Unverified50% confidenceBenchmarkExact time
Evaluation on BLAST and multiple zero-shot benchmarks shows that FlexCast's accuracy is comparable to or even better than strong TSFM baselines, while inference latency is significantly reduced.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
11/16/2026
First Seen
Valid until: 2/14/2027
Sources
Related Entities
Related Claims
UnverifiedThere is a trade-off between speed and accuracy: the TE recipe offers higher accuracy (loss only 0.87% higher) but lower throughput, while the mxfp scheme offers higher throughput but greater accuracy loss.74% similarUnverifiedModels deployed at lower numerical precision are cheaper to run; precision is simultaneously a cost lever and a quality lever.66% similarUnverifiedIn actual deployment, lower precision is usually used for weights and slightly higher precision is retained for activations, to strike a balance between image quality and speed.66% similarUnverifiedASR识别准确率仍强烈依赖信噪比,当SNR低于0dB时模型性能呈现悬崖式跌落而非优雅降级65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1174097API
curl https://kongchang.com/api/v1/knowledge/claims/1174097MCP
get_claim(id=1174097)