = [(BertModel,       BertTokenizer,      'bert-base-uncased'),\n          (OpenAIGPTModel,  OpenAIGPTTokenizer, 'openai-gpt'),\n          (GPT2Model,       GPT2Tokenizer,      'gpt2'),\n          (TransfoXLModel,  TransfoXLTokenizer, 'transfo-xl-wt103'),\n          (XLNetModel,      XLNetTokenizer,     'xlnet-base-cased'),\n          (XLMModel,        XLMTokenizer,       'xlm-mlm-enfr-1024'),\n          (RobertaModel,    RobertaTokenizer,   'roberta-base')]\n\n# Let's encode some text in a sequence of hidden-states using each model:\nfor model_class, tokenizer_class, pretrained_weights in MODELS:\n    # Load pretrained model/tokenizer\n    tokenizer = tokenizer_class.from_pretrained(pretrained_weights)\n    model = model_class.from_pretrained(pretrained_weights)\n\n    # Encode text\n    input_ids = torch.tensor([tokenizer.encode(\"Here is some text to encode\", add_special_tokens=True)])  # Add special tokens takes care of adding [CLS], [SEP], <s>... tokens in the right way for each model.\n    with torch.no_grad():\n        last_hidden_states = model(input_ids)[0]  # Models outputs are now tuples\n\n# Each architecture is provided with several class for fine-tuning on down-stream tasks, e.g.\nBERT_MODEL_CLASSES = [BertModel, BertForPreTraining, BertForMaskedLM, BertForNextSentencePrediction,\n                      BertForSequenceClassification, BertForMultipleChoice, BertForTokenClassification,\n                      BertForQuestionAnswering]\n\n# All the classes for an architecture can be initiated from pretrained weights for this architecture\n# Note that additional weights added for fine-tuning are only initialized\n# and need to be trained on the down-stream task\ntokenizer = BertTokenizer.from_pretrained('bert-base-uncased')\nfor model_class in BERT_MODEL_CLASSES:\n    # Load pretrained model/tokenizer\n    model = model_class.from_pretrained('bert-base-uncased')\n\n# Models can return full list of hidden-states & attentions weights at each layer\nmodel = model_class.from_pretrained(pretrained_weights,\n                                    output_hidden_states=True,\n                                    output_attentions=True)\ninput_ids = torch.tensor([tokenizer.encode(\"Let's see all hidden-states and attentions on this text\")])\nall_hidden_states, all_attentions = model(input_ids)[-2:]\n\n# Models are compatible with Torchscript\nmodel = model_class.from_pretrained(pretrained_weights, torchscript=True)\ntraced_model = torch.jit.trace(model, (input_ids,))\n\n# Simple serialization for models and tokenizers\nmodel.save_pretrained('./directory/to/save/')  # save\nmodel = model_class.from_pretrained('./directory/to/save/')  # re-load\ntokenizer.save_pretrained('./directory/to/save/')  # save\ntokenizer = tokenizer_class.from_pretrained('./directory/to/save/')  # re-load\n\n# SOTA examples for GLUE, SQUAD, text generation...\n```\n\n## Quick tour of the fine-tuning/usage scripts\n\nThe library comprises several example scripts with SOTA performances for NLU and NLG tasks:\n\n- `run_glue.py`: an example fine-tuning Bert, XLNet and XLM on nine different GLUE tasks (*sequence-level classification*)\n- `run_squad.py`: an example fine-tuning Bert, XLNet and XLM on the question answering dataset SQuAD 2.0 (*token-level classification*)\n- `run_generation.py`: an example using GPT, GPT-2, Transformer-XL and XLNet for conditional language generation\n- other model-specific examples (see the documentation).\n\nHere are three quick usage examples for these scripts:\n\n### `run_glue.py`: Fine-tuning on GLUE tasks for sequence classification\n\nThe [General Language Understanding Evaluation (GLUE) benchmark](https://gluebenchmark.com/) is a collection of nine sentence- or sentence-pair language understanding tasks for evaluating and analyzing natural language understanding systems.