Unverified50% confidenceBenchmarkExact time
When using SGUID, Qwen3-4B improved by 0.5 percentage points on HMMT25 after the first round, and by a cumulative 1.1 percentage points after the second round.
1
Sources
50%
Confidence
Long-term
Relevance
10/13/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedIn the Qwen3-4B comparative experiment, updating the model directly with unfiltered skills without using SGUID filtering decreased performance by 0.3 percentage points on HMMT25.76% similarUnverifiedAt a throughput of 259k, Qwen3.5 achieves approximately a 19× speed improvement, with hybrid attention contributing approximately 5×, MoE sparsification approximately 2×, and Multi-Token Prediction approximately 1.5×.67% similarUnverifiedGPT-4o and Qwen 2.5 Coder 32B tied for second place in DataBugBench, both scoring 4/5 with an accuracy of 80%.67% similarUnverifiedGlimmer's intelligence index score of 35 surpasses Gemma 2 27B (30) and nearly matches Qwen 2.5 (36), with Qwen 2.5 72B topping the leaderboard at 38.67% similarUnverifiedLlama 3.1 405B's score on MMLU is already close to the level of GPT-4o, and Qwen 2.5 72B has surpassed GPT-4 Turbo on certain metrics in math and code generation benchmarks.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1013309API
curl https://kongchang.com/api/v1/knowledge/claims/1013309MCP
get_claim(id=1013309)