Spaces:
Runtime error
Runtime error
Commit
•
2c99f68
1
Parent(s):
a6b011a
Update run.py
Browse files
run.py
CHANGED
|
@@ -15,8 +15,7 @@ from unsloth import FastLanguageModel
|
|
| 15 |
import torch
|
| 16 |
|
| 17 |
|
| 18 |
-
max_seq_length = 2048 # Choose any! We auto support RoPE Scaling internally!
|
| 19 |
-
dtype = None # None for auto detection. Float16 for Tesla T4, V100, Bfloat16 for Ampere+
|
| 20 |
load_in_4bit = True # Use 4bit quantization to reduce memory usage. Can be False.
|
| 21 |
|
| 22 |
|
|
@@ -32,7 +31,7 @@ from unsloth import FastLanguageModel
|
|
| 32 |
model, tokenizer = FastLanguageModel.from_pretrained(
|
| 33 |
model_name = "traversaal-llm-regional-languages/Unsloth_Urdu_Llama3_1_4bit_PF100",
|
| 34 |
max_seq_length = max_seq_length,
|
| 35 |
-
dtype =
|
| 36 |
load_in_4bit = load_in_4bit
|
| 37 |
)
|
| 38 |
FastLanguageModel.for_inference(model)
|
|
|
|
| 15 |
import torch
|
| 16 |
|
| 17 |
|
| 18 |
+
max_seq_length = 2048 # Choose any! We auto support RoPE Scaling internally! # None for auto detection. Float16 for Tesla T4, V100, Bfloat16 for Ampere+
|
|
|
|
| 19 |
load_in_4bit = True # Use 4bit quantization to reduce memory usage. Can be False.
|
| 20 |
|
| 21 |
|
|
|
|
| 31 |
model, tokenizer = FastLanguageModel.from_pretrained(
|
| 32 |
model_name = "traversaal-llm-regional-languages/Unsloth_Urdu_Llama3_1_4bit_PF100",
|
| 33 |
max_seq_length = max_seq_length,
|
| 34 |
+
dtype = 'Auto',
|
| 35 |
load_in_4bit = load_in_4bit
|
| 36 |
)
|
| 37 |
FastLanguageModel.for_inference(model)
|