Add files via upload

openseek/competition/LongContext-ICL-Annotation/src/main.py

@@ -1,92 +1,159 @@
-import json, os, argparse
-from tqdm import tqdm, trange
-from transformers import AutoTokenizer
-
-# from method import build_prompt, select_examples, annotate
-
-from method import build_prompt, select_examples
-
-from method import annotate_nvidia as annotate # For Nvidia GPU
-# from method import annotate_ascend as annotate # For Huawei Ascend
-
-TASK_FILES = {
-    1: './data/openseek-1_closest_integers.json',
-    2: './data/openseek-2_count_nouns_verbs.json',
-    3: './data/openseek-3_collatz_conjecture.json',
-    4: './data/openseek-4_conala_concat_strings.json',
-    5: './data/openseek-5_semeval_2018_task1_tweet_sadness_detection.json',
-    6: './data/openseek-6_mnli_same_genre_classification.json',
-    7: './data/openseek-7_jeopardy_answer_generation_all.json',
-    8: '../data/openseek-8_kernel_generation.json',
-}
-
-def parser_args():
-    parser = argparse.ArgumentParser()
-    parser.add_argument('--task_id', type=int, required=True,
-                        help='Task ID to evaluate, should be in [1, 7].')
-    parser.add_argument('--max_input_length', type=int, default=10_000,
-                        help='Maximum input length for the model.')
-    parser.add_argument('--log_path_prefix', type=str, 
-                        default='../outputs/',
-                        help='Prefix path to save the evaluation logs.')
-    parser.add_argument('--tokenizer_path', type=str,
-                        default='/share/project/wuhaiming/spaces/data_agent/OpenSeek-main/openseek/competition/LongContext-ICL-Annotation/src/Qwen3-4B')
-    args = parser.parse_args()
-    return args
-
-def evaluate(task_id:int, 
-             qwen_tokenizer:AutoTokenizer,
-             max_input_length:int=128_000,
-             log_path_prefix:str='./outputs/'
-        )->float:
-    assert task_id in [i for i in range(1, 9)],\
-        f"task_id should be in [1, 8], but got {task_id}."
-
-    task_file = TASK_FILES[task_id]
-    with open(task_file, 'r') as f:
-        task_dict = json.load(f)
-
-    task_name = task_dict['task_name']
-    task_description = task_dict['Definition'][0]
-    icl_examples = task_dict['examples'][:100]
-    test_samples = task_dict['test_samples']
-
-    version = 1
-    output_file = f'{log_path_prefix}openseek-{task_id}-v{version}.jsonl'
-    output_path = os.path.dirname(output_file)
-    os.makedirs(output_path, exist_ok=True)
-    while os.path.exists(output_file):
-        version += 1
-        output_file = f'{log_path_prefix}openseek-{task_id}-v{version}.jsonl'
-    with open(output_file, 'w') as f:
-        pass
-
-    examples_str = None
-    for test_sample in tqdm(test_samples, desc=f'Evaluation on Task {task_id}: {task_name}'):
-        test_record = dict()
-
-        test_sample_id = test_sample['id']
-        test_record['test_sample_id'] = test_sample_id
-
-
-        text2annotate = test_sample['input']
-        prompt = build_prompt(task_description, text2annotate)
-        if examples_str is None:
-            examples_str = select_examples(icl_examples, task_description, text2annotate)
-        input_prompt = prompt.replace("[[EXAMPLES]]\n\n", examples_str+'\n\n')
-
-        # tokenized_input = qwen_tokenizer(input_prompt, return_tensors="pt")
-        # if tokenized_input['input_ids'].shape[1] > max_input_length:
-        #     test_record['prediction'] = None
-        # else:
-        #     prediction = annotate(input_prompt)
-        #     test_record['prediction'] = prediction
-        prediction = annotate(input_prompt)
-        test_record['prediction'] = prediction
-        with open(output_file, 'a') as f:
-            f.write(json.dumps(test_record)+'\n')
-
-if __name__ == '__main__':
-    args = parser_args()
-    qwen_tokenizer = AutoTokenizer.from_pretrained(args.tokenizer_path)
+import json, os, argparse
+from tqdm import tqdm, trange
+from transformers import AutoTokenizer
+
+# from method import build_prompt, select_examples, annotate
+
+from method import build_prompt, select_examples
+
+# from method import annotate_nvidia as annotate # For Nvidia GPU
+from method import annotate_ascend as annotate # For Huawei Ascend
+from method import annotate_batch
+from method import annotate_with_self_consistency                         
+
+DATA_DIR = '/root/flagos/OpenSeek/openseek/competition/LongContext-ICL-Annotation/data'
+OUTPUT_DIR = '/root/flagos/OpenSeek/openseek/competition/LongContext-ICL-Annotation/outputs'
+
+TASK_FILES = {
+    1: f'{DATA_DIR}/openseek-1_closest_integers.json',
+    2: f'{DATA_DIR}/openseek-2_count_nouns_verbs.json',
+    3: f'{DATA_DIR}/openseek-3_collatz_conjecture.json',
+    4: f'{DATA_DIR}/openseek-4_conala_concat_strings.json',
+    5: f'{DATA_DIR}/openseek-5_semeval_2018_task1_tweet_sadness_detection.json',
+    6: f'{DATA_DIR}/openseek-6_mnli_same_genre_classification.json',
+    7: f'{DATA_DIR}/openseek-7_jeopardy_answer_generation_all.json',
+    8: f'{DATA_DIR}/openseek-8_kernel_generation.json',
+}
+
+def parser_args():
+    parser = argparse.ArgumentParser()
+    parser.add_argument('--task_id', type=int, required=True,
+                        help='Task ID to evaluate, should be in [1, 7].')
