init

experiments/main.sh

@@ -1,7 +1,7 @@
 MODEL="cardiffnlp/twitter-roberta-base"
 
 MODEL="jhu-clsp/bernice" # nerd[ukri], topic[hawk]
-MODEL="roberta-base"  # nerd[hawk], topic [hawk]
+MODEL="roberta-base"  # ner[hawk], nerd[hawk], topic [hawk]
 MODEL="vinai/bertweet-base" # nerd [stone], topic [ukri]
 
 
@@ -27,7 +27,7 @@ python model_finetuning_ner.py -m "${MODEL}" -d "ner_random3_seed2"
 # NERD
 python model_finetuning_nerd.py -m "${MODEL}" -d "nerd_temporal"
 
-python model_finetuning_nerd.py -m "${MODEL}" -d "nerd_random0_seed0"
+python model_finetuning_nerd.py -m "${MODEL}" -d "nerd_random0_seed0" --skip-train --skip-test
 python model_finetuning_nerd.py -m "${MODEL}" -d "nerd_random1_seed0"
 python model_finetuning_nerd.py -m "${MODEL}" -d "nerd_random2_seed0"
 python model_finetuning_nerd.py -m "${MODEL}" -d "nerd_random3_seed0"

experiments/model_finetuning_ner.py

@@ -107,8 +107,7 @@ def main(
             truncation=True,
             is_split_into_words=True,
             padding="max_length",
-            max_length=128 if model in ["jhu-clsp/bernice", "vinai/bertweet-base"] else 256
-        )
+            max_length=128)
         all_labels = examples["gold_label_sequence"]
         new_labels = []
         for ind, labels in enumerate(all_labels):

experiments/model_finetuning_topic.py

@@ -94,7 +94,7 @@ def main(
             [preprocess(model, t) for t in x["text"]],
             padding="max_length",
             truncation=True,
-            max_length=128 if model in ["jhu-clsp/bernice", "vinai/bertweet-base"] else 256),
+            max_length=128),
         batched=True
     )
     tokenized_datasets = tokenized_datasets.rename_column("gold_label_list", "label")

statistics.py

@@ -0,0 +1,17 @@
+import pandas as pd
+from datasets import load_dataset
+
+stats = []
+for i in ["topic_temporal", "nerd_temporal", "ner_temporal"]:
+    for s in ["train", "validation", "test"]:
+    # for s in ["train", "validation", "test", "test_1", "test_2", "test_3", "test_4"]:
+        dataset = load_dataset("tweettemposhift/tweet_temporal_shift", i, split=s)
+        df = dataset.to_pandas()
+        stats.append({
+            "data": i,
+            "split": s,
+            "size": len(dataset),
+            "date": f'{str(pd.to_datetime(df.date).max()).split(" ")[0]}/{str(pd.to_datetime(df.date).max()).split(" ")[0]}',
+         })
+df = pd.DataFrame(stats)
+