Upload config.json with huggingface_hub
Browse files- config.json +191 -0
config.json
ADDED
|
@@ -0,0 +1,191 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"MiniCPMForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"attention_bias": false,
|
| 6 |
+
"attention_dropout": 0.0,
|
| 7 |
+
"auto_map": {
|
| 8 |
+
"AutoConfig": "configuration_minicpm.MiniCPMConfig",
|
| 9 |
+
"AutoModel": "modeling_minicpm.MiniCPMModel",
|
| 10 |
+
"AutoModelForCausalLM": "modeling_minicpm.MiniCPMForCausalLM",
|
| 11 |
+
"AutoModelForSeq2SeqLM": "modeling_minicpm.MiniCPMForCausalLM",
|
| 12 |
+
"AutoModelForSequenceClassification": "modeling_minicpm.MiniCPMForSequenceClassification"
|
| 13 |
+
},
|
| 14 |
+
"bos_token_id": 1,
|
| 15 |
+
"dim_model_base": 256,
|
| 16 |
+
"eos_token_id": 73440,
|
| 17 |
+
"head_dim": 128,
|
| 18 |
+
"hidden_act": "silu",
|
| 19 |
+
"hidden_size": 4096,
|
| 20 |
+
"initializer_range": 0.1,
|
| 21 |
+
"intermediate_size": 16384,
|
| 22 |
+
"max_position_embeddings": 65536,
|
| 23 |
+
"mlp_bias": false,
|
| 24 |
+
"model_type": "minicpm",
|
| 25 |
+
"num_attention_heads": 32,
|
| 26 |
+
"num_hidden_layers": 32,
|
| 27 |
+
"num_key_value_heads": 2,
|
| 28 |
+
"pad_token_id": 73440,
|
| 29 |
+
"pretraining_tp": 1,
|
| 30 |
+
"quantization_config": {
|
| 31 |
+
"bits": 4,
|
| 32 |
+
"checkpoint_format": "gptq",
|
| 33 |
+
"damp_percent": 0.01,
|
| 34 |
+
"desc_act": false,
|
| 35 |
+
"group_size": 128,
|
| 36 |
+
"lm_head": false,
|
| 37 |
+
"model_file_base_name": null,
|
| 38 |
+
"model_name_or_path": null,
|
| 39 |
+
"quant_method": "gptq",
|
| 40 |
+
"static_groups": true,
|
| 41 |
+
"sym": true,
|
| 42 |
+
"true_sequential": true
|
| 43 |
+
},
|
| 44 |
+
"rms_norm_eps": 1e-06,
|
| 45 |
+
"rope_scaling": {
|
| 46 |
+
"factor": 1.0,
|
| 47 |
+
"long_factor": [
|
| 48 |
+
0.9982316082870437,
|
| 49 |
+
1.033048153422584,
|
| 50 |
+
1.0749920956484724,
|
| 51 |
+
1.1255096879436193,
|
| 52 |
+
1.1863348602111476,
|
| 53 |
+
1.259543828902579,
|
| 54 |
+
1.3476188888731149,
|
| 55 |
+
1.4535223827776373,
|
| 56 |
+
1.5807816745852985,
|
| 57 |
+
1.7335856049489526,
|
| 58 |
+
1.9168922912975785,
|
| 59 |
+
2.1365471404135326,
|
| 60 |
+
2.3994084200118646,
|
| 61 |
+
2.713475511863602,
|
| 62 |
+
3.0880118452194134,
|
| 63 |
+
3.533650295140154,
|
| 64 |
+
4.062463396503134,
|
| 65 |
+
4.687974098908333,
|
| 66 |
+
5.425075306704039,
|
| 67 |
+
6.289818967956352,
|
| 68 |
+
7.29902962722721,
|
| 69 |
+
8.469695779093664,
|
| 70 |
+
9.81809877306655,
|
| 71 |
+
11.358657902065282,
|
| 72 |
+
13.102505860712087,
|
| 73 |
+
15.055862949967128,
|
| 74 |
+
17.218348131364184,
