Unverified50% confidenceBenchmarkExact time
FSRS-6 is the sixth-generation iteration of the FSRS series, and its accuracy in predicting forgetting curves in benchmark tests is significantly better than previous generations.
1
Sources
50%
Confidence
Long-term
Relevance
10/10/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedGradient checkpointing and mixed-precision training (FP16/BF16) are common memory-reduction measures.66% similarVerifiedFP8 has two variants: E4M3 (4-bit exponent + 3-bit mantissa), which has higher precision and is suited for forward inference, and E5M2 (5-bit exponent + 2-bit mantissa), which has a larger dynamic range and is suited for gradient computation.64% similarUnverifiedCompared to FP16 original precision, FP8 quantization typically causes no more than 1–2% performance degradation on mainstream benchmarks.61% similarUnverifiedGrok 4.6 在 WANDR 上能够匹配 Fable 5 的测试结果，而成本降低了 60% 以上60% similarUnverifiedState-of-the-art supervised models on the LIAR benchmark still hover around 28%–32% six-class accuracy.60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/994015API
curl https://kongchang.com/api/v1/knowledge/claims/994015MCP
get_claim(id=994015)