Unverified50% confidenceBenchmarkExact time
Numerical correctness validation of the engine shows that, compared with the CPU fp32 transformers reference implementation, the English paragraph PPL is 5.233 vs 5.280, average KL ≤ 0.003, and greedy answer matches are 63/64 and 64/64.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/17/2026
First Seen
Valid until: 1/15/2027
Sources
Related Entities
Related Claims
UnverifiedThe C runtime's perplexity is 26.8407, and the PyTorch reference implementation's is 26.8188, a difference of 0.082%, meeting the threshold of <=1%.68% similarUnverifiedLPIPS scores published by Unsloth relative to BF16 Turbo transformer are: Q8_0=0.036, Q4_K_M=0.153, Q2_K=0.374.66% similarUnverifiedIn the task of completing the last Token of an FP8 quantization function, after DeepSeek-V4-Flash-Base adds equal-sign decorations before the code, when the number of equal signs modulo 4 is 0 or 1 the error rate is as high as 71.3%, and when the remainder is 2 or 3 the accuracy returns to 91.5%.66% similarUnverified在细粒度汽车分类的k-NN实验中，SigLIP2 SO400M达到约92%准确率，CLIP ViT-L约59%，DINOv2 Giant约41%64% similarUnverifiedAt fp16 precision, each value occupies 2 bytes.63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1053043API
curl https://kongchang.com/api/v1/knowledge/claims/1053043MCP
get_claim(id=1053043)