Momentum, Commentary Tone, and Earnings Surprise in LLM Financial Forecasts
Resumo indisponível. Consulte o paper original.
PAPER / ARXIV:2610.08882
Alina Khaybullina
RESUMO
FinVector-Market-4B adapts Qwen/Qwen3.5-4B with rank-16 LoRA on a 22,000-example corpus for structured financial tasks. We evaluate the base and adapted models on the same 600-example benchmark under implicit and explicit JSON-schema contracts. Supplying the schema alone raises base-model JSON validity from 0% to 91.3%. Under matched explicit prompting, scores improve substantially across FinQA exact match, calculator-expression correctness, scenario branch-label agreement, and implication-direction agreement. The results show that compact financial domain adaptation can produce substantial task-specific gains beyond output-format learning under matched prompting.
NO MESMO MAPA
Resumo indisponível. Consulte o paper original.
Resumo indisponível. Consulte o paper original.
Resumo indisponível. Consulte o paper original.
Resumo indisponível. Consulte o paper original.
Resumo indisponível. Consulte o paper original.
Resumo indisponível. Consulte o paper original.