\n\nBefore running anyone of these GLUE tasks you should download the\n[GLUE data](https://gluebenchmark.com/tasks) by running\n[this script](https://gist.github.com/W4ngatang/60c2bdb54d156a41194446737ce03e2e)\nand unpack it to some directory `$GLUE_DIR`.\n\nYou should also install the additional packages required by the examples:\n\n```shell\npip install -r ./examples/requirements.txt\n```\n\n```shell\nexport GLUE_DIR=/path/to/glue\nexport TASK_NAME=MRPC\n\npython ./examples/run_glue.py \\\n    --model_type bert \\\n    --model_name_or_path bert-base-uncased \\\n    --task_name $TASK_NAME \\\n    --do_train \\\n    --do_eval \\\n    --do_lower_case \\\n    --data_dir $GLUE_DIR/$TASK_NAME \\\n    --max_seq_length 128 \\\n    --per_gpu_eval_batch_size=8   \\\n    --per_gpu_train_batch_size=8   \\\n    --learning_rate 2e-5 \\\n    --num_train_epochs 3.0 \\\n    --output_dir /tmp/$TASK_NAME/\n```\n\nwhere task name can be one of CoLA, SST-2, MRPC, STS-B, QQP, MNLI, QNLI, RTE, WNLI.\n\nThe dev set results will be present within the text file 'eval_results.txt' in the specified output_dir. In case of MNLI, since there are two separate dev sets, matched and mismatched, there will be a separate output folder called '/tmp/MNLI-MM/' in addition to '/tmp/MNLI/'.\n\n#### Fine-tuning XLNet model on the STS-B regression task\n\nThis example code fine-tunes XLNet on the STS-B corpus using parallel training on a server with 4 V100 GPUs.\nParallel training is a simple way to use several GPUs (but is slower and less flexible than distributed training, see below).\n\n```shell\nexport GLUE_DIR=/path/to/glue\n\npython ./examples/run_glue.py \\\n    --model_type xlnet \\\n    --model_name_or_path xlnet-large-cased \\\n    --do_train  \\\n    --do_eval   \\\n    --task_name=sts-b     \\\n    --data_dir=${GLUE_DIR}/STS-B  \\\n    --output_dir=./proc_data/sts-b-110   \\\n    --max_seq_length=128   \\\n    --per_gpu_eval_batch_size=8   \\\n    --per_gpu_train_batch_size=8   \\\n    --gradient_accumulation_steps=1 \\\n    --max_steps=1200  \\\n    --model_name=xlnet-large-cased   \\\n    --overwrite_output_dir   \\\n    --overwrite_cache \\\n    --warmup_steps=120\n```\n\nOn this machine we thus have a batch size of 32, please increase `gradient_accumulation_steps` to reach the same batch size if you have a smaller machine. These hyper-parameters should result in a Pearson correlation coefficient of `+0.917` on the development set.\n\n#### Fine-tuning Bert model on the MRPC classification task\n\nThis example code fine-tunes the Bert Whole Word Masking model on the Microsoft Research Paraphrase Corpus (MRPC) corpus using distributed training on 8 V100 GPUs to reach a F1 > 92.\n\n```bash\npython -m torch.distributed.launch --nproc_per_node 8 ./examples/run_glue.py   \\\n    --model_type bert \\\n    --model_name_or_path bert-large-uncased-whole-word-masking \\\n    --task_name MRPC \\\n    --do_train   \\\n    --do_eval   \\\n    --do_lower_case   \\\n    --data_dir $GLUE_DIR/MRPC/   \\\n    --max_seq_length 128   \\\n    --per_gpu_eval_batch_size=8   \\\n    --per_gpu_train_batch_size=8   \\\n    --learning_rate 2e-5   \\\n    --num_train_epochs 3.0  \\\n    --output_dir /tmp/mrpc_output/ \\\n    --overwrite_output_dir   \\\n    --overwrite_cache \\\n```\n\nTraining with these hyper-parameters gave us the following results:\n\n```bash\n  acc = 0.8823529411764706\n  acc_and_f1 = 0.901702786377709\n  eval_loss = 0.3418912578906332\n  f1 = 0.9210526315789473\n  global_step = 174\n  loss = 0.07231863956341798\n```\n\n### `run_squad.py`: Fine-tuning on SQuAD for question-answering\n\nThis example code fine-tunes BERT on the SQuAD dataset