Unverified50% confidenceBenchmarkExact time
In EvoMap's benchmark test, sub-agents actually found 373 correct answers, but after the main agent merged the results, only 217 were retained in the final output.
1
Sources
50%
Confidence
Long-term
Relevance
10/6/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe model gave the correct answer of 34 in a multimodal image recognition test counting pencils, at a speed of 55 tokens per second at that time.69% similarUnverified早期Codex在HumanEval基准测试中以37%的准确率独立解决编程题，达到当时最高水平66% similarUnverifiedThe fine-tuned 406M model achieved 36.33 ROUGE-1 on the test set, while the authors' own 1.2B system achieved 35.41 ROUGE-1, and the two cannot be statistically distinguished.64% similarUnverifiedMistral Small 4 scored only 0.699 on unit conversion inputs and produced 265 errors that differed from the correct label by exactly a power of ten.63% similarUnverifiedRegardless of the encoding scheme used, 4-bit precision can only distinguish 16 different numerical levels.63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/981941API
curl https://kongchang.com/api/v1/knowledge/claims/981941MCP
get_claim(id=981941)