model update

README.md

@@ -0,0 +1,215 @@
+
+---
+license: cc-by-4.0
+metrics:
+- bleu4
+- meteor
+- rouge-l
+- bertscore
+- moverscore
+language: fr
+datasets:
+- lmqg/qg_frquad
+pipeline_tag: text2text-generation
+tags:
+- question generation
+- answer extraction
+widget:
+- text: "generate question: Créateur » (Maker), lui aussi au singulier, « <hl> le Suprême Berger <hl> » (The Great Shepherd) ; de l'autre, des réminiscences de la théologie de l'Antiquité : le tonnerre, voix de Jupiter, « Et souvent ta voix gronde en un tonnerre terrifiant », etc."
+  example_title: "Question Generation Example 1" 
+- text: "generate question: Ce black dog peut être lié à des évènements traumatisants issus du monde extérieur, tels que son renvoi de l'Amirauté après la catastrophe des Dardanelles, lors de la <hl> Grande Guerre <hl> de 14-18, ou son rejet par l'électorat en juillet 1945."
+  example_title: "Question Generation Example 2" 
+- text: "generate question: contre <hl> Normie Smith <hl> et 15 000 dollars le 28 novembre 1938."
+  example_title: "Question Generation Example 3" 
+- text: "extract answers: Pourtant, la strophe spensérienne, utilisée cinq fois avant que ne commence le chœur, constitue en soi un vecteur dont les répétitions structurelles, selon Ricks, relèvent du pur lyrisme tout en constituant une menace potentielle. Après les huit sages pentamètres iambiques, l'alexandrin final <hl> permet une pause <hl>, « véritable illusion d'optique » qu'accentuent les nombreuses expressions archaïsantes telles que did swoon, did seem, did go, did receive, did make, qui doublent le prétérit en un temps composé et paraissent à la fois « très précautionneuses et très peu pressées »."
+  example_title: "Answer Extraction Example 1" 
+- text: "extract answers: Néanmoins, une fois encore, l'arithmétique modulaire est insuffisante pour venir à bout du théorème. Dirichlet utilise de nombreuses techniques analytiques, comme les séries entières et l'analyse complexe. Le fruit de ces travaux donne naissance à une nouvelle branche des mathématiques : la théorie analytique des nombres. L'un des points cruciaux de cette théorie provient de l'unique article de <hl> Bernhard Riemann <hl> en théorie des nombres : Sur le nombre de nombres premiers inférieurs à une taille donnée. Il conjecture une localisation des racines de sa fonction ζ. La recherche de la position des racines, initiée par Dirichlet, devient une préoccupation centrale et reste l'une des conjectures pressenties comme les plus difficiles des mathématiques de notre époque."
+  example_title: "Answer Extraction Example 2" 
+model-index:
+- name: lmqg/mbart-large-cc25-frquad-qg-ae
+  results:
+  - task:
+      name: Text2text Generation
+      type: text2text-generation
+    dataset:
+      name: lmqg/qg_frquad
+      type: default
+      args: default
+    metrics:
+    - name: BLEU4 (Question Generation)
+      type: bleu4_question_generation
+      value: 0.91
+    - name: ROUGE-L (Question Generation)
+      type: rouge_l_question_generation
+      value: 18.54
+    - name: METEOR (Question Generation)
+      type: meteor_question_generation
+      value: 8.56
+    - name: BERTScore (Question Generation)
+      type: bertscore_question_generation
+      value: 72.56
+    - name: MoverScore (Question Generation)
+      type: moverscore_question_generation
+      value: 50.46
+    - name: QAAlignedF1Score-BERTScore (Question & Answer Generation (with Gold Answer))
+      type: qa_aligned_f1_score_bertscore_question_answer_generation_with_gold_answer
+      value: 77.72
+    - name: QAAlignedRecall-BERTScore (Question & Answer Generation (with Gold Answer))
+      type: qa_aligned_recall_bertscore_question_answer_generation_with_gold_answer
+      value: 78.58
+    - name: QAAlignedPrecision-BERTScore (Question & Answer Generation (with Gold Answer))
+      type: qa_aligned_precision_bertscore_question_answer_generation_with_gold_answer
+      value: 76.9
+    - name: QAAlignedF1Score-MoverScore (Question & Answer Generation (with Gold Answer))
+      type: qa_aligned_f1_score_moverscore_question_answer_generation_with_gold_answer
+      value: 51.65
+    - name: QAAlignedRecall-MoverScore (Question & Answer Generation (with Gold Answer))
+      type: qa_aligned_recall_moverscore_question_answer_generation_with_gold_answer
+      value: 52.16
