Updating modelcards for ModelUsage (#209)

* updating modelcards for ModelUsage

* bug fix

allennlp_models/modelcards/coref-spanbert.json

@@ -4,7 +4,6 @@
  "registered_predictor_name": null,
  "display_name": "Coreference Resolution",
  "task_id": "coref",
- "archive_file": "coref-spanbert-large-2020.02.27.tar.gz",
  "model_details": {
  "description": "The basic outline of this model is to get an embedded representation of each span in the document. These span representations are scored and used to prune away spans that are unlikely to occur in a coreference cluster. For the remaining spans, the model decides which antecedent span (if any) they are coreferent with. The resulting coreference links, after applying transitivity, imply a clustering of the spans in the document. The GloVe embeddings in the original paper have been substituted with SpanBERT embeddings.",
  "short_description": "Higher-order coref with coarse-to-fine inference (with SpanBERT embeddings).",
@@ -19,8 +18,7 @@
  "url": "https://api.semanticscholar.org/CorpusID:4891749"
  },
  "license": null,
- "contact": "[email protected]",
- "training_config": "coref/coref_spanbert_large.jsonnet"
+ "contact": "[email protected]"
  },
  "intended_use": {
  "primary_uses": null,
@@ -39,15 +37,17 @@
  "evaluation_data": {
  "dataset": {
  "name": "Ontonotes 5.0",
- "url": "https://catalog.ldc.upenn.edu/LDC2013T19"
+ "url": "https://catalog.ldc.upenn.edu/LDC2013T19",
+ "notes": "The Coreference model was evaluated on the CoNLL 2012 dataset. Unfortunately we cannot release this data due to licensing restrictions by the LDC. To compile the data in the right format for evaluating the Coreference model, please see scripts/compile_coref_data.sh. This script requires the Ontonotes 5.0 dataset, available on the LDC website."
  },
  "motivation": null,
  "preprocessing": null
  },
  "training_data": {
  "dataset": {
  "name": "Ontonotes 5.0",
- "url": "https://catalog.ldc.upenn.edu/LDC2013T19"
+ "url": "https://catalog.ldc.upenn.edu/LDC2013T19",
+ "notes": "The Coreference model was evaluated on the CoNLL 2012 dataset. Unfortunately we cannot release this data due to licensing restrictions by the LDC. To compile the data in the right format for evaluating the Coreference model, please see scripts/compile_coref_data.sh. This script requires the Ontonotes 5.0 dataset, available on the LDC website."
  },
  "motivation": null,
  "preprocessing": null
@@ -61,5 +61,10 @@
  },
  "model_ethical_considerations": {
  "ethical_considerations": null
+ },
+ "model_usage": {
+ "archive_file": "coref-spanbert-large-2020.02.27.tar.gz",
+ "training_config": "coref/coref_spanbert_large.jsonnet",
+ "install_instructions": "pip install allennlp==1.0.0 allennlp-models==1.0.0"
  }
 }

allennlp_models/modelcards/generation-bart.json

@@ -3,7 +3,6 @@
  "registered_model_name": "bart",
  "registered_predictor_name": null,
  "display_name": "BART",
- "archive_file": "bart-2020.07.25.tar.gz",
  "model_details": {
  "description": "The BART model here uses a language modeling head, and therefore can be used for generation. The BART encoder, implemented as a `Seq2SeqEncoder`, which assumes it operates on already embedded inputs. This means that we remove the token and position embeddings from BART in this module. For the typical use case of using BART to encode inputs to your model (where we include the token and position embeddings from BART), you should use `PretrainedTransformerEmbedder(bart_model_name, sub_module=\"encoder\")` instead of this.",
  "short_description": "BART with a language model head for generation.",
@@ -37,15 +36,17 @@
  "evaluation_data": {
  "dataset": {
  "name": "CNN/DailyMail",
- "url": "https:/abisee/cnn-dailymail"
+ "url": "https:/abisee/cnn-dailymail",
+ "notes": "Please download the data from the url provided."
  },
  "motivation": null,
  "preprocessing": null
  },
  "training_data": {
  "dataset": {
  "name": "CNN/DailyMail",
- "url": "https:/abisee/cnn-dailymail"
+ "url": "https:/abisee/cnn-dailymail",
+ "notes": "Please download the data from the url provided."
  },
  "motivation": null,
  "preprocessing": null
@@ -59,5 +60,10 @@
  },
  "model_ethical_considerations": {
  "ethical_considerations": null
+ },
+ "model_usage": {
+ "archive_file": "bart-2020.07.25.tar.gz",
+ "training_config": "generation/bart_cnn_dm.jsonnet",
+ "install_instructions": "pip install allennlp==1.0.0 allennlp-models==1.0.0"
  }
 }

