Spaces:
Paused
Paused
Commit
·
cb0f371
1
Parent(s):
85bb959
Update app_v3.py
Browse files
app_v3.py
CHANGED
|
@@ -19,7 +19,7 @@ pretrained_model_dir = "FPHam/Jackson_The_Formalizer_V2_13b_GPTQ"
|
|
| 19 |
|
| 20 |
#st.write(f'{os.listdir(quantized_model_dir)}')
|
| 21 |
#model_name_or_path = quantized_model_dir
|
| 22 |
-
|
| 23 |
|
| 24 |
#os.environ['CUDA_VISIBLE_DEVICES'] = '0'
|
| 25 |
|
|
@@ -33,7 +33,7 @@ use_triton = False
|
|
| 33 |
tokenizer = AutoTokenizer.from_pretrained(pretrained_model_dir, use_fast=True)
|
| 34 |
model = AutoGPTQForCausalLM.from_quantized(
|
| 35 |
pretrained_model_dir,
|
| 36 |
-
|
| 37 |
use_safetensors=True,
|
| 38 |
device="cuda:0",
|
| 39 |
#use_triton=use_triton,
|
|
|
|
| 19 |
|
| 20 |
#st.write(f'{os.listdir(quantized_model_dir)}')
|
| 21 |
#model_name_or_path = quantized_model_dir
|
| 22 |
+
model_basename = "Jackson2-4bit-128g-GPTQ"
|
| 23 |
|
| 24 |
#os.environ['CUDA_VISIBLE_DEVICES'] = '0'
|
| 25 |
|
|
|
|
| 33 |
tokenizer = AutoTokenizer.from_pretrained(pretrained_model_dir, use_fast=True)
|
| 34 |
model = AutoGPTQForCausalLM.from_quantized(
|
| 35 |
pretrained_model_dir,
|
| 36 |
+
model_basename=model_basename,
|
| 37 |
use_safetensors=True,
|
| 38 |
device="cuda:0",
|
| 39 |
#use_triton=use_triton,
|