From 970cc1fb6f35af89e9236ef4a4b6c91e6bb5179c Mon Sep 17 00:00:00 2001
From: louishsu <is.louishsu@foxmail.com>
Date: Sun, 10 Oct 2021 12:53:08 +0800
Subject: [PATCH] 2021/10/10 nezha-legal-fgm2.0-lsr0.1_f82.73_p76.19_r90.49

---
 main.py             |  8 ++++++-
 main_local.py       |  8 ++++++-
 scripts/run_span.sh | 56 +++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 70 insertions(+), 2 deletions(-)

diff --git a/main.py b/main.py
index 3f7c184..6bddf88 100644
--- a/main.py
+++ b/main.py
@@ -66,7 +66,13 @@ def main():
     # n_splits = 5
     # seed=42
     # --------------------------
-    version = "nezha-legal-100k-fgm1.0-lsr0.1"
+    # version = "nezha-legal-100k-fgm1.0-lsr0.1"
+    # model_type = "nezha_span"
+    # dataset_name = "cail_ner"
+    # n_splits = 5
+    # seed=42
+    # --------------------------
+    version = "nezha-legal-fgm2.0-lsr0.1"
     model_type = "nezha_span"
     dataset_name = "cail_ner"
     n_splits = 5
diff --git a/main_local.py b/main_local.py
index 0859bf1..11a3225 100644
--- a/main_local.py
+++ b/main_local.py
@@ -66,7 +66,13 @@ def main():
     # n_splits = 5
     # seed=42
     # --------------------------
-    version = "nezha-legal-fgm1.0-lsr0.1-ema3"
+    # version = "nezha-legal-fgm1.0-lsr0.1-ema3"
+    # model_type = "nezha_span"
+    # dataset_name = "cail_ner"
+    # n_splits = 5
+    # seed=42
+    # --------------------------
+    version = "nezha-legal-fgm2.0-lsr0.1"
     model_type = "nezha_span"
     dataset_name = "cail_ner"
     n_splits = 5
diff --git a/scripts/run_span.sh b/scripts/run_span.sh
index e748b4e..9a19fc0 100644
--- a/scripts/run_span.sh
+++ b/scripts/run_span.sh
@@ -565,6 +565,62 @@ python run_span.py \
     --seed=42
 done
 
+# Further-pretrain LSR
+for k in 0 1 2 3 4
+do
+python run_span.py \
+    --version=nezha-legal-fgm2.0-lsr0.1-fold${k} \
+    --data_dir=./data/ner-ctx0-5fold-seed42/ \
+    --train_file=train.${k}.json \
+    --dev_file=dev.${k}.json \
+    --test_file=dev.${k}.json \
+    --model_type=nezha_span \
+    --model_name_or_path=/home/louishsu/NewDisk/Code/CAIL2021/nezha-legal-cn-base-wwm/ \
+    --do_train \
+    --overwrite_output_dir \
+    --evaluate_during_training \
+    --evaluate_each_epoch \
+    --save_best_checkpoints \
+    --max_span_length=40 \
+    --width_embedding_dim=128 \
+    --train_max_seq_length=512 \
+    --eval_max_seq_length=512 \
+    --do_lower_case \
+    --per_gpu_train_batch_size=8 \
+    --per_gpu_eval_batch_size=16 \
+    --gradient_accumulation_steps=2 \
+    --learning_rate=5e-5 \
+    --other_learning_rate=1e-3 \
+    --num_train_epochs=8.0 \
+    --warmup_proportion=0.1 \
+    --do_fgm --fgm_epsilon=2.0 \
+    --loss_type=lsr --label_smooth_eps=0.1 \
+    --seed=42
+done
+# main_local
+# avg
+# {'p': 0.9046767380798356, 'r': 0.8917144893289825, 'f': 0.8981488477521723}
+# 犯罪嫌疑人
+# {'p': 0.963254593175853, 'r': 0.9653411728299551, 'f': 0.9642967542503864}
+# 受害人
+# {'p': 0.9311955168119551, 'r': 0.9623552123552124, 'f': 0.9465189873417722}
+# 被盗货币
+# {'p': 0.8584686774941995, 'r': 0.8087431693989071, 'f': 0.8328643781654473}
+# 物品价值
+# {'p': 0.9754571703561117, 'r': 0.9698564593301435, 'f': 0.9726487523992322}
+# 盗窃获利
+# {'p': 0.8848484848484849, 'r': 0.9106029106029107, 'f': 0.8975409836065574}
+# 被盗物品
+# {'p': 0.8189794091316025, 'r': 0.7912125929769936, 'f': 0.8048565898293155}
+# 作案工具
+# {'p': 0.7943166441136671, 'r': 0.7986394557823129, 'f': 0.796472184531886}
+# 时间
+# {'p': 0.9484612532443456, 'r': 0.9251356238698011, 'f': 0.9366532405712193}
+# 地点
+# {'p': 0.874439461883408, 'r': 0.8316747227750924, 'f': 0.8525211308656367}
+# 组织机构
+# {'p': 0.8808618504435995, 'r': 0.8622828784119106, 'f': 0.8714733542319749}
+
 # Further-pretrain 100k steps, LSR
 for k in 0 1 2 3 4
 do