+    - name: QAAlignedPrecision-MoverScore (Question & Answer Generation (with Gold Answer))
+      type: qa_aligned_precision_moverscore_question_answer_generation_with_gold_answer
+      value: 51.15
+    - name: BLEU4 (Answer Extraction)
+      type: bleu4_answer_extraction
+      value: 0.0
+    - name: ROUGE-L (Answer Extraction)
+      type: rouge_l_answer_extraction
+      value: 3.48
+    - name: METEOR (Answer Extraction)
+      type: meteor_answer_extraction
+      value: 3.24
+    - name: BERTScore (Answer Extraction)
+      type: bertscore_answer_extraction
+      value: 58.41
+    - name: MoverScore (Answer Extraction)
+      type: moverscore_answer_extraction
+      value: 45.72
+    - name: AnswerF1Score (Answer Extraction)
+      type: answer_f1_score__answer_extraction
+      value: 3.66
+    - name: AnswerExactMatch (Answer Extraction)
+      type: answer_exact_match_answer_extraction
+      value: 0.0
+---
+
+# Model Card of `lmqg/mbart-large-cc25-frquad-qg-ae`
+This model is fine-tuned version of [facebook/mbart-large-cc25](https://huggingface.co/facebook/mbart-large-cc25) for question generation and answer extraction jointly on the [lmqg/qg_frquad](https://huggingface.co/datasets/lmqg/qg_frquad) (dataset_name: default) via [`lmqg`](https://github.com/asahi417/lm-question-generation).
+
+
+### Overview
+- **Language model:** [facebook/mbart-large-cc25](https://huggingface.co/facebook/mbart-large-cc25)   
+- **Language:** fr  
+- **Training data:** [lmqg/qg_frquad](https://huggingface.co/datasets/lmqg/qg_frquad) (default)
+- **Online Demo:** [https://autoqg.net/](https://autoqg.net/)
+- **Repository:** [https://github.com/asahi417/lm-question-generation](https://github.com/asahi417/lm-question-generation)
+- **Paper:** [https://arxiv.org/abs/2210.03992](https://arxiv.org/abs/2210.03992)
+
+### Usage
+- With [`lmqg`](https://github.com/asahi417/lm-question-generation#lmqg-language-model-for-question-generation-)
+```python
+from lmqg import TransformersQG
+
+# initialize model
+model = TransformersQG(language="fr", model="lmqg/mbart-large-cc25-frquad-qg-ae")
+
+# model prediction
+question_answer_pairs = model.generate_qa("Créateur » (Maker), lui aussi au singulier, « le Suprême Berger » (The Great Shepherd) ; de l'autre, des réminiscences de la théologie de l'Antiquité : le tonnerre, voix de Jupiter, « Et souvent ta voix gronde en un tonnerre terrifiant », etc.")
+
+```
+
+- With `transformers`
+```python
+from transformers import pipeline
+
+pipe = pipeline("text2text-generation", "lmqg/mbart-large-cc25-frquad-qg-ae")
+
+# answer extraction
+answer = pipe("generate question: Créateur » (Maker), lui aussi au singulier, « <hl> le Suprême Berger <hl> » (The Great Shepherd) ; de l'autre, des réminiscences de la théologie de l'Antiquité : le tonnerre, voix de Jupiter, « Et souvent ta voix gronde en un tonnerre terrifiant », etc.")
+
+# question generation
+question = pipe("extract answers: Pourtant, la strophe spensérienne, utilisée cinq fois avant que ne commence le chœur, constitue en soi un vecteur dont les répétitions structurelles, selon Ricks, relèvent du pur lyrisme tout en constituant une menace potentielle. Après les huit sages pentamètres iambiques, l'alexandrin final <hl> permet une pause <hl>, « véritable illusion d'optique » qu'accentuent les nombreuses expressions archaïsantes telles que did swoon, did seem, did go, did receive, did make, qui doublent le prétérit en un temps composé et paraissent à la fois « très précautionneuses et très peu pressées ».")
+
+```
+
+## Evaluation
+
+
+- ***Metric (Question Generation)***: [raw metric file](https://huggingface.co/lmqg/mbart-large-cc25-frquad-qg-ae/raw/main/eval/metric.first.sentence.paragraph_answer.question.lmqg_qg_frquad.default.json) 
+
+|            |   Score | Type    | Dataset                                                          |
+|:-----------|--------:|:--------|:-----------------------------------------------------------------|
+| BERTScore  |   72.56 | default | [lmqg/qg_frquad](https://huggingface.co/datasets/lmqg/qg_frquad) |
+| Bleu_1     |   16.16 | default | [lmqg/qg_frquad](https://huggingface.co/datasets/lmqg/qg_frquad) |
+| Bleu_2     |    4.88 | default | [lmqg/qg_frquad](https://huggingface.co/datasets/lmqg/qg_frquad) |
