Unverified50% confidenceBenchmarkExact time
In the benchmark, the Dyna-Q algorithm obtained an average of approximately 0.62 badges.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
8/29/2026
First Seen
Valid until: 11/27/2026
Sources
Related Entities
Related Claims
UnverifiedQwen3.8 27B achieved an overall score of 52 on the Artificial Analysis evaluation platform.63% similarUnverifiedThe UP host fine-tuned Qwen3-4B with approximately 100 to 200 additional identity-stability-related data samples specifically to resolve the character confusion issue.61% similarUnverifiedIQ2_XXS以约2.06位的平均比特率编码权重，IQ3_S约3.44位，Q4_K_M约4.83位61% similarUnverifiedQwen 3.6 Max in neutral mode has an ambiguity detection F1 of only 16% and asks an average of only 0.07 questions per task, but once it does ask for clarification, the accuracy reaches 94.7%61% similarUnverifiedThe Q4_K_M quantization scheme averages approximately 4.8 bits per weight, while the Q8_0 quantization scheme uses 8 bits per weight.59% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/822091API
curl https://kongchang.com/api/v1/knowledge/claims/822091MCP
get_claim(id=822091)