allennlp_models/modelcards/glove-sst.json

@@ -4,7 +4,6 @@
  "registered_predictor_name": null,
  "display_name": "GLoVe-LSTM",
  "task_id": "sentiment-analysis",
- "archive_file": "basic_stanford_sentiment_treebank-2020.06.09.tar.gz",
  "model_details": {
  "description": "This model uses GloVe embeddings and is trained on the binary classification setting of the Stanford Sentiment Treebank. It achieves about 87% on the test set.",
  "short_description": "LSTM binary classifier with GloVe embeddings.",
@@ -15,8 +14,7 @@
  "model_type": "LSTM",
  "paper": null,
  "license": null,
- "contact": "[email protected]",
- "training_config": "classification/basic_stanford_sentiment_treebank.jsonnet"
+ "contact": "[email protected]"
  },
  "intended_use": {
  "primary_uses": null,
@@ -35,15 +33,17 @@
  "evaluation_data": {
  "dataset": {
  "name": "Stanford Sentiment Treebank",
- "url": "https://nlp.stanford.edu/sentiment/treebank.html"
+ "url": "https://nlp.stanford.edu/sentiment/treebank.html",
+ "processed_url": "https://allennlp.s3.amazonaws.com/datasets/sst/test.txt"
  },
  "motivation": null,
  "preprocessing": null
  },
  "training_data": {
  "dataset": {
  "name": "Stanford Sentiment Treebank",
- "url": "https://nlp.stanford.edu/sentiment/treebank.html"
+ "url": "https://nlp.stanford.edu/sentiment/treebank.html",
+ "processed_url": "https://allennlp.s3.amazonaws.com/datasets/sst/train.txt"
  },
  "motivation": null,
  "preprocessing": "Binary classification setting"
@@ -57,5 +57,10 @@
  },
  "model_ethical_considerations": {
  "ethical_considerations": null
+ },
+ "model_usage": {
+ "archive_file": "basic_stanford_sentiment_treebank-2020.06.09.tar.gz",
+ "training_config": "classification/basic_stanford_sentiment_treebank.jsonnet",
+ "install_instructions": "pip install allennlp==1.0.0 allennlp-models==1.0.0"
  }
 }

allennlp_models/modelcards/lm-masked-language-model.json

@@ -4,7 +4,6 @@
  "registered_predictor_name": null,
  "display_name": "BERT-based Masked Language Model",
  "task_id": "masked-language-modeling",
- "archive_file": "bert-masked-lm-2020-10-07.tar.gz",
  "model_details": {
  "description": "The `MaskedLanguageModel` embeds some input tokens (including some which are masked), contextualizes them, then predicts targets for the masked tokens, computing a loss against known targets.",
  "short_description": "BERT-based masked language model",
@@ -19,8 +18,7 @@
  "url": "https://api.semanticscholar.org/CorpusID:52967399"
  },
  "license": null,
- "contact": "[email protected]",
- "training_config": null
+ "contact": "[email protected]"
  },
  "intended_use": {
  "primary_uses": null,
@@ -55,5 +53,10 @@
  },
  "model_ethical_considerations": {
  "ethical_considerations": "BERT demonstrates gender bias in that it thinks the doctor is more likely a man ('his') than a woman ('her'). An important issue in NLP is how to understand and address such biases in our linguistic models."
+ },
+ "model_usage": {
+ "archive_file": "bert-masked-lm-2020-10-07.tar.gz",
+ "training_config": "lm/bidirectional_language_model.jsonnet",
+ "install_instructions": "pip install allennlp==1.0.0 allennlp-models==1.0.0"
  }
 }

