Spaces:
Paused
Paused
truongghieu
commited on
Commit
•
84de4b9
1
Parent(s):
96d175f
Update app.py
Browse files
app.py
CHANGED
@@ -3,7 +3,7 @@ from transformers import AutoTokenizer, AutoModelForCausalLM, GenerationConfig,B
|
|
3 |
|
4 |
import torch
|
5 |
|
6 |
-
model_id = "truongghieu/deci-
|
7 |
|
8 |
# Check if a GPU is available
|
9 |
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
@@ -16,14 +16,16 @@ bnb_config = BitsAndBytesConfig(
|
|
16 |
|
17 |
tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
|
18 |
# Load model in this way if use GPU
|
19 |
-
|
20 |
-
model = AutoModelForCausalLM.from_pretrained(
|
|
|
|
|
21 |
# Move the model to the GPU if available
|
22 |
|
23 |
generation_config = GenerationConfig(
|
24 |
penalty_alpha=0.6,
|
25 |
do_sample=True,
|
26 |
-
top_k=
|
27 |
temperature=0.5,
|
28 |
repetition_penalty=1.2,
|
29 |
max_new_tokens=50,
|
|
|
3 |
|
4 |
import torch
|
5 |
|
6 |
+
model_id = "truongghieu/deci-finetuned_Prj2"
|
7 |
|
8 |
# Check if a GPU is available
|
9 |
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
|
|
16 |
|
17 |
tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
|
18 |
# Load model in this way if use GPU
|
19 |
+
if device == "cuda":
|
20 |
+
model = AutoModelForCausalLM.from_pretrained(model_id, trust_remote_code=True, quantization_config=bnb_config)
|
21 |
+
else:
|
22 |
+
model = AutoModelForCausalLM.from_pretrained("truongghieu/deci-finetuned", trust_remote_code=True)
|
23 |
# Move the model to the GPU if available
|
24 |
|
25 |
generation_config = GenerationConfig(
|
26 |
penalty_alpha=0.6,
|
27 |
do_sample=True,
|
28 |
+
top_k=3,
|
29 |
temperature=0.5,
|
30 |
repetition_penalty=1.2,
|
31 |
max_new_tokens=50,
|