Fix q_group_size
#1
by
mike-ravkine
- opened
README.md
CHANGED
|
@@ -54,7 +54,7 @@ tokenizer = AutoTokenizer.from_pretrained(config.tokenizer_name)
|
|
| 54 |
w_bit = 4
|
| 55 |
q_config = {
|
| 56 |
"zero_point": True,
|
| 57 |
-
"q_group_size":
|
| 58 |
}
|
| 59 |
|
| 60 |
load_quant = snapshot_download('abhinavkulkarni/tiiuae-falcon-40b-instruct-w4-g128-awq')
|
|
|
|
| 54 |
w_bit = 4
|
| 55 |
q_config = {
|
| 56 |
"zero_point": True,
|
| 57 |
+
"q_group_size": 128,
|
| 58 |
}
|
| 59 |
|
| 60 |
load_quant = snapshot_download('abhinavkulkarni/tiiuae-falcon-40b-instruct-w4-g128-awq')
|