FEA/update readme and docstring

ShannonAI · Oct 3, 2020 · bbf4a29 · bbf4a29
1 parent 564e90e
commit bbf4a29
Show file tree

Hide file tree

Showing 19 changed files with 21 additions and 165 deletions.
diff --git a/README.md b/README.md
@@ -44,3 +44,23 @@ dataset path, bert model path and log path, respectively.
 and save the topk checkpoints to `default_root_dir`.
 
 To evaluate them, use `evaluate.py`
+
+
+## Results
+With training scripts in `scripts/reproduce`, one should be able to reproduce results for four datasets below:
+
+### Flat NER Datasets
+
+| Dataset | Zh-MSRA | Zh-OntoNotes4.0 | 
+|---|---|---|
+| Previous SOTA | 95.54  | 81.63 | 
+| Our method | **95.75** | **82.11** | 
+|  | **(+0.21)** | **(+0.48)** | 
+
+### Nested NER Datasets
+
+| Dataset | ACE 2004 | ACE 2005 | 
+|---|---|---|
+| Previous SOTA | 84.7 | 84.33 | 
+| Our method | **85.98** | **86.88** |
+|  | **(+1.28)** | **(+2.55)** |
diff --git a/datasets/collate_functions.py b/datasets/collate_functions.py
@@ -1,14 +1,4 @@
 # encoding: utf-8
-"""
-@author: Yuxian Meng
-@contact: yuxian_meng@shannonai.com
-
-@version: 1.0
-@file: collate_functions
-@time: 2020/9/12 14:21
-@desc: 
-
-"""
 
 import torch
 from typing import List

diff --git a/datasets/mrc_ner_dataset.py b/datasets/mrc_ner_dataset.py
@@ -1,14 +1,4 @@
 # encoding: utf-8
-"""
-@author: Yuxian Meng
-@contact: yuxian_meng@shannonai.com
-
-@version: 1.0
-@file: mrc_ner_dataset
-@time: 2020/9/6 14:27
-@desc: 
-
-"""
 
 
 import json

diff --git a/datasets/truncate_dataset.py b/datasets/truncate_dataset.py
@@ -1,14 +1,4 @@
 # encoding: utf-8
-"""
-@author: Yuxian Meng
-@contact: yuxian_meng@shannonai.com
-
-@version: 1.0
-@file: TruncateDatset
-@time: 2020/7/3 17:48
-
-    Truncate Datset, mostly used for debug
-"""
 
 from torch.utils.data import Dataset
 

diff --git a/evaluate.py b/evaluate.py
@@ -1,14 +1,5 @@
 # encoding: utf-8
-"""
-@author: Yuxian Meng
-@contact: yuxian_meng@shannonai.com
 
-@version: 1.0
-@file: evaluate
-@time: 2020/8/28 21:36
-@desc: 
-
-"""
 
 import os
 from pytorch_lightning import Trainer

diff --git a/loss/adaptive_dice_loss.py b/loss/adaptive_dice_loss.py
@@ -1,14 +1,4 @@
 # encoding: utf-8
-"""
-@author: Yuxian Meng
-@contact: yuxian_meng@shannonai.com
-
-@version: 1.0
-@file: adaptive_dice_loss
-@time: 2020/9/14 17:24
-@desc: todo
-
-"""
 
 
 import torch

diff --git a/loss/dice_loss.py b/loss/dice_loss.py
@@ -1,14 +1,4 @@
 # encoding: utf-8
-"""
-@author: Yuxian Meng
-@contact: yuxian_meng@shannonai.com
-
-@version: 1.0
-@file: dice_loss
-@time: 2020/9/9 10:55
-@desc: todo
-
-"""
 
 
 import torch

diff --git a/metrics/functional/query_span_f1.py b/metrics/functional/query_span_f1.py
@@ -1,14 +1,4 @@
 # encoding: utf-8
-"""
-@author: Yuxian Meng
-@contact: yuxian_meng@shannonai.com
-
-@version: 1.0
-@file: query_span_f1
-@time: 2020/9/6 20:05
-@desc: 
-
-"""
 
 
 import torch

diff --git a/metrics/query_span_f1.py b/metrics/query_span_f1.py
@@ -1,14 +1,4 @@
 # encoding: utf-8
-"""
-@author: Yuxian Meng
-@contact: yuxian_meng@shannonai.com
-
-@version: 1.0
-@file: query_span_f1
-@time: 2020/9/7 20:40
-@desc: 
-
-"""
 
 
 from pytorch_lightning.metrics.metric import TensorMetric

diff --git a/models/bert_query_ner.py b/models/bert_query_ner.py
@@ -1,14 +1,5 @@
 # encoding: utf-8
-"""
-@author: Yuxian Meng
-@contact: yuxian_meng@shannonai.com
 
