MCPcopy Create free account

hub / github.com/alibaba/vstyle / functions

Functions23 in github.com/alibaba/vstyle

⨍ Functions 23◇ Types & classes 0
↓ 5 callersFunctionsafe_float
(x: Any)
lalm_eval/gemini_eval.py:119
↓ 4 callersFunctionensure_dir
(p: Path)
lalm_eval/gemini_eval.py:125
↓ 2 callersFunctioncompute_individual_correlations
Compute correlations between individual annotators and model
human_align/compute_model_human_spearman_r.py:73
↓ 2 callersFunctioncompute_overall_correlation
Compute overall correlation between human average and model
human_align/compute_model_human_spearman_r.py:117
↓ 2 callersFunctioncompute_pairwise_human_correlations
Compute pairwise correlations between human annotators
human_align/compute_model_human_spearman_r.py:93
↓ 2 callersFunctionfilter_by_category
Filter data dictionary by category
human_align/compute_model_human_spearman_r.py:143
↓ 2 callersFunctionget_major_category
Extract major category from ability string
human_align/compute_model_human_spearman_r.py:137
↓ 2 callersFunctionparse_score
Extract score from model reply: 1. Find all segments like [[...]] 2. Search backwards for the first value that can be converted to float
lalm_eval/gemini_eval.py:128
↓ 1 callersFunctionanalyze_language
Analyze correlations for a single language
human_align/compute_model_human_spearman_r.py:302
↓ 1 callersFunctionanalyze_scores
(scored_jsonl: Path, out_dir: Path)
lalm_eval/gemini_eval.py:412
↓ 1 callersFunctioncall_gemini_api
Call model using official Gemini API
lalm_eval/gemini_eval.py:65
↓ 1 callersFunctioncompute_category_correlations
Compute correlations for each major category
human_align/compute_model_human_spearman_r.py:153
↓ 1 callersFunctionconstruct_prompt
(template: str, instruction: str, ability: str)
lalm_eval/gemini_eval.py:59
↓ 1 callersFunctioneval_all_samples
( root_dir: Path, metadata_path: Path, prompts_dir: Path, out_dir: Path, max_per_ability:
lalm_eval/gemini_eval.py:312
↓ 1 callersFunctionget_ordered_abilities
Get abilities in the order defined by SUB_CATS
lalm_eval/gemini_eval.py:403
↓ 1 callersFunctionload_human_scores
Load human annotation jsonl files, return: - human_scores: {annotator: {key: mos}} - ability_map: {key: ability} where key = f"{did}_{iid
human_align/compute_model_human_spearman_r.py:8
↓ 1 callersFunctionload_model_scores
Load model evaluation jsonl file, return: - model_scores: {key: score} - ability_map: updated {key: ability}
human_align/compute_model_human_spearman_r.py:43
↓ 1 callersFunctionload_processed_ids_and_clean_failures
Read existing metadata_with_score.jsonl, return set of successfully processed IDs, and clean failed records in-place Returns:
lalm_eval/gemini_eval.py:223
↓ 1 callersFunctionload_prompts
(prompts_dir: Path)
lalm_eval/gemini_eval.py:51
↓ 1 callersFunctionmain
()
lalm_eval/gemini_eval.py:609
↓ 1 callersFunctionprint_summary_table
Print a summary table comparing results across languages
human_align/compute_model_human_spearman_r.py:201
↓ 1 callersFunctionremove_duplicates_from_tasks
Remove duplicate samples based on ID, keeping the first occurrence Args: tasks: List of sample dictionaries Returns
lalm_eval/gemini_eval.py:278
Functionevaluate_one
Evaluate single sample and write complete output to txt file
lalm_eval/gemini_eval.py:146