using distributed training on 8 V100 GPUs and Bert Whole Word Masking uncased model to reach a F1 > 93 on SQuAD:\n\n```bash\npython -m torch.distributed.launch --nproc_per_node=8 ./examples/run_squad.py \\\n    --model_type bert \\\n    --model_name_or_path bert-large-uncased-whole-word-masking \\\n    --do_train \\\n    --do_eval \\\n    --do_lower_case \\\n    --train_file $SQUAD_DIR/train-v1.1.json \\\n    --predict_file $SQUAD_DIR/dev-v1.1.json \\\n    --learning_rate 3e-5 \\\n    --num_train_epochs 2 \\\n    --max_seq_length 384 \\\n    --doc_stride 128 \\\n    --output_dir ../models/wwm_uncased_finetuned_squad/ \\\n    --per_gpu_eval_batch_size=3   \\\n    --per_gpu_train_batch_size=3   \\\n```\n\nTraining with these hyper-parameters gave us the following results:\n\n```bash\npython $SQUAD_DIR/evaluate-v1.1.py $SQUAD_DIR/dev-v1.1.json ../models/wwm_uncased_finetuned_squad/predictions.json\n{\"exact_match\": 86.91579943235573, \"f1\": 93.1532499015869}\n```\n\nThis is the model provided as `bert-large-uncased-whole-word-masking-finetuned-squad`.\n\n### `run_generation.py`: Text generation with GPT, GPT-2, Transformer-XL and XLNet\n\nA conditional generation script is also included to generate text from a prompt.\nThe generation script includes the [tricks](https://github.com/rusiaaman/XLNet-gen#methodology) proposed by by Aman Rusia to get high quality generation with memory models like Transformer-XL and XLNet (include a predefined text to make short inputs longer).\n\nHere is how to run the script with the small version of OpenAI GPT-2 model:\n\n```shell\npython ./examples/run_generation.py \\\n    --model_type=gpt2 \\\n    --length=20 \\\n    --model_name_or_path=gpt2 \\\n```\n\n## Migrating from pytorch-pretrained-bert to pytorch-transformers\n\nHere is a quick summary of what you should take care of when migrating from `pytorch-pretrained-bert` to `pytorch-transformers`\n\n### Models always output `tuples`\n\nThe main breaking change when migrating from `pytorch-pretrained-bert` to `pytorch-transformers` is that the models forward method always outputs a `tuple` with various elements depending on the model and the configuration parameters.\n\nThe exact content of the tuples for each model are detailed in the models' docstrings and the [documentation](https://huggingface.co/pytorch-transformers/).\n\nIn pretty much every case, you will be fine by taking the first element of the output as the output you previously used in `pytorch-pretrained-bert`.\n\nHere is a `pytorch-pretrained-bert` to `pytorch-transformers` conversion example for a `BertForSequenceClassification` classification model:\n\n```python\n# Let's load our model\nmodel = BertForSequenceClassification.from_pretrained('bert-base-uncased')\n\n# If you used to have this line in pytorch-pretrained-bert:\nloss = model(input_ids, labels=labels)\n\n# Now just use this line in pytorch-transformers to extract the loss from the output tuple:\noutputs = model(input_ids, labels=labels)\nloss = outputs[0]\n\n# In pytorch-transformers you can also have access to the logits:\nloss, logits = outputs[:2]\n\n# And even the attention weights if you configure the model to output them (and other outputs too, see the docstrings and documentation)\nmodel = BertForSequenceClassification.from_pretrained('bert-base-uncased', output_attentions=True)\noutputs = model(input_ids, labels=labels)\nloss, logits, attentions = outputs\n```\n\n### Serialization\n\nBreaking change in the `from_pretrained()`method:\n\n1. Models are now set in evaluation mode by default when instantiated with the `from_pretrained()` method. To train them don't forget to set them back in training mode (`model.train()`) to activate the dropout modules.