Upload project files

Text_encoder/model_best/config.json

@@ -0,0 +1,25 @@
+{
+  "_name_or_path": "uie_base_en",
+  "architectures": [
+    "UIE"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "classifier_dropout": null,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-12,
+  "max_position_embeddings": 512,
+  "model_type": "bert",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "pad_token_id": 0,
+  "position_embedding_type": "absolute",
+  "torch_dtype": "float32",
+  "transformers_version": "4.20.0",
+  "type_vocab_size": 4,
+  "use_cache": true,
+  "vocab_size": 30522
+}

Text_encoder/model_best/pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:afea4b2ea4e7389c794ed4d71e169bb00284abb83d36e3c89f7883203c7887b0
+size 456930115

Text_encoder/model_best/special_tokens_map.json

@@ -0,0 +1,7 @@
+{
+  "cls_token": "[CLS]",
+  "mask_token": "[MASK]",
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "unk_token": "[UNK]"
+}

Text_encoder/model_best/tokenizer_config.json

@@ -0,0 +1,15 @@
+{
+  "cls_token": "[CLS]",
+  "do_basic_tokenize": true,
+  "do_lower_case": true,
+  "mask_token": "[MASK]",
+  "name_or_path": "uie_base_en",
+  "never_split": null,
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "special_tokens_map_file": "uie_base_en/special_tokens_map.json",
+  "strip_accents": null,
+  "tokenize_chinese_chars": true,
+  "tokenizer_class": "BertTokenizer",
+  "unk_token": "[UNK]"
+}

checkpoints/masc.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2d0e1703c7c4c6c41b80ba2f83aff61be6a3803b1deec53f87088cc4f4387924
+size 1223522938

checkpoints/mate.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:894fe8959e077a45a3d50fe1880abe88e7eb07db00ce4cec2870130d129fa654
+size 1223403721

eval.sh

@@ -1,15 +1,22 @@
 #!/usr/bin/env bash 
-export CUDA_VISIBLE_DEVICES="1"
+# Evaluation script for DASCO models
+# Supports MATE, MASC, and MABSA evaluation
 
+export CUDA_VISIBLE_DEVICES="0"
+
+# ============================================
 # MATE evaluation
+# ============================================
 
-CHECKPOINT_DIR="./checkpoints/MATE_2017"
-TEST_DATA="./finetune_dataset/twitter17/test"
+CHECKPOINT_DIR="./checkpoints/MATE_custom"
+TEST_DATA="./finetune_dataset/custom/test"
 
 best_stats_values=(0 0 0 0 0 0 "None")  # [Correct, Label, Prediction, Accuracy, Recall, F1, Model]
 declare -r COR=0 LABEL=1 PRED=2 ACC=3 REC=4 F1=5 MODEL=6
  
 for model in "${CHECKPOINT_DIR}"/*.pt; do 
+    [ -f "$model" ] || continue  # Skip if no .pt files found
+    
     output=$(python eval_tools.py  \
         --MATE_model "${model}" \
         --test_ds "${TEST_DATA}" \
@@ -47,25 +54,26 @@ for model in "${CHECKPOINT_DIR}"/*.pt; do
     fi
 done 
 
-echo -e "\nBest Model: ${best_stats_values[$MODEL]}"
+echo -e "\n========== MATE Best Results =========="
+echo "Best Model: ${best_stats_values[$MODEL]}"
 echo "F1      : ${best_stats_values[$F1]}"
 echo "Accuracy: ${best_stats_values[$ACC]}"
 echo "Recall  : ${best_stats_values[$REC]}"
-echo "Correct : ${best_stats_values[$COR]}"
-echo "Label   : ${best_stats_values[$LABEL]}"
-echo "Prediction: ${best_stats_values[$PRED]}"
-
 
 
-# MASC evaluation
+# ============================================
+# MASC evaluation (uncomment to use)
+# ============================================
 
-# CHECKPOINT_DIR="./checkpoints/MASC_2017"
-# TEST_DATA="./finetune_dataset/twitter17/test"
+# CHECKPOINT_DIR="./checkpoints/MASC_custom"
+# TEST_DATA="./finetune_dataset/custom/test"
 
 # best_stats_values=(0 0 0 0 0 "None")  # [Correct, Label, Prediction, Accuracy, Macro_F1, Model]
 # declare -r COR=0 LABEL=1 PRED=2 ACC=3 MacroF1=4 MODEL=5
  
 # for model in "${CHECKPOINT_DIR}"/*.pt; do 
+#     [ -f "$model" ] || continue
+#     
 #     output=$(python eval_tools.py  \
 #         --MASC_model "${model}" \
 #         --test_ds "${TEST_DATA}" \
@@ -88,7 +96,7 @@ echo "Prediction: ${best_stats_values[$PRED]}"
 
 #     if [[ "${f1:-0}" =~ ^[0-9.]+$ ]]; then
 #         is_better=$(awk -v f1="$f1" -v best="${best_stats_values[$MacroF1]}" 'BEGIN { print (f1 > best) ? 1 : 0 }')
-        
+#         
 #         if [ "$is_better" -eq 1 ]; then
 #             best_stats_values[$COR]=${correct:-0}
 #             best_stats_values[$LABEL]=${label:-0}
@@ -100,20 +108,20 @@ echo "Prediction: ${best_stats_values[$PRED]}"
 #     fi
 # done 
 
-# echo -e "\nBest Model: ${best_stats_values[$MODEL]}"
-# echo "F1      : ${best_stats_values[$MacroF1]}"
+# echo -e "\n========== MASC Best Results =========="
+# echo "Best Model: ${best_stats_values[$MODEL]}"
+# echo "Macro F1: ${best_stats_values[$MacroF1]}"
 # echo "Accuracy: ${best_stats_values[$ACC]}"
-# echo "Correct : ${best_stats_values[$COR]}"
-# echo "Label   : ${best_stats_values[$LABEL]}"
-# echo "Prediction: ${best_stats_values[$PRED]}"
 
 
+# ============================================
+# MABSA evaluation (uncomment to use)
+# ============================================
 
-# MABSA evaluation
 # python eval_tools.py \
-#    --MATE_model ./DASCO/checkpoints/MATE_2017/best_f1:94.933.pt \
-#    --MASC_model ./DASCO/checkpoints/MASC_2017/best_f1:77.616.pt \
-#    --test_ds ./finetune_dataset/twitter17/test \
+#    --MATE_model ./checkpoints/MATE_custom/best_f1:XX.XXX.pt \
+#    --MASC_model ./checkpoints/MASC_custom/best_f1:XX.XXX.pt \
+#    --test_ds ./finetune_dataset/custom/test \
 #    --task MABSA \
 #    --gcn_layers 4 \
 #    --device cuda:0