Unverified50% confidenceFactExact time

Porting LLaVA-1.5's token IDs as-is to two models with disjoint vocabularies drops F1 straight to 0.00, whereas using tokenizer-derived readings keeps F1 stable between 0.80 and 0.92.

1

Sources

50%

Confidence

Long-term

Relevance

10/20/2026

First Seen

Sources

The POPE Benchmark's Yes/No Extraction Pitfall: VLM Hallucination Evaluation Confounded by Token Choice

Related Entities

LLaVA-1.5model SentencePiececoncept BPEconcept WordPiececoncept

Related Claims

UnverifiedUpgrading from v0.27.0 to v0.27.1 is generally low-risk, allowing users to gain new features while maintaining system stability.51% similar Unverifiedv0.40.2 increments from v0.40.1 and is a patch-level update.51% similar UnverifiedMeta Llama系列开源版本曾因量化配置错误导致性能下降50% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/1074068

API

curl https://kongchang.com/api/v1/knowledge/claims/1074068

MCP

get_claim(id=1074068)