update to gptq format
Browse filesSigned-off-by: wenhuach <wenhuach87@gmail.com>
- config.json +1 -2
config.json
CHANGED
|
@@ -38,7 +38,6 @@
|
|
| 38 |
"qk_nope_head_dim": 128,
|
| 39 |
"qk_rope_head_dim": 64,
|
| 40 |
"quantization_config": {
|
| 41 |
-
"backend": "auto_round:gptq:exllamav2",
|
| 42 |
"amp": true,
|
| 43 |
"autoround_version": "0.4.3",
|
| 44 |
"batch_size": 8,
|
|
@@ -56,7 +55,7 @@
|
|
| 56 |
"lr": 0.005,
|
| 57 |
"minmax_lr": 0.005,
|
| 58 |
"nsamples": 512,
|
| 59 |
-
"quant_method": "
|
| 60 |
"scale_dtype": "torch.float16",
|
| 61 |
"seqlen": 512,
|
| 62 |
"sym": true,
|
|
|
|
| 38 |
"qk_nope_head_dim": 128,
|
| 39 |
"qk_rope_head_dim": 64,
|
| 40 |
"quantization_config": {
|
|
|
|
| 41 |
"amp": true,
|
| 42 |
"autoround_version": "0.4.3",
|
| 43 |
"batch_size": 8,
|
|
|
|
| 55 |
"lr": 0.005,
|
| 56 |
"minmax_lr": 0.005,
|
| 57 |
"nsamples": 512,
|
| 58 |
+
"quant_method": "gptq",
|
| 59 |
"scale_dtype": "torch.float16",
|
| 60 |
"seqlen": 512,
|
| 61 |
"sym": true,
|