\n\n2. The additional `*input` and `**kwargs` arguments supplied to the `from_pretrained()` method used to be directly passed to the underlying model's class `__init__()` method. They are now used to update the model configuration attribute instead which can break derived model classes build based on the previous `BertForSequenceClassification` examples. We are working on a way to mitigate this breaking change in [#866](https://github.com/huggingface/pytorch-transformers/pull/866) by forwarding the the model `__init__()` method (i) the provided positional arguments and (ii) the keyword arguments which do not match any configuration class attributes.\n\nAlso, while not a breaking change, the serialization methods have been standardized and you probably should switch to the new method `save_pretrained(save_directory)` if you were using any other serialization method before.\n\nHere is an example:\n\n```python\n### Let's load a model and tokenizer\nmodel = BertForSequenceClassification.from_pretrained('bert-base-uncased')\ntokenizer = BertTokenizer.from_pretrained('bert-base-uncased')\n\n### Do some stuff to our model and tokenizer\n# Ex: add new tokens to the vocabulary and embeddings of our model\ntokenizer.add_tokens(['[SPECIAL_TOKEN_1]', '[SPECIAL_TOKEN_2]'])\nmodel.resize_token_embeddings(len(tokenizer))\n# Train our model\ntrain(model)\n\n### Now let's save our model and tokenizer to a directory\nmodel.save_pretrained('./my_saved_model_directory/')\ntokenizer.save_pretrained('./my_saved_model_directory/')\n\n### Reload the model and the tokenizer\nmodel = BertForSequenceClassification.from_pretrained('./my_saved_model_directory/')\ntokenizer = BertTokenizer.from_pretrained('./my_saved_model_directory/')\n```\n\n### Optimizers: BertAdam & OpenAIAdam are now AdamW, schedules are standard PyTorch schedules\n\nThe two optimizers previously included, `BertAdam` and `OpenAIAdam`, have been replaced by a single `AdamW` optimizer which has a few differences:\n\n- it only implements weights decay correction,\n- schedules are now externals (see below),\n- gradient clipping is now also external (see below).\n\nThe new optimizer `AdamW` matches PyTorch `Adam` optimizer API and let you use standard PyTorch or apex methods for the schedule and clipping.\n\nThe schedules are now standard [PyTorch learning rate schedulers](https://pytorch.org/docs/stable/optim.html#how-to-adjust-learning-rate) and not part of the optimizer anymore.\n\nHere is a conversion examples from `BertAdam` with a linear warmup and decay schedule to `AdamW` and the same schedule:\n\n```python\n# Parameters:\nlr = 1e-3\nmax_grad_norm = 1.0\nnum_total_steps = 1000\nnum_warmup_steps = 100\nwarmup_proportion = float(num_warmup_steps) / float(num_total_steps)  # 0.1\n\n### Previously BertAdam optimizer was instantiated like this:\noptimizer = BertAdam(model.parameters(), lr=lr, schedule='warmup_linear', warmup=warmup_proportion, t_total=num_total_steps)\n### and used like this:\nfor batch in train_data:\n    loss = model(batch)\n    loss.backward()\n    optimizer.step()\n\n### In PyTorch-Transformers, optimizer and schedules are splitted and instantiated like this:\noptimizer = AdamW(model.parameters(), lr=lr, correct_bias=False)  # To reproduce BertAdam specific behavior set correct_bias=False\nscheduler = WarmupLinearSchedule(optimizer, warmup_steps=num_warmup_steps, t_total=num_total_steps)  # PyTorch scheduler\n### and used like this:\nfor batch in train_data:\n    loss = model(batch)\n    loss.backward()\n    torch.nn.utils.clip_grad_norm_(model.parameters(), max_grad_norm)  # Gradient clipping is not in AdamW anymore (so you can use amp without issue)\n    optimizer.step()\n    scheduler.step()\n    optimizer.zero_grad()\n```\n\n## Citation\n\nAt the moment, there is no paper associated to PyTorch-Transformers but we are working on preparing one. In the meantime, please include a mention of the library and a link to the present repository if you use this work in a published or open-source project.\n\n\n",