+    parser.add_argument('--max_input_length', type=int, default=10_000,
+                        help='Maximum input length for the model.')
+    parser.add_argument('--log_path_prefix', type=str, 
+                        default='/root/flagos/OpenSeek/openseek/competition/LongContext-ICL-Annotation/outputs/',
+                        help='Prefix path to save the evaluation logs.')
+    parser.add_argument('--tokenizer_path', type=str,
+                        default='/root/flagos/Qwen3-4B')
+    args = parser.parse_args()
+    return args
+
+def evaluate(task_id:int, 
+             qwen_tokenizer:AutoTokenizer,
+             max_input_length:int=128_000,
+             log_path_prefix:str='./outputs/'
+        )->float:
+    assert task_id in [i for i in range(1, 9)],\
+        f"task_id should be in [1, 8], but got {task_id}."
+
+    task_file = TASK_FILES[task_id]
+    with open(task_file, 'r') as f:
+        task_dict = json.load(f)
+
+    task_name = task_dict['task_name']
+    task_description = task_dict['Definition'][0]
+    icl_examples = task_dict['examples'][:50]
+    test_samples = task_dict['test_samples']
+
+    version = 1
+    output_file = f'{log_path_prefix}openseek-{task_id}-v{version}.jsonl'
+    output_path = os.path.dirname(output_file)
+    os.makedirs(output_path, exist_ok=True)
+    while os.path.exists(output_file):
+        version += 1
+        output_file = f'{log_path_prefix}openseek-{task_id}-v{version}.jsonl'
+    with open(output_file, 'w') as f:
+        pass
+
+    examples_str = None
+    batch_size = 8
+    prompts_batch = []
+    sample_ids_batch = []
+
+    # Task 8 is code generation, needs more tokens and different post-processing
+    max_tokens = 1024 if task_id == 8 else 256
+    use_count_answer = False if task_id == 8 else True
+
+
+
+    use_self_consistency = (task_id == 6 or (1 <= task_id <= 4))
+
+
+    use_social_media_enhancement = (task_id == 5)
+
+    for test_sample in tqdm(test_samples, desc=f'Evaluation on Task {task_id}: {task_name}'):
+        test_sample_id = test_sample['id']
+        text2annotate = test_sample['input']
+
+
+        prompt = build_prompt(task_description, text2annotate, task_id=task_id, use_social_media_enhancement=use_social_media_enhancement)
+        if examples_str is None:
+            # Task 8 is code generation task
+            is_code_generation = (task_id == 8)
+
+
+            use_cot = (1 <= task_id <= 4)                
+
+            balance_sentiment = (task_id == 5)
+
+            use_importance_weighting = (1 <= task_id <= 4)
+
+            use_contrastive_learning = (1 <= task_id <= 4 or task_id == 6)
+            examples_str = select_examples(icl_examples, task_description, text2annotate, 
+                                          is_code_generation=is_code_generation,
+                                          use_task_aware=True, task_id=task_id,
+                                          use_quality_filter=True, quality_threshold=0.5,
+                                          use_diversity=False, use_similarity=False,
+                                          use_cot=use_cot,
+                                          balance_sentiment=balance_sentiment,
+                                          use_importance_weighting=use_importance_weighting,
+                                          use_contrastive_learning=use_contrastive_learning)
+        input_prompt = prompt.replace("[[EXAMPLES]]\n\n", examples_str+'\n\n')
+
+
+        if use_self_consistency:
+
+            final_answer, confidence, all_predictions = annotate_with_self_consistency(
+                input_prompt, 
+                num_samples=5,          
+                temperature_range=[0.7, 0.85, 1.0, 1.1, 1.2],
+                max_tokens=max_tokens,
+                task_id=task_id,
+                confidence_threshold=0.4
+            )
+            test_record = {
+                'test_sample_id': test_sample_id, 
+                'prediction': final_answer,
+                'confidence': confidence,
+                'all_predictions': all_predictions
+            }
+            with open(output_file, 'a') as f:
+                f.write(json.dumps(test_record)+'\n')
+        else:
+
+            prompts_batch.append(input_prompt)
+            sample_ids_batch.append(test_sample_id)
+
+            # Process batch when full
+            if len(prompts_batch) >= batch_size:
+                results = annotate_batch(prompts_batch, num_workers=4, max_tokens=max_tokens, use_count_answer=use_count_answer, task_id=task_id)
+                for sid, (pred, _) in zip(sample_ids_batch, results):
+                    test_record = {'test_sample_id': sid, 'prediction': pred}
+                    with open(output_file, 'a') as f:
+                        f.write(json.dumps(test_record)+'\n')
+                prompts_batch = []
+                sample_ids_batch = []
+
+
+    if not use_self_consistency and prompts_batch:
+        results = annotate_batch(prompts_batch, num_workers=4, max_tokens=max_tokens, use_count_answer=use_count_answer, task_id=task_id)
+        for sid, (pred, _) in zip(sample_ids_batch, results):
+            test_record = {'test_sample_id': sid, 'prediction': pred}
+            with open(output_file, 'a') as f:
+                f.write(json.dumps(test_record)+'\n')
+
+if __name__ == '__main__':
+    args = parser_args()
+    qwen_tokenizer = AutoTokenizer.from_pretrained(args.tokenizer_path)
     evaluate(args.task_id, qwen_tokenizer, args.max_input_length, args.log_path_prefix)