Spaces:
Running
on
Zero
Running
on
Zero
remove flash
Browse files- app.py +0 -1
- requirements.txt +0 -1
app.py
CHANGED
|
@@ -14,7 +14,6 @@ model = AutoModelForCausalLM.from_pretrained(
|
|
| 14 |
device_map="cuda",
|
| 15 |
torch_dtype="auto",
|
| 16 |
trust_remote_code=True,
|
| 17 |
-
attn_implementation='flash_attention_2',
|
| 18 |
).cuda()
|
| 19 |
|
| 20 |
generation_config = GenerationConfig.from_pretrained(model_path)
|
|
|
|
| 14 |
device_map="cuda",
|
| 15 |
torch_dtype="auto",
|
| 16 |
trust_remote_code=True,
|
|
|
|
| 17 |
).cuda()
|
| 18 |
|
| 19 |
generation_config = GenerationConfig.from_pretrained(model_path)
|
requirements.txt
CHANGED
|
@@ -7,6 +7,5 @@ scipy
|
|
| 7 |
soundfile
|
| 8 |
pillow
|
| 9 |
accelerate
|
| 10 |
-
flash-attn
|
| 11 |
transformers
|
| 12 |
backoff
|
|
|
|
| 7 |
soundfile
|
| 8 |
pillow
|
| 9 |
accelerate
|
|
|
|
| 10 |
transformers
|
| 11 |
backoff
|