update Pytorch version

README.md

@@ -1,10 +1,16 @@
-# TDEER
+[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/tdeer-an-efficient-translating-decoding/joint-entity-and-relation-extraction-on-nyt)](https://paperswithcode.com/sota/joint-entity-and-relation-extraction-on-nyt?p=tdeer-an-efficient-translating-decoding)
+[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/tdeer-an-efficient-translating-decoding/joint-entity-and-relation-extraction-on-1)](https://paperswithcode.com/sota/joint-entity-and-relation-extraction-on-1?p=tdeer-an-efficient-translating-decoding)
+[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/tdeer-an-efficient-translating-decoding/relation-extraction-on-nyt)](https://paperswithcode.com/sota/relation-extraction-on-nyt?p=tdeer-an-efficient-translating-decoding)
+[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/tdeer-an-efficient-translating-decoding/relation-extraction-on-webnlg)](https://paperswithcode.com/sota/relation-extraction-on-webnlg?p=tdeer-an-efficient-translating-decoding)
 
-Official Code For TDEER: An Efficient Translating Decoding Schema for Joint Extraction of Entities and Relations (EMNLP2021)
+
+# TDEER 🦌
+
+Official Code For [TDEER: An Efficient Translating Decoding Schema for Joint Extraction of Entities and Relations](https://aclanthology.org/2021.emnlp-main.635/) (EMNLP2021)
 
 ## Overview
 
-TDEER is an efficient model for joint extraction of entities and relations. Unlike the common decoding approach that predicting the relation between subject and object, we adopt the proposed translating decoding schema: subject + relation -> objects, to decode triples. By the proposed translating decoding schema, TDEER can handle the overlapping triple problem effectively and efficiently. The following figure is an illustration of our models.
+TDEER is an efficient model for joint extraction of entities and relations. Unlike the common decoding approach that predicts the relation between subject and object, we adopt the proposed translating decoding schema: subject + relation -> objects, to decode triples. By the proposed translating decoding schema, TDEER can handle the overlapping triple problem effectively and efficiently. The following figure is an illustration of our models.
 
 ![overview](docs/TDEER-Overview.png)
 
@@ -15,7 +21,9 @@ TDEER is an efficient model for joint extraction of entities and relations. Unli
 
 We conducted experiments under python3.7 and used GPUs device to accelerate computing. 
 
-You can install the required dependencies by the following script.
+You should first prepare the tensorflow version in terms of your GPU environment. For tensorflow version, we recommend `tensorflow-gpu==1.15.0`.
+
+Then, you can install the other required dependencies by the following script.
 
 ```bash
 pip install -r requirements.txt
@@ -37,127 +45,26 @@ Click 👉[BERT-Base-Cased](https://storage.googleapis.com/bert_models/2018_10_1
 
 ### 4. Train & Eval
 
-You can use `run.py` with `--do_train` to train the model. After training, you can also use `run.py` with `--do_test` to evaluate data.
-
-Our training and evaluating commands are as follows:
-
-1\. NYT
-
-train:
-
-```bash
-CUDA_VISIBLE_DEVICES=0 nohup python -u run.py \
---do_train \
---model_name NYT \
---rel_path data/NYT/rel2id.json \
---train_path data/NYT/train_triples.json \
---dev_path data/NYT/test_triples.json \
---bert_dir pretrained-bert/cased_L-12_H-768_A-12 \
---save_path ckpts/nyt.model \
---learning_rate 0.00005 \
---neg_samples 2 \
---epoch 200 \
---verbose 2 > nyt.log &
-```
-
-evaluate:
+we have already refactor this code to pytorch version, you could download the pretrain model directly from [huggingface](https://huggingface.co/bert-base-cased/tree/main).
 
-```
-CUDA_VISIBLE_DEVICES=0 python run.py \
---do_test \
---model_name NYT \
---rel_path data/NYT/rel2id.json \
---test_path data/NYT/test_triples.json \
---bert_dir pretrained-bert/cased_L-12_H-768_A-12 \
---ckpt_path ckpts/nyt.model \
---max_len 512 \
---verbose 1
-```
 
-You can evaluate other data by specifying `--test_path`.
+if you want to fast vaild the code 
 
-2\. WebNLG
+please excuete the following command:
 
-train:
+#### for training
 
 ```bash
-CUDA_VISIBLE_DEVICES=0 nohup python -u run.py \
---do_train \
---model_name WebNLG \
---rel_path data/WebNLG/rel2id.json \
---train_path data/WebNLG/train_triples.json \
---dev_path data/WebNLG/test_triples.json \
---bert_dir pretrained-bert/cased_L-12_H-768_A-12 \
---save_path ckpts/webnlg.model \
---max_sample_triples 5 \
---neg_samples 5 \
---learning_rate 0.00005 \
---epoch 300 \
---verbose 2 > webnlg.log &
-```
-
-evaluate:
 