        "description_content_type": "text/markdown",
        "docs_url": null,
        "download_url": "",
        "downloads": {
            "last_day": -1,
            "last_month": -1,
            "last_week": -1
        },
        "home_page": "https://github.com/huggingface/pytorch-transformers",
        "keywords": "NLP deep learning transformer pytorch BERT GPT GPT-2 google openai CMU",
        "license": "Apache",
        "maintainer": "",
        "maintainer_email": "",
        "name": "pytorch-transformers",
        "package_url": "https://pypi.org/project/pytorch-transformers/",
        "platform": "",
        "project_url": "https://pypi.org/project/pytorch-transformers/",
        "project_urls": {
            "Homepage": "https://github.com/huggingface/pytorch-transformers"
        },
        "release_url": "https://pypi.org/project/pytorch-transformers/1.2.0/",
        "requires_dist": [
            "torch (>=1.0.0)",
            "numpy",
            "boto3",
            "requests",
            "tqdm",
            "regex",
            "sentencepiece",
            "sacremoses"
        ],
        "requires_python": "",
        "summary": "Repository of pre-trained NLP Transformer models: BERT & RoBERTa, GPT & GPT-2, Transformer-XL, XLNet and XLM",
        "version": "1.2.0"
    },
    "last_serial": 5780726,
    "releases": {
        "0.7.0": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "96d666aeae9d7e1d530174c4e159a746",
                    "sha256": "f21184abaf9704b9fb7a1cf361cf3e6d4cfad33f8c1324180314744ed951f922"
                },
                "downloads": -1,
                "filename": "pytorch_transformers-0.7.0-py3-none-any.whl",
                "has_sig": false,
                "md5_digest": "96d666aeae9d7e1d530174c4e159a746",
                "packagetype": "bdist_wheel",
                "python_version": "py3",
                "requires_python": null,
                "size": 150690,
                "upload_time": "2019-07-05T10:23:15",
                "url": "https://files.pythonhosted.org/packages/2d/b4/2592044f6cefe77e6bf9bb418e6b15cbff7385b434c79e499b5e6e72812f/pytorch_transformers-0.7.0-py3-none-any.whl"
            },
            {
                "comment_text": "",
                "digests": {
                    "md5": "c5e3ef8b995fa33d1c14650e27d15311",
                    "sha256": "164dd082f4186e91822d9d44074cceb226b1fe634813a1cb1f02519c5f3376d5"
                },
                "downloads": -1,
                "filename": "pytorch_transformers-0.7.0.tar.gz",
                "has_sig": false,
                "md5_digest": "c5e3ef8b995fa33d1c14650e27d15311",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 168148,
                "upload_time": "2019-07-05T10:23:18",
                "url": "https://files.pythonhosted.org/packages/c1/5f/143e82a823339ffc503d013c47878890aab383a9101582220a335e6bc877/pytorch_transformers-0.7.0.tar.gz"
            }
        ],
        "1.0.0": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "8898512f9728bfc3586c791f767da89c",
                    "sha256": "9e36f64a497d3e43aed5effd826d9d9ba0e0b335c279359a56562c8238966c78"
                },
                "downloads": -1,
                "filename": "pytorch_transformers-1.0.0-py3-none-any.whl",
                "has_sig": false,
                "md5_digest": "8898512f9728bfc3586c791f767da89c",
                "packagetype": "bdist_wheel",
                "python_version": "py3",
                "requires_python": null,
