Unverified50% confidenceBenchmarkExact time
Using Qwen3-4B as the base model, applying supervised fine-tuning with only counterexample data causes the model's ability to recognize true theorems to collapse from 0.27 to 0.00.
1
Sources
50%
Confidence
Long-term
Relevance
10/5/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedA 70B parameter model after Q4 quantization often still has a lower perplexity than a 13B model under Q8 or even FP16.71% similarUnverifiedThe higher the quantization level (e.g., Q4), the less VRAM is consumed, but generation quality will decrease accordingly.69% similarUnverifiedOllama downloads Q4 quantized models by default, which have small VRAM usage but lower precision and more pronounced hallucination behavior.66% similarUnverifiedThe gap between this prompt-only method and the fine-tuned single-model state of the art has been narrowed to just 0.38 points.66% similarUnverifiedThe zero-shot span-based GLiNER model almost completely fails on entity types that depend on institutional function, achieving a micro F1 of only 0.243.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/975472API
curl https://kongchang.com/api/v1/knowledge/claims/975472MCP
get_claim(id=975472)