Unverified50% confidenceBenchmarkExact time
The model with the extended vocabulary improved its pass rate on HumanEval from 45.7% to 52.4%.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/23/2026
First Seen
Valid until: 1/21/2027
Sources
Related Entities
Related Claims
UnverifiedProgram-Agent pushed the base model and the supervised fine-tuned model to 55.71% and 56.25%, respectively.68% similarUnverifiedIn the same test, proposal quality rose from 44.7% to 53%, an increase of 8.3 percentage points.66% similarUnverifiedAfter RL post-training, the small model's accuracy improves from 4% to 45% (pass@1).65% similarUnverifiedInjecting concept representation into the draft model of DFlash2 can improve the mean acceptance length by 4.17%.65% similarUnverifiedIn providing correct, domain-specific justifications, performance across different models ranged from 25.6% all the way to 76.9%.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1091083API
curl https://kongchang.com/api/v1/knowledge/claims/1091083MCP
get_claim(id=1091083)