Unverified50% confidenceBenchmarkExact time
After introducing the one-word summary prompt and lowering the head learning rate, the accuracies of the two small encoders rose to 41.7% and 42.5%, and Qwen3-8B reached 46.0%±0.2 (three random seeds).
1
Sources
50%
Confidence
Long-term
Relevance
11/19/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedAfter 50 practice runs, Qwen's capped answers dropped from 93% to 33%, the math score rose from 7.9 to 48.1, elementary word problems rose from 14 to 67, and the instruction-following test rose from 21 to 62; when given only 10 or 25 practice runs, the loop barely faded and the math score remained low.75% similarUnverifiedIn reinforcement learning post-training, Deletion Tuning's prior-task accuracy remains almost at the base model level, whereas full fine-tuning, LoRA, and random sparse fine-tuning lose up to 5.4 percentage points.74% similarUnverifiedOn the Qwen2.5 series models, from 0.5B to 14B parameters, model accuracy on indication questions improved by 18.3 percentage points, but accuracy on contraindication questions remained almost stagnant near the level of random guessing.73% similarUnverifiedRandom accuracy is about 0.33 for three-class classification and about 0.25 for four-class classification; only the magnitude of improvement relative to the random level can reflect whether the model has truly learned anything.72% similarUnverifiedIQ2_XXS以约2.06位的平均比特率对权重编码,相比IQ3_S(约3.44位)或Q4_K_M(约4.83位)压缩率更高但信息损失更大72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1186763API
curl https://kongchang.com/api/v1/knowledge/claims/1186763MCP
get_claim(id=1186763)