Unverified50% confidenceBenchmarkExact time
The full 340M model trains 4.9 times faster.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/30/2026
First Seen
Valid until: 1/28/2027
Sources
Related Entities
Related Claims
Unverified340M and 1.3B are parameter scales, and 4K context means a training/evaluation sequence length of 4096 tokens.69% similarUnverifiedA 4B model trained with RPL outperforms all 8B model baselines.68% similarUnverifiedThe CS-4's inference speed is an order of magnitude faster (about 30x) than GPU solutions.64% similarUnverifiedThe 3.11x is kernel-level forward acceleration, while 34% is end-to-end training step speed improvement; the two are measured on different bases.64% similarUnverifiedControlled measurements show that branch pruning brings a 1.426x speedup in step time.63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1114898API
curl https://kongchang.com/api/v1/knowledge/claims/1114898MCP
get_claim(id=1114898)