                "size": 137195,
                "upload_time": "2019-07-16T14:13:32",
                "url": "https://files.pythonhosted.org/packages/40/b5/2d78e74001af0152ee61d5ad4e290aec9a1e43925b21df2dc74ec100f1ab/pytorch_transformers-1.0.0-py3-none-any.whl"
            },
            {
                "comment_text": "",
                "digests": {
                    "md5": "d9d7886f07ecfba433873168fa044964",
                    "sha256": "5ee90ec47b1862f9c94304abe60359d06ab641e7235d9346d142559740ee7854"
                },
                "downloads": -1,
                "filename": "pytorch_transformers-1.0.0.tar.gz",
                "has_sig": false,
                "md5_digest": "d9d7886f07ecfba433873168fa044964",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 118954,
                "upload_time": "2019-07-16T14:13:38",
                "url": "https://files.pythonhosted.org/packages/aa/89/5f24ad07fe768fb3c6e2f210541d578e87a78f2587b18f7a37868b8bbedb/pytorch_transformers-1.0.0.tar.gz"
            }
        ],
        "1.1.0": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "6f888189cbe38f77c987bc8ba8714738",
                    "sha256": "cd670284de1b1b0bfe2f4a27531e873b5a76cd163960661971a340f33c3f4290"
                },
                "downloads": -1,
                "filename": "pytorch_transformers-1.1.0-py3-none-any.whl",
                "has_sig": false,
                "md5_digest": "6f888189cbe38f77c987bc8ba8714738",
                "packagetype": "bdist_wheel",
                "python_version": "py3",
                "requires_python": null,
                "size": 158066,
                "upload_time": "2019-08-15T21:14:28",
                "url": "https://files.pythonhosted.org/packages/50/89/ad0d6bb932d0a51793eaabcf1617a36ff530dc9ab9e38f765a35dc293306/pytorch_transformers-1.1.0-py3-none-any.whl"
            },
            {
                "comment_text": "",
                "digests": {
                    "md5": "8727e7ba90e3be94807a607f218c141e",
                    "sha256": "f1eadbd394564c0dbe0e12e3ebb5b1470169af275c2a71e20c2480ee5a93e94c"
                },
                "downloads": -1,
                "filename": "pytorch_transformers-1.1.0.tar.gz",
                "has_sig": false,
                "md5_digest": "8727e7ba90e3be94807a607f218c141e",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 133084,
                "upload_time": "2019-08-15T21:14:31",
                "url": "https://files.pythonhosted.org/packages/72/a7/68f2d8fd6d927e374181af34f827a9f4a469487d99d0c017177c5db8726e/pytorch_transformers-1.1.0.tar.gz"
            }
        ],
        "1.2.0": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "885d028952907e3b9425567892e89d6f",
                    "sha256": "15f12a04424c0f6d3a7c7b57d6c79628dc9c117a204fb7db8c1ea330c77a6898"
                },
                "downloads": -1,
                "filename": "pytorch_transformers-1.2.0-py2-none-any.whl",
                "has_sig": false,
                "md5_digest": "885d028952907e3b9425567892e89d6f",
                "packagetype": "bdist_wheel",
                "python_version": "py2",
                "requires_python": null,
                "size": 176428,
                "upload_time": "2019-09-04T11:36:34",
                "url": "https://files.pythonhosted.org/packages/a9/be/587dd871b8abcdb34bbfbfac976b403bb22bc5c41d3d18cff329bf669fd3/pytorch_transformers-1.2.0-py2-none-any.whl"
            },
            {
                "comment_text": "",
                "digests": {
                    "md5": "2dba40cda929caa2d2c8c843bf327336",
                    "sha256": "bdb606fe1f2d27586710ed03cfa49dbbd80215c38bf965862daada0c137fd7ce"
                },
                "downloads": -1,
