marlin
PublicFP16xINT4 LLM inference kernel that can achieve near-ideal ~4x speedups up to medium batchsizes of 16-32 tokens.
Creat:2024-01-17T21:07:53
Update:2025-03-26T21:37:20
870
Stars
0
Stars Increase
FP16xINT4 LLM inference kernel that can achieve near-ideal ~4x speedups up to medium batchsizes of 16-32 tokens.