allennlp_models/modelcards/lm-next-token-lm-gpt2.json

@@ -4,7 +4,6 @@
  "registered_predictor_name": null,
  "display_name": "GPT2-based Next Token Language Model",
  "task_id": "language-modeling",
- "archive_file": "gpt2-next-word-lm-2020.06.30.tar.gz",
  "model_details": {
  "description": "This is the public 345M parameter OpenAI GPT-2 language model for generating sentences. The model embeds some input tokens, contextualizes them, then predicts the next word, computing a loss against known target. \nIf `BeamSearch` is given, this model will predict a sequence of next tokens.",
  "short_description": "OpenAI's GPT-2 language model that generates the next token.",
@@ -19,8 +18,7 @@
  "url": "https://api.semanticscholar.org/CorpusID:160025533"
  },
  "license": null,
- "contact": "[email protected]",
- "training_config": null
+ "contact": "[email protected]"
  },
  "intended_use": {
  "primary_uses": null,
@@ -61,5 +59,10 @@
  },
  "model_ethical_considerations": {
  "ethical_considerations": null
+ },
+ "model_usage": {
+ "archive_file": "gpt2-next-word-lm-2020.06.30.tar.gz",
+ "training_config": null,
+ "install_instructions": "pip install allennlp==1.0.0 allennlp-models==1.0.0"
  }
 }

allennlp_models/modelcards/mc-roberta-commonsenseqa.json

@@ -4,7 +4,6 @@
  "registered_predictor_name": "transformer_mc",
  "display_name": "RoBERTa Common Sense QA",
  "task_id": "mc",
- "archive_file": "commonsenseqa.2020-07-08.tar.gz",
  "model_details": {
  "description": "This is a multiple choice model patterned after the BERT architecture. It calculates a score for each sequence on top of the CLS token, and then chooses the alternative with the highest score.",
  "short_description": "RoBERTa-based multiple choice model for CommonSenseQA.",
@@ -19,8 +18,7 @@
  "url": "https://api.semanticscholar.org/CorpusID:198953378"
  },
  "license": null,
- "contact": "[email protected]",
- "training_config": null
+ "contact": "[email protected]"
  },
  "intended_use": {
  "primary_uses": null,
@@ -39,15 +37,17 @@
  "evaluation_data": {
  "dataset": {
  "name": "CommonSenseQA (validation set)",
- "url": "https:/jonathanherzig/commonsenseqa"
+ "url": "https:/jonathanherzig/commonsenseqa",
+ "notes": "Please download the data from the url provided."
  },
  "motivation": null,
  "preprocessing": null
  },
  "training_data": {
  "dataset": {
  "name": "CommonSenseQA (train set)",
- "url": "https:/jonathanherzig/commonsenseqa"
+ "url": "https:/jonathanherzig/commonsenseqa",
+ "notes": "Please download the data from the url provided."
  },
  "motivation": null,
  "preprocessing": null
@@ -61,5 +61,10 @@
  },
  "model_ethical_considerations": {
  "ethical_considerations": null
+ },
+ "model_usage": {
+ "archive_file": "commonsenseqa.2020-07-08.tar.gz",
+ "training_config": "mc/commonsenseqa.jsonnet",
+ "install_instructions": "pip install allennlp==1.0.0 allennlp-models==1.0.0"
  }
 }

