Unverified50% confidenceBenchmarkExact time
On the CLEVR dataset of the VQA task, Latent-Lens achieves an accuracy of 81.0±7.9%, versus 58.3±5.7% for the text communication baseline.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
11/2/2026
First Seen
Valid until: 1/31/2027
Sources
Related Entities
Related Claims
UnverifiedDirect use of Kev achieves 64.00% accuracy, and Qwen3.5-4B outputting only the final answer achieves 68.80% accuracy.71% similarUnverifiedIn agentgateway's testing, qwen3.8 (27.3B, Q4_K_M quantization, 262K context) had a valid answer rate of 90.0% (54/60), an accuracy rate among answered questions of 98.1% (53/54), and an overall accuracy rate of 88.3% (53/60).69% similarUnverifiedThe V1 approach achieved a self-tested top-1 accuracy of 93%, but scored 0/137 top-5 accuracy on a benchmark set of 137 real photographs.68% similarUnverifiedThe label-image accuracy gaps of GPT-4o, Qwen-VL Max, and LLaMA reach 9.5, 13.8, and 12.6 percentage points respectively.67% similarUnverifiedOn the Qwen3-VL-4B model, fine-tuning with procedurally generated data raises its accuracy on SpaceCast-Bench from 34.0% to 65.7%, and achieves a macro-average gain across six out-of-domain benchmarks.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1125565API
curl https://kongchang.com/api/v1/knowledge/claims/1125565MCP
get_claim(id=1125565)