-@version: 1.0
-@file: bert_query_ner
-@time: 2020/9/6 17:01
-@desc: 
-
-"""
 
 import torch
 import torch.nn as nn

diff --git a/models/classifier.py b/models/classifier.py
@@ -1,14 +1,4 @@
 # encoding: utf-8
-"""
-@author: Yuxian Meng
-@contact: yuxian_meng@shannonai.com
-
-@version: 1.0
-@file: classifier
-@time: 2020/9/6 17:02
-@desc: 
-
-"""
 
 
 import torch.nn as nn

diff --git a/models/query_ner_config.py b/models/query_ner_config.py
@@ -1,14 +1,4 @@
 # encoding: utf-8
-"""
-@author: Yuxian Meng
-@contact: yuxian_meng@shannonai.com
-
-@version: 1.0
-@file: query_ner_config
-@time: 2020/9/6 22:18
-@desc: 
-
-"""
 
 
 from transformers import BertConfig

diff --git a/ner2mrc/genia2mrc.py b/ner2mrc/genia2mrc.py
@@ -1,14 +1,4 @@
 # encoding: utf-8
-"""
-@author: Yuxian Meng
-@contact: yuxian_meng@shannonai.com
-
-@version: 1.0
-@file: genia2mrc.py
-@time: 2020/9/12 20:19
-@desc: 
-
-"""
 
 
 import os

diff --git a/ner2mrc/msra2mrc.py b/ner2mrc/msra2mrc.py
@@ -1,14 +1,5 @@
 # encoding: utf-8
-"""
-@author: Yuxian Meng
-@contact: yuxian_meng@shannonai.com
 
-@version: 1.0
-@file: msra2mrc
-@time: 2020/9/9 20:56
-@desc: Convert MSRA raw data to MRC format
-
-"""
 
 import os
 from utils.bmes_decode import bmes_decode

diff --git a/scripts/reproduce/ace05.sh b/scripts/reproduce/ace05.sh
@@ -34,4 +34,4 @@ python trainer.py \
 --warmup_steps $WARMUP \
 --max_length $MAXLEN \
 --gradient_clip_val $MAXNORM \
---optimizer "sgd"
+--optimizer "adamw"
diff --git a/trainer.py b/trainer.py
@@ -1,14 +1,4 @@
 # encoding: utf-8
-"""
-@author: Yuxian Meng
-@contact: yuxian_meng@shannonai.com
-
-@version: 1.0
-@file: trainer
-@time: 2020/9/6 14:26
-@desc: pytorch-lightning trainer
-
-"""
 
 
 import argparse

diff --git a/utils/bmes_decode.py b/utils/bmes_decode.py
@@ -1,14 +1,4 @@
 # encoding: utf-8
-"""
-@author: Yuxian Meng
-@contact: yuxian_meng@shannonai.com
-
-@version: 1.0
-@file: bmes_decode
-@time: 2020/9/8 10:55
-@desc: 
-
-"""
 
 
 from typing import Tuple, List

diff --git a/utils/get_parser.py b/utils/get_parser.py
@@ -1,13 +1,5 @@
 # encoding: utf-8
-"""
-@author: Yuxian Meng
-@contact: yuxian_meng@shannonai.com
 
-@version: 1.0
-@file: get_parser
-@time: 2020/7/9 15:48
-
-"""
 
 import argparse
 

diff --git a/utils/radom_seed.py b/utils/radom_seed.py
@@ -1,14 +1,5 @@
 # encoding: utf-8
-"""
-@author: Yuxian Meng
-@contact: yuxian_meng@shannonai.com
 
-@version: 1.0
-@file: radom_seed
-@time: 2020/7/9 15:53
-
-    这一行开始写关于本文件的说明与解释
-"""
 
 import numpy as np
 import torch