+| Bleu_3     |    1.85 | default | [lmqg/qg_frquad](https://huggingface.co/datasets/lmqg/qg_frquad) |
+| Bleu_4     |    0.91 | default | [lmqg/qg_frquad](https://huggingface.co/datasets/lmqg/qg_frquad) |
+| METEOR     |    8.56 | default | [lmqg/qg_frquad](https://huggingface.co/datasets/lmqg/qg_frquad) |
+| MoverScore |   50.46 | default | [lmqg/qg_frquad](https://huggingface.co/datasets/lmqg/qg_frquad) |
+| ROUGE_L    |   18.54 | default | [lmqg/qg_frquad](https://huggingface.co/datasets/lmqg/qg_frquad) |
+
+
+- ***Metric (Question & Answer Generation)***:  [raw metric file](https://huggingface.co/lmqg/mbart-large-cc25-frquad-qg-ae/raw/main/eval/metric.first.answer.paragraph.questions_answers.lmqg_qg_frquad.default.json)
+
+|                                 |   Score | Type    | Dataset                                                          |
+|:--------------------------------|--------:|:--------|:-----------------------------------------------------------------|
+| QAAlignedF1Score (BERTScore)    |   77.72 | default | [lmqg/qg_frquad](https://huggingface.co/datasets/lmqg/qg_frquad) |
+| QAAlignedF1Score (MoverScore)   |   51.65 | default | [lmqg/qg_frquad](https://huggingface.co/datasets/lmqg/qg_frquad) |
+| QAAlignedPrecision (BERTScore)  |   76.9  | default | [lmqg/qg_frquad](https://huggingface.co/datasets/lmqg/qg_frquad) |
+| QAAlignedPrecision (MoverScore) |   51.15 | default | [lmqg/qg_frquad](https://huggingface.co/datasets/lmqg/qg_frquad) |
+| QAAlignedRecall (BERTScore)     |   78.58 | default | [lmqg/qg_frquad](https://huggingface.co/datasets/lmqg/qg_frquad) |
+| QAAlignedRecall (MoverScore)    |   52.16 | default | [lmqg/qg_frquad](https://huggingface.co/datasets/lmqg/qg_frquad) |
+
+
+- ***Metric (Answer Extraction)***: [raw metric file](https://huggingface.co/lmqg/mbart-large-cc25-frquad-qg-ae/raw/main/eval/metric.first.answer.paragraph_sentence.answer.lmqg_qg_frquad.default.json)
+
+|                  |   Score | Type    | Dataset                                                          |
+|:-----------------|--------:|:--------|:-----------------------------------------------------------------|
+| AnswerExactMatch |    0    | default | [lmqg/qg_frquad](https://huggingface.co/datasets/lmqg/qg_frquad) |
+| AnswerF1Score    |    3.66 | default | [lmqg/qg_frquad](https://huggingface.co/datasets/lmqg/qg_frquad) |
+| BERTScore        |   58.41 | default | [lmqg/qg_frquad](https://huggingface.co/datasets/lmqg/qg_frquad) |
+| Bleu_1           |    2.56 | default | [lmqg/qg_frquad](https://huggingface.co/datasets/lmqg/qg_frquad) |
+| Bleu_2           |    0.76 | default | [lmqg/qg_frquad](https://huggingface.co/datasets/lmqg/qg_frquad) |
+| Bleu_3           |    0    | default | [lmqg/qg_frquad](https://huggingface.co/datasets/lmqg/qg_frquad) |
+| Bleu_4           |    0    | default | [lmqg/qg_frquad](https://huggingface.co/datasets/lmqg/qg_frquad) |
+| METEOR           |    3.24 | default | [lmqg/qg_frquad](https://huggingface.co/datasets/lmqg/qg_frquad) |
+| MoverScore       |   45.72 | default | [lmqg/qg_frquad](https://huggingface.co/datasets/lmqg/qg_frquad) |
+| ROUGE_L          |    3.48 | default | [lmqg/qg_frquad](https://huggingface.co/datasets/lmqg/qg_frquad) |
+
+
+
+## Training hyperparameters
+
+The following hyperparameters were used during fine-tuning:
+ - dataset_path: lmqg/qg_frquad
+ - dataset_name: default
+ - input_types: ['paragraph_answer', 'paragraph_sentence']
+ - output_types: ['question', 'answer']
+ - prefix_types: ['qg', 'ae']
+ - model: facebook/mbart-large-cc25
+ - max_length: 512
+ - max_length_output: 32
+ - epoch: 5
+ - batch: 2
+ - lr: 0.0001
+ - fp16: False
+ - random_seed: 1
+ - gradient_accumulation_steps: 32
+ - label_smoothing: 0.15
+
+The full configuration can be found at [fine-tuning config file](https://huggingface.co/lmqg/mbart-large-cc25-frquad-qg-ae/raw/main/trainer_config.json).
+
+## Citation
+```
+@inproceedings{ushio-etal-2022-generative,
+    title = "{G}enerative {L}anguage {M}odels for {P}aragraph-{L}evel {Q}uestion {G}eneration",
+    author = "Ushio, Asahi  and
+        Alva-Manchego, Fernando  and
+        Camacho-Collados, Jose",
+    booktitle = "Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing",
+    month = dec,
+    year = "2022",
+    address = "Abu Dhabi, U.A.E.",
+    publisher = "Association for Computational Linguistics",
+}
+
+```

