Fetching metadata from the HF Docker repository...
-
Step 1: research foundation (RESEARCH.md)
devadit15 +1: Claude Fable 5 Every parameter and template traced to named sources: SummEval, SummaC,
RAGAS, TruLens triad, G-Eval, MT-Bench biases, HELM, IFEval, FActScore,
classic readability/lexical stats, HumanEval; prompting guidance from
Anthropic/OpenAI/Google docs plus CoT, self-consistency, few-shot
selection, and lost-in-the-middle papers. Locks model checkpoints:
bge-small-en-v1.5 + nli-deberta-v3-small.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>