feat: add initial AutoGPTQ backend implementation

This commit is contained in:
Ettore Di Giacinto 2023-08-07 22:39:10 +02:00
parent 91d49cfe9f
commit a843e64fc2
37 changed files with 660 additions and 148 deletions

View file

@ -54,6 +54,11 @@ type Config struct {
RMSNormEps float32 `yaml:"rms_norm_eps"`
NGQA int32 `yaml:"ngqa"`
// AutoGPTQ
ModelBaseName string `yaml:"model_base_name"`
Device string `yaml:"device"`
Triton bool `yaml:"triton"`
}
type Functions struct {