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "lmqg_output/mbart-large-cc25-frquad-qg-ae/best_model",
+  "_name_or_path": "facebook/mbart-large-cc25",
   "_num_labels": 3,
   "activation_dropout": 0.0,
   "activation_function": "gelu",

eval/metric.first.answer.paragraph.questions_answers.lmqg_qg_frquad.default.json

@@ -0,0 +1 @@
+{"test": {"QAAlignedF1Score (BERTScore)": 0.7772413075848769, "QAAlignedRecall (BERTScore)": 0.7858170717744499, "QAAlignedPrecision (BERTScore)": 0.7689549139583383, "QAAlignedF1Score (MoverScore)": 0.5164507832847365, "QAAlignedRecall (MoverScore)": 0.5215983072529093, "QAAlignedPrecision (MoverScore)": 0.5114646311900342, "Bleu_1": 0.037782235103238186, "Bleu_2": 0.015199909819190933, "Bleu_3": 0.004885828523775546, "Bleu_4": 0.0010360112893287412, "METEOR": 0.1054760059084704, "ROUGE_L": 0.07329315468535748, "BERTScore": 0.5734855355142509, "MoverScore": 0.49132126390282654}, "validation": {"QAAlignedF1Score (BERTScore)": 0.7856529529386986, "QAAlignedRecall (BERTScore)": 0.7828424381920649, "QAAlignedPrecision (BERTScore)": 0.7885869638043304, "QAAlignedF1Score (MoverScore)": 0.5210219089023471, "QAAlignedRecall (MoverScore)": 0.5205554081240038, "QAAlignedPrecision (MoverScore)": 0.5215353446160117, "Bleu_1": 0.11868833101890196, "Bleu_2": 0.05962247943466257, "Bleu_3": 0.018656299341997678, "Bleu_4": 0.006695070088080167, "METEOR": 0.1826567214417883, "ROUGE_L": 0.1712360131717798, "BERTScore": 0.6651338213899484, "MoverScore": 0.5104089792922674}}

