AWQ: Up to 2.66x higher throughput (#2566)

This commit is contained in:
Casper
2024-01-26 23:53:17 -08:00
committed by GitHub
parent 390b495ff3
commit beb89f68b4
4 changed files with 127 additions and 1 deletions
+8
View File
@@ -70,6 +70,14 @@ torch::Tensor awq_gemm(
torch::Tensor _scaling_factors,
torch::Tensor _zeros,
int split_k_iters);
torch::Tensor awq_dequantize(
torch::Tensor _kernel,
torch::Tensor _scaling_factors,
torch::Tensor _zeros,
int split_k_iters,
int thx,
int thy);
#endif
void squeezellm_gemm(