Add files via upload

[shaoyuheng1]

No description provided.

[gemini-code-assist]

Code Review

This pull request significantly enhances the data annotation pipeline by introducing advanced example selection strategies, including similarity-based, diversity-based, and contrastive learning approaches. It adds support for task-specific guidance, Chain-of-Thought reasoning, self-consistency voting, and social media sentiment analysis. Additionally, the implementation now supports batch processing and parallel annotation requests. Feedback focuses on improving code portability by removing hardcoded absolute paths, cleaning up formatting artifacts (extraneous backslashes), refining broad exception handling, and reducing duplication in API configurations.

[gemini-code-assist]

Hardcoding absolute paths specific to a local environment (/root/flagos/...) makes the code non-portable. It is better to use relative paths or environment variables to define these directories.

Suggested change

	DATA_DIR = '/root/flagos/OpenSeek/openseek/competition/LongContext-ICL-Annotation/data'
	OUTPUT_DIR = '/root/flagos/OpenSeek/openseek/competition/LongContext-ICL-Annotation/outputs'
	DATA_DIR = './data'
	OUTPUT_DIR = './outputs'

[gemini-code-assist]

The default values for --log_path_prefix and --tokenizer_path are hardcoded to absolute paths. These should be changed to relative paths or made configurable to ensure the script can run in different environments.

[gemini-code-assist]

The tokenizer path is hardcoded to an absolute path. This should be passed as a parameter or retrieved from a configuration to maintain portability.

    tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen3-4B", trust_remote_code=True)

[gemini-code-assist]

There are numerous occurrences of backslashes (\) on empty lines throughout this file (e.g., lines 12-14, 29-32, etc.). This appears to be a formatting artifact or error that should be removed to improve code readability.

[gemini-code-assist]

The bare except: block is too broad. It is recommended to catch specific exceptions (e.g., Exception or even better, specific sklearn errors) to avoid catching system signals like KeyboardInterrupt.

    except Exception:
        tfidf_matrix = vectorizer.transform(texts)

[gemini-code-assist]

The API base URL and model name are hardcoded and duplicated in multiple functions (e.g., annotate_ascend and annotate_with_self_consistency). These should be defined as constants or passed as arguments.