                "filename": "pytorch_transformers-1.2.0-py3-none-any.whl",
                "has_sig": false,
                "md5_digest": "2dba40cda929caa2d2c8c843bf327336",
                "packagetype": "bdist_wheel",
                "python_version": "py3",
                "requires_python": null,
                "size": 176430,
                "upload_time": "2019-09-04T11:36:36",
                "url": "https://files.pythonhosted.org/packages/a3/b7/d3d18008a67e0b968d1ab93ad444fc05699403fa662f634b2f2c318a508b/pytorch_transformers-1.2.0-py3-none-any.whl"
            },
            {
                "comment_text": "",
                "digests": {
                    "md5": "48e994040a609c050cf559764e430cb6",
                    "sha256": "293e4a864ae9d9401f9fba13f16b8696e4a1cb38bcd0b56562d03af5489daeb9"
                },
                "downloads": -1,
                "filename": "pytorch_transformers-1.2.0.tar.gz",
                "has_sig": false,
                "md5_digest": "48e994040a609c050cf559764e430cb6",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 152283,
                "upload_time": "2019-09-04T11:36:39",
                "url": "https://files.pythonhosted.org/packages/39/46/60ade12cd10f3e3dc7cb109361a73ebd8a8530c35bed71f681d2588aa277/pytorch_transformers-1.2.0.tar.gz"
            }
        ]
    },
    "urls": [
        {
            "comment_text": "",
            "digests": {
                "md5": "885d028952907e3b9425567892e89d6f",
                "sha256": "15f12a04424c0f6d3a7c7b57d6c79628dc9c117a204fb7db8c1ea330c77a6898"
            },
            "downloads": -1,
            "filename": "pytorch_transformers-1.2.0-py2-none-any.whl",
            "has_sig": false,
            "md5_digest": "885d028952907e3b9425567892e89d6f",
            "packagetype": "bdist_wheel",
            "python_version": "py2",
            "requires_python": null,
            "size": 176428,
            "upload_time": "2019-09-04T11:36:34",
            "url": "https://files.pythonhosted.org/packages/a9/be/587dd871b8abcdb34bbfbfac976b403bb22bc5c41d3d18cff329bf669fd3/pytorch_transformers-1.2.0-py2-none-any.whl"
        },
        {
            "comment_text": "",
            "digests": {
                "md5": "2dba40cda929caa2d2c8c843bf327336",
                "sha256": "bdb606fe1f2d27586710ed03cfa49dbbd80215c38bf965862daada0c137fd7ce"
            },
            "downloads": -1,
            "filename": "pytorch_transformers-1.2.0-py3-none-any.whl",
            "has_sig": false,
            "md5_digest": "2dba40cda929caa2d2c8c843bf327336",
            "packagetype": "bdist_wheel",
            "python_version": "py3",
            "requires_python": null,
            "size": 176430,
            "upload_time": "2019-09-04T11:36:36",
            "url": "https://files.pythonhosted.org/packages/a3/b7/d3d18008a67e0b968d1ab93ad444fc05699403fa662f634b2f2c318a508b/pytorch_transformers-1.2.0-py3-none-any.whl"
        },
        {
            "comment_text": "",
            "digests": {
                "md5": "48e994040a609c050cf559764e430cb6",
                "sha256": "293e4a864ae9d9401f9fba13f16b8696e4a1cb38bcd0b56562d03af5489daeb9"
            },
            "downloads": -1,
            "filename": "pytorch_transformers-1.2.0.tar.gz",
            "has_sig": false,
            "md5_digest": "48e994040a609c050cf559764e430cb6",
            "packagetype": "sdist",
            "python_version": "source",
            "requires_python": null,
            "size": 152283,
            "upload_time": "2019-09-04T11:36:39",
            "url": "https://files.pythonhosted.org/packages/39/46/60ade12cd10f3e3dc7cb109361a73ebd8a8530c35bed71f681d2588aa277/pytorch_transformers-1.2.0.tar.gz"
        }
    ]
}