Unverified50% confidenceBenchmarkExact time
In the experiments, the 4B model's accuracy before training was about fifty percent, while the 27B model reached over seventy-five percent.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
10/17/2026
First Seen
Valid until: 10/31/2026
Sources
Related Entities
Related Claims
UnverifiedAfter training for most of a day using only about 1,000 edge tasks that are 'sometimes done right and sometimes done wrong,' the 4B model recovered more than half the gap to the 27B model.76% similarUnverifiedAfter RL post-training, the small model's accuracy improves from 4% to 45% (pass@1).70% similarUnverifiedIn a sound classification experiment on the ESC-50 dataset, the top-1 hit rate was approximately 25% before training and rose to approximately 66% after about seven and a half minutes of training on an A100.70% similarUnverifiedEarly in training, the model's output is close to a uniform distribution and α is low; as training progresses, the model learns to concentrate probability mass on the correct token, and α rises.69% similarUnverified4-bit量化模型在大多数任务上的表现已能达到原始精度的95%以上68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1056464API
curl https://kongchang.com/api/v1/knowledge/claims/1056464MCP
get_claim(id=1056464)