allennlp_models/modelcards/mc-roberta-piqa.json

@@ -4,7 +4,6 @@
  "registered_predictor_name": "transformer_mc",
  "display_name": "Physical Interaction Question Answering",
  "task_id": "mc",
- "archive_file": "piqa.2020-07-08.tar.gz",
  "model_details": {
  "description": "This is a multiple choice model patterned after the BERT architecture. It calculates a score for each sequence on top of the CLS token, and then chooses the alternative with the highest score.",
  "short_description": "RoBERTa-based multiple choice model for PIQA.",
@@ -19,8 +18,7 @@
  "url": "https://api.semanticscholar.org/CorpusID:198953378"
  },
  "license": null,
- "contact": "[email protected]",
- "training_config": null
+ "contact": "[email protected]"
  },
  "intended_use": {
  "primary_uses": null,
@@ -39,15 +37,17 @@
  "evaluation_data": {
  "dataset": {
  "name": "PIQA (validation set)",
- "url": "https://yonatanbisk.com/piqa/"
+ "url": "https://yonatanbisk.com/piqa/",
+ "notes": "Please download the data from the url provided."
  },
  "motivation": null,
  "preprocessing": null
  },
  "training_data": {
  "dataset": {
  "name": "PIQA (train set)",
- "url": "https://yonatanbisk.com/piqa/"
+ "url": "https://yonatanbisk.com/piqa/",
+ "notes": "Please download the data from the url provided."
  },
  "motivation": null,
  "preprocessing": null
@@ -61,5 +61,10 @@
  },
  "model_ethical_considerations": {
  "ethical_considerations": null
+ },
+ "model_usage": {
+ "archive_file": "piqa.2020-07-08.tar.gz",
+ "training_config": "mc/piqa.jsonnet",
+ "install_instructions": "pip install allennlp==1.0.0 allennlp-models==1.0.0"
  }
 }

allennlp_models/modelcards/mc-roberta-swag.json

@@ -4,7 +4,6 @@
  "registered_predictor_name": "transformer_mc",
  "display_name": "RoBERTa SWAG",
  "task_id": "mc",
- "archive_file": "swag.2020-07-08.tar.gz",
  "model_details": {
  "description": "This is a multiple choice model patterned after the BERT architecture. It calculates a score for each sequence on top of the CLS token, and then chooses the alternative with the highest score.",
  "short_description": "RoBERTa-based multiple choice model for SWAG.",
@@ -19,8 +18,7 @@
  "url": "https://api.semanticscholar.org/CorpusID:198953378"
  },
  "license": null,
- "contact": "[email protected]",
- "training_config": null
+ "contact": "[email protected]"
  },
  "intended_use": {
  "primary_uses": null,
@@ -39,15 +37,17 @@
  "evaluation_data": {
  "dataset": {
  "name": "SWAG (validation set)",
- "url": "https://rowanzellers.com/swag/"
+ "url": "https://rowanzellers.com/swag/",
+ "notes": "Please download the data from the url provided."
  },
  "motivation": null,
  "preprocessing": null
  },
  "training_data": {
  "dataset": {
  "name": "SWAG (train set)",
- "url": "https://rowanzellers.com/swag/"
+ "url": "https://rowanzellers.com/swag/",
+ "notes": "Please download the data from the url provided."
  },
  "motivation": null,
  "preprocessing": null
@@ -61,5 +61,10 @@
  },
  "model_ethical_considerations": {
  "ethical_considerations": null
+ },
+ "model_usage": {
+ "archive_file": "swag.2020-07-08.tar.gz",
+ "training_config": "mc/swag.jsonnet",
+ "install_instructions": "pip install allennlp==1.0.0 allennlp-models==1.0.0"
  }
 }

allennlp_models/modelcards/modelcard-template.json

@@ -4,7 +4,12 @@
  "registered_predictor_name": null,
  "display_name": null,
  "task_id": null,
- "archive_file": null,
+ "model_usage": {
+ "archive_file": null,
+ "training_config": null,
+ "overrides": null,
+ "install_instructions": "pip install allennlp==1.0.0 allennlp-models==1.0.0",
+ },
  "model_details": {
  "description": null,
  "short_description": null,
@@ -17,7 +22,6 @@
  "citation": null,
  "license": null,
  "contact": "[email protected]",
- "training_config": null
  },
  "intended_use": {
  "primary_uses": null,