Muennighoff
commited on
Commit
•
a2975b0
1
Parent(s):
900f290
Update modeling_gritlm7b.py
Browse files- modeling_gritlm7b.py +6 -4
modeling_gritlm7b.py
CHANGED
@@ -46,11 +46,13 @@ from transformers.utils import (
|
|
46 |
from transformers import MistralConfig
|
47 |
|
48 |
|
|
|
49 |
try:
|
50 |
-
|
51 |
-
|
52 |
-
|
53 |
-
|
|
|
54 |
except:
|
55 |
pass
|
56 |
|
|
|
46 |
from transformers import MistralConfig
|
47 |
|
48 |
|
49 |
+
# transformers has a bug where it will try to import everything from a custom model file unless there's try/except
|
50 |
try:
|
51 |
+
if is_flash_attn_2_available():
|
52 |
+
from flash_attn import flash_attn_func, flash_attn_varlen_func
|
53 |
+
from flash_attn.bert_padding import index_first_axis, pad_input, unpad_input # noqa
|
54 |
+
|
55 |
+
_flash_supports_window_size = "window_size" in list(inspect.signature(flash_attn_func).parameters)
|
56 |
except:
|
57 |
pass
|
58 |
|