Update config.json
Browse files- config.json +1 -11
config.json
CHANGED
|
@@ -21,23 +21,14 @@
|
|
| 21 |
"backend": "auto",
|
| 22 |
"batch_size": 1,
|
| 23 |
"bits": 4,
|
| 24 |
-
"block_name_to_quantize":
|
| 25 |
"cache_block_outputs": true,
|
| 26 |
"checkpoint_format": "gptq",
|
| 27 |
"damp_percent": 0.1,
|
| 28 |
"dataset": "c4",
|
| 29 |
"desc_act": false,
|
| 30 |
-
"exllama_config": {
|
| 31 |
-
"version": 1
|
| 32 |
-
},
|
| 33 |
"group_size": 128,
|
| 34 |
"max_input_length": null,
|
| 35 |
-
"meta": {
|
| 36 |
-
"quantizer": [
|
| 37 |
-
"optimum:2.1.0",
|
| 38 |
-
"gptqmodel:5.6.12"
|
| 39 |
-
]
|
| 40 |
-
},
|
| 41 |
"model_seqlen": null,
|
| 42 |
"module_name_preceding_first_block": null,
|
| 43 |
"modules_in_block_to_quantize": null,
|
|
@@ -47,7 +38,6 @@
|
|
| 47 |
"tokenizer": null,
|
| 48 |
"true_sequential": true,
|
| 49 |
"use_cuda_fp16": false,
|
| 50 |
-
"use_exllama": true
|
| 51 |
},
|
| 52 |
"rms_norm_eps": 1e-06,
|
| 53 |
"rope_scaling": {
|
|
|
|
| 21 |
"backend": "auto",
|
| 22 |
"batch_size": 1,
|
| 23 |
"bits": 4,
|
| 24 |
+
"block_name_to_quantize": "model.language_model.layers",
|
| 25 |
"cache_block_outputs": true,
|
| 26 |
"checkpoint_format": "gptq",
|
| 27 |
"damp_percent": 0.1,
|
| 28 |
"dataset": "c4",
|
| 29 |
"desc_act": false,
|
|
|
|
|
|
|
|
|
|
| 30 |
"group_size": 128,
|
| 31 |
"max_input_length": null,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 32 |
"model_seqlen": null,
|
| 33 |
"module_name_preceding_first_block": null,
|
| 34 |
"modules_in_block_to_quantize": null,
|
|
|
|
| 38 |
"tokenizer": null,
|
| 39 |
"true_sequential": true,
|
| 40 |
"use_cuda_fp16": false,
|
|
|
|
| 41 |
},
|
| 42 |
"rms_norm_eps": 1e-06,
|
| 43 |
"rope_scaling": {
|