-```bash
-CUDA_VISIBLE_DEVICES=0 python run.py \
---do_test \
---model_name WebNLG \
---rel_path data/WebNLG/rel2id.json \
---test_path data/WebNLG/test_triples.json \
---bert_dir pretrained-bert/cased_L-12_H-768_A-12 \
---ckpt_path ckpts/webnlg.model \
---max_len 512 \
---verbose 1
+python run.py --do_train --model_name NYT_quick_test --rel_path data/data/NYT/rel2id.json --train_path data/data/NYT/train_triples.json --dev_path data/data/NYT/test_triples.json --bert_model bert-base-cased --save_path ckpts/nyt_quick_test.model --learning_rate 0.00005 --neg_samples 2 --epoch 5 --batch_size 32 --max_len 120 --max_sample_triples 100 --eval_steps 100 --num_workers 4 --use_amp --subset_size 1000
 ```
 
-You can evaluate other data by specifying `--test_path`.
-
-
-3\. NYT11-HRL
-
-train:
+#### for testing
 
 ```bash
-CUDA_VISIBLE_DEVICES=0 nohup python -u run.py \
---do_train \
---model_name NYT11-HRL \
---rel_path data/NYT11-HRL/rel2id.json \
---train_path data/NYT11-HRL/train_triples.json \
---dev_path data/NYT11-HRL/test_triples.json \
---bert_dir pretrained-bert/cased_L-12_H-768_A-12 \
---save_path ckpts/nyt11hrl.model \
---learning_rate 0.00005 \
---neg_samples 1 \
---epoch 100 \
---verbose 2 > nyt11hrl.log &
+python run.py --do_test --model_name NYT_full_test --rel_path data/data/NYT/rel2id.json --test_path data/data/NYT/test_triples.json --bert_model bert-base-cased --ckpt_path ckpts/nyt_full_train.model --batch_size 32 --max_len 120
 ```
 
-evaluate:
-
-```
-CUDA_VISIBLE_DEVICES=0 python run.py \
---do_test \
---model_name NYT11-HRL \
---rel_path data/NYT/rel2id.json \
---test_path data/NYT11-HRL/test_triples.json \
---bert_dir pretrained-bert/cased_L-12_H-768_A-12 \
---ckpt_path ckpts/nyt11hrl.model \
---max_len 512 \
---verbose 1
-```
-
-
-### Pre-trained Models
-
-We released our pre-trained models for NYT, WebNLG, and NYT11-HRL datasets, and uploaded them to this repository via git-lfs.
-
-You can download pre-trained models and then decompress them (`ckpts.zip`) to the `ckpts` folder.
-
-To use the pre-trained models, you need to download our processed datasets and specify `--rel_path` to our processed `rel2id.json`.
-
-To evaluate by the pre-trained models, you can use above commands and specify `--ckpt_path` to specific model.
-
-
 In our setting, NYT, WebNLG, and NYT11-HRL achieve the best result on Epoch 86, 174, and 23 respectively.
 
 1\. NYT
@@ -194,14 +101,30 @@ If you use our code in your research, please cite our work:
 
 
 ```bibtex
-@inproceedings{li2021tdeer,
-   title={TDEER: An Efficient Translating Decoding Schema for Joint Extraction of Entities and Relations},
-   author={Li, Xianming and Luo, Xiaotian and Dong, Chenghao and Yang, Daichuan and Luan, Beidi and He, Zhen},
-   booktitle={Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing (EMNLP)},
-   year={2021}
+@inproceedings{li-etal-2021-tdeer,
+    title = "{TDEER}: An Efficient Translating Decoding Schema for Joint Extraction of Entities and Relations",
+    author = "Li, Xianming  and
+      Luo, Xiaotian  and
+      Dong, Chenghao  and
+      Yang, Daichuan  and
+      Luan, Beidi  and
+      He, Zhen",
+    booktitle = "Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing",
+    month = nov,
+    year = "2021",
+    address = "Online and Punta Cana, Dominican Republic",
+    publisher = "Association for Computational Linguistics",
+    url = "https://aclanthology.org/2021.emnlp-main.635",
+    pages = "8055--8064",
 }
+
 ```
 
+## Acknowledgement
+
+Some of our codes are inspired by [weizhepei/CasRel](https://github.com/weizhepei/CasRel). Thanks for their excellent work.
+
+
 ## Contact
 
 If you have any questions about the paper or code, you can

data/README.md

@@ -1,3 +1,3 @@
 We follow [weizhepei/CasRel](https://github.com/weizhepei/CasRel) to prepare datas.
 
-You could download our preprocessed datasets ([Google Drive]() | [Baidu NetDisk]()).
+Please use `git-lfs` to download the preprocessed data.