eval/metric.first.answer.paragraph_answer.question.lmqg_qg_frquad.default.json

@@ -0,0 +1 @@
+{"validation": {"Bleu_1": 0.16304445241082266, "Bleu_2": 0.04359089100104815, "Bleu_3": 0.01563846849594309, "Bleu_4": 0.006834006918163857}, "test": {"Bleu_1": 0.16118480014339126, "Bleu_2": 0.04850750055849233, "Bleu_3": 0.018321994368103143, "Bleu_4": 0.008914098623388059}}

eval/metric.first.answer.paragraph_sentence.answer.lmqg_qg_frquad.default.json

@@ -0,0 +1 @@
+{"validation": {"Bleu_1": 0.031053952321203823, "Bleu_2": 0.008912110057459717, "Bleu_3": 0.001275715526012735, "Bleu_4": 8.771777990582571e-08, "METEOR": 0.035440713934410256, "ROUGE_L": 0.03976014741067617, "BERTScore": 0.5888344976242393, "MoverScore": 0.4594421933634128, "AnswerF1Score": 4.397795867894216, "AnswerExactMatch": 0.0}, "test": {"Bleu_1": 0.02556461731493042, "Bleu_2": 0.007614699024995999, "Bleu_3": 1.1486877120295986e-08, "Bleu_4": 1.4418621186928819e-11, "METEOR": 0.03242150774410041, "ROUGE_L": 0.03476518310190234, "BERTScore": 0.5841268705790734, "MoverScore": 0.4571808179187319, "AnswerF1Score": 3.662940194122861, "AnswerExactMatch": 0.0}}

eval/metric.first.sentence.paragraph_answer.question.lmqg_qg_frquad.default.json

@@ -0,0 +1 @@
+{"validation": {"Bleu_1": 0.16420308483290114, "Bleu_2": 0.04418468099154152, "Bleu_3": 0.01583842777449354, "Bleu_4": 0.0069411827995547876, "METEOR": 0.08164915030915688, "ROUGE_L": 0.19507497362063778, "BERTScore": 0.722660510506299, "MoverScore": 0.5040415700487906}, "test": {"Bleu_1": 0.16155802126675542, "Bleu_2": 0.04875364461552838, "Bleu_3": 0.018514125978873273, "Bleu_4": 0.009054516092973287, "METEOR": 0.08556904020375079, "ROUGE_L": 0.18541590755594925, "BERTScore": 0.7256057500260548, "MoverScore": 0.5046048262328876}}

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8ca7308192a6d07270e334ca92dba3302998e89c16583976c82ad4392b643ddf
-size 2444580125
+oid sha256:468a76648b10b87d53978dca873afdb50bd49186d142a9636e471b48b49128af
+size 2444587421

tokenizer_config.json

@@ -12,7 +12,7 @@
     "single_word": false
   },
   "model_max_length": 1024,
-  "name_or_path": "lmqg_output/mbart-large-cc25-frquad-qg-ae/best_model",
+  "name_or_path": "facebook/mbart-large-cc25",
   "pad_token": "<pad>",
   "sep_token": "</s>",
   "special_tokens_map_file": null,

trainer_config.json

@@ -0,0 +1 @@
+{"dataset_path": "lmqg/qg_frquad", "dataset_name": "default", "input_types": ["paragraph_answer", "paragraph_sentence"], "output_types": ["question", "answer"], "prefix_types": ["qg", "ae"], "model": "facebook/mbart-large-cc25", "max_length": 512, "max_length_output": 32, "epoch": 5, "batch": 2, "lr": 0.0001, "fp16": false, "random_seed": 1, "gradient_accumulation_steps": 32, "label_smoothing": 0.15}