Upload folder using huggingface_hub
Browse files- config.json +35 -31
config.json
CHANGED
|
@@ -21,38 +21,42 @@
|
|
| 21 |
"pad_token_id": -1,
|
| 22 |
"pretraining_tp": 1,
|
| 23 |
"quantization_config": {
|
| 24 |
-
"
|
| 25 |
-
"exclude": [
|
| 26 |
-
"lm_head"
|
| 27 |
-
],
|
| 28 |
-
"export": {
|
| 29 |
-
"kv_cache_group": [],
|
| 30 |
-
"pack_method": "reorder",
|
| 31 |
-
"weight_format": "real_quantized",
|
| 32 |
-
"weight_merge_groups": null
|
| 33 |
-
},
|
| 34 |
-
"global_quant_config": {
|
| 35 |
-
"bias": null,
|
| 36 |
-
"input_tensors": null,
|
| 37 |
-
"output_tensors": null,
|
| 38 |
-
"target_device": null,
|
| 39 |
-
"weight": {
|
| 40 |
-
"ch_axis": null,
|
| 41 |
-
"dtype": "int8",
|
| 42 |
-
"group_size": null,
|
| 43 |
-
"is_dynamic": false,
|
| 44 |
-
"observer_cls": "PerTensorMinMaxObserver",
|
| 45 |
-
"qscheme": "per_tensor",
|
| 46 |
-
"round_method": "half_even",
|
| 47 |
-
"scale_type": "float",
|
| 48 |
-
"symmetric": true
|
| 49 |
-
}
|
| 50 |
-
},
|
| 51 |
-
"layer_quant_config": {},
|
| 52 |
-
"layer_type_quant_config": {},
|
| 53 |
-
"pack_method": "reorder",
|
| 54 |
"quant_method": "quark",
|
| 55 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 56 |
},
|
| 57 |
"rms_norm_eps": 1e-06,
|
| 58 |
"rope_scaling": null,
|
|
|
|
| 21 |
"pad_token_id": -1,
|
| 22 |
"pretraining_tp": 1,
|
| 23 |
"quantization_config": {
|
| 24 |
+
"library": "quark",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 25 |
"quant_method": "quark",
|
| 26 |
+
"quark_config": {
|
| 27 |
+
"algo_config": null,
|
| 28 |
+
"exclude": [
|
| 29 |
+
"lm_head"
|
| 30 |
+
],
|
| 31 |
+
"export": {
|
| 32 |
+
"kv_cache_group": [],
|
| 33 |
+
"pack_method": "reorder",
|
| 34 |
+
"weight_format": "real_quantized",
|
| 35 |
+
"weight_merge_groups": null
|
| 36 |
+
},
|
| 37 |
+
"global_quant_config": {
|
| 38 |
+
"bias": null,
|
| 39 |
+
"input_tensors": null,
|
| 40 |
+
"output_tensors": null,
|
| 41 |
+
"target_device": null,
|
| 42 |
+
"weight": {
|
| 43 |
+
"ch_axis": null,
|
| 44 |
+
"dtype": "int8",
|
| 45 |
+
"group_size": null,
|
| 46 |
+
"is_dynamic": false,
|
| 47 |
+
"observer_cls": "PerTensorMinMaxObserver",
|
| 48 |
+
"qscheme": "per_tensor",
|
| 49 |
+
"round_method": "half_even",
|
| 50 |
+
"scale_type": "float",
|
| 51 |
+
"symmetric": true
|
| 52 |
+
}
|
| 53 |
+
},
|
| 54 |
+
"layer_quant_config": {},
|
| 55 |
+
"layer_type_quant_config": {},
|
| 56 |
+
"pack_method": "reorder",
|
| 57 |
+
"quant_method": "quark",
|
| 58 |
+
"quant_mode": 1
|
| 59 |
+
}
|
| 60 |
},
|
| 61 |
"rms_norm_eps": 1e-06,
|
| 62 |
"rope_scaling": null,
|