Fetching metadata from the HF Docker repository... miras_dev
changes
c12c960 - article_cache Last updated code by Ilyas
- data codes with mbc mode
- scripts Improve FACTScore precision: negative assertion filtering, dense retrieval, LLM synthesis mode
- test_ownerhsip_history upds
- 1.66 kB first commit
- 230 Bytes Clean repo: remove old results, caches, and redundant data
- 26.7 kB first commit
- 11.5 kB Code review: fix README inaccuracies, remove dead code, consolidate get_llm()
- 24.4 kB Add architecture diagrams, mode comparison, and BBC state flow
- 426 Bytes first commit
- 7.57 kB Comprehensive code review: fix MBFC methodology, add rating systems, cleanup
- 19.9 kB Align MBC implementation with paper and add template-based evaluation
- 53.4 kB Code review: fix README inaccuracies, remove dead code, consolidate get_llm()
- 28.8 kB Comprehensive code review: fix MBFC methodology, add rating systems, cleanup
- 2.93 kB Fix README factual errors: align scoring tables, formulas, and metrics with source code
- 25.3 kB Add benchmark log analysis script for leakage, QA, performance, and fact stats
- 18.9 kB Set default model to gpt-5-mini-2025-08-07 across all components
- 4.98 kB Add benchmark enhancements: article caching, per-fact recall detail, logging, and 100-word truncation
- 23.3 kB Fix search mode data leakage: expand sanitization regex, sanitize debug fields
- 11 kB Set default model to gpt-5-mini-2025-08-07 across all components
- 6.89 kB codes with mbc mode
- 12.5 kB Fix Wikipedia API 403: add User-Agent header to all Wikipedia requests
- 5.57 kB Add single-outlet evaluation script and sample hybrid output
- 1.02 MB Improve system mode bias/factuality MAE and FACTScore precision
- 2.34 MB first commit
- 74 kB Set default model to gpt-5-mini-2025-08-07 across all components
- 15.2 kB upd extract_subset
- 3.28 kB results of 6 modes in 30 samples
- 103 kB first commit
- 7.97 kB first commit
- 3.16 kB Set default model to gpt-5-mini-2025-08-07 across all components
- 11.8 kB Improve system/hybrid to top benchmark positions
- 40.9 kB Improve FACTScore precision: negative assertion filtering, dense retrieval, LLM synthesis mode
- 3.5 kB Add single-outlet evaluation script and sample hybrid output
- 380 Bytes add results
- 8.25 kB Set default model to gpt-5-mini-2025-08-07 across all components
- 10.1 kB Set default model to gpt-5-mini-2025-08-07 across all components
- 20.2 kB Save full intermediate metric data to JSONL and add offline recomputation
- 110 kB Set default model to gpt-5-mini-2025-08-07 across all components
- 7.43 kB Set default model to gpt-5-mini-2025-08-07 across all components
- 264 Bytes first commit
- 7.71 kB upds
- 104 kB Set default model to gpt-5-mini-2025-08-07 across all components
- 2.44 kB new results
- 21.3 kB Save full intermediate metric data to JSONL and add offline recomputation
- 2.01 kB first commit
- 45.7 kB Add LLM score calibration and --cached-only filter for system mode
- 18.1 kB Improve system & hybrid mode accuracy across all metrics
- 12.4 kB Fix ModuleNotFoundError: change 'from ddgs import DDGS' to 'from duckduckgo_search import DDGS'
- 3.7 kB first commit
