Add GPTQ support (#916)
This commit is contained in:
+12
@@ -77,3 +77,15 @@ void squeezellm_gemm(
|
||||
torch::Tensor mat,
|
||||
torch::Tensor mul,
|
||||
torch::Tensor lookup_table);
|
||||
|
||||
torch::Tensor gptq_gemm(
|
||||
torch::Tensor a,
|
||||
torch::Tensor b_q_weight,
|
||||
torch::Tensor b_gptq_qzeros,
|
||||
torch::Tensor b_gptq_scales,
|
||||
torch::Tensor b_g_idx,
|
||||
bool use_exllama);
|
||||
|
||||
void gptq_shuffle(
|
||||
torch::Tensor q_weight,
|
||||
torch::Tensor q_perm);
|
||||
|
||||
Reference in New Issue
Block a user