Unverified50% confidenceBenchmarkExact time
After training on the Yamaha-CMU Off-Road (YCOR) dataset, SegFormer-B2 achieves a validation set mIoU of approximately 57.4%.
1
Sources
50%
Confidence
Long-term
Relevance
10/8/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedGradient checkpointing and mixed-precision training (FP16/BF16) are common memory-reduction measures.59% similarUnverifiedbuffalo_l在MS1MV3数据集（约300万张图像）上训练，LFW准确率接近99.8%58% similarUnverifiedThe combination of EfficientNet-B3 and SVM achieved the strongest baseline performance, with an accuracy of 96.31% and a macro F1 score of 96.36%.58% similarUnverifiedPathway reported that a BDH-based system achieved a 97.4% Top-1 accuracy rate on approximately 250,000 Sudoku Extreme puzzles without using CoT or solution backtracking.57% similarUnverifiedThe CAPI-DINO ViT-B/14 model achieved a linear probing accuracy of 68.7% on the iNat21 dataset.56% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/987189API
curl https://kongchang.com/api/v1/knowledge/claims/987189MCP
get_claim(id=987189)