|
| 75 |
+
19.581439255386453,
|
| 76 |
+
22.127353314656723,
|
| 77 |
+
24.828633849376587,
|
| 78 |
+
27.6486820771775,
|
| 79 |
+
30.54334096108829,
|
| 80 |
+
33.46345345363812,
|
| 81 |
+
36.358112337548896,
|
| 82 |
+
39.17816056534983,
|
| 83 |
+
41.879441100069684,
|
| 84 |
+
44.425355159339965,
|
| 85 |
+
46.78844628336223,
|
| 86 |
+
48.95093146475928,
|
| 87 |
+
50.90428855401433,
|
| 88 |
+
52.648136512661125,
|
| 89 |
+
54.18869564165987,
|
| 90 |
+
55.537098635632745,
|
| 91 |
+
56.7077647874992,
|
| 92 |
+
57.71697544677006,
|
| 93 |
+
58.58171910802236,
|
| 94 |
+
59.31882031581807,
|
| 95 |
+
59.94433101822328,
|
| 96 |
+
60.47314411958625,
|
| 97 |
+
60.918782569507,
|
| 98 |
+
61.29331890286281,
|
| 99 |
+
61.60738599471455,
|
| 100 |
+
61.87024727431288,
|
| 101 |
+
62.089902123428836,
|
| 102 |
+
62.27320880977746,
|
| 103 |
+
62.42601274014111,
|
| 104 |
+
62.55327203194878,
|
| 105 |
+
62.65917552585329,
|
| 106 |
+
62.74725058582382,
|
| 107 |
+
62.82045955451526,
|
| 108 |
+
62.88128472678279,
|
| 109 |
+
62.931802319077946,
|
| 110 |
+
62.97374626130382,
|
| 111 |
+
63.008562806439365
|
| 112 |
+
],
|
| 113 |
+
"original_max_position_embeddings": 65536,
|
| 114 |
+
"rope_type": "longrope",
|
| 115 |
+
"short_factor": [
|
| 116 |
+
0.9982316082870437,
|
| 117 |
+
1.033048153422584,
|
| 118 |
+
1.0749920956484724,
|
| 119 |
+
1.1255096879436193,
|
| 120 |
+
1.1863348602111476,
|
| 121 |
+
1.259543828902579,
|
| 122 |
+
1.3476188888731149,
|
| 123 |
+
1.4535223827776373,
|
| 124 |
+
1.5807816745852985,
|
| 125 |
+
1.7335856049489526,
|
| 126 |
+
1.9168922912975785,
|
| 127 |
+
2.1365471404135326,
|
| 128 |
+
2.3994084200118646,
|
| 129 |
+
2.713475511863602,
|
| 130 |
+
3.0880118452194134,
|
| 131 |
+
3.533650295140154,
|
| 132 |
+
4.062463396503134,
|
| 133 |
+
4.687974098908333,
|
| 134 |
+
5.425075306704039,
|
| 135 |
+
6.289818967956352,
|
| 136 |
+
7.29902962722721,
|
| 137 |
+
8.469695779093664,
|
| 138 |
+
9.81809877306655,
|
| 139 |
+
11.358657902065282,
|
| 140 |
+
13.102505860712087,
|
| 141 |
+
15.055862949967128,
|
| 142 |
+
17.218348131364184,
|
| 143 |
+
19.581439255386453,
|
| 144 |
+
22.127353314656723,
|
| 145 |
+
24.828633849376587,
|
| 146 |
+
27.6486820771775,
|
| 147 |
+
30.54334096108829,
|
| 148 |
+
33.46345345363812,
|
| 149 |
+
36.358112337548896,
|
| 150 |
+
39.17816056534983,
|
| 151 |
+
41.879441100069684,
|
| 152 |
+
44.425355159339965,
|
| 153 |
+
46.78844628336223,
|
| 154 |
+
48.95093146475928,
|
| 155 |
+
50.90428855401433,
|
| 156 |
+
52.648136512661125,
|
| 157 |
+
54.18869564165987,
|
| 158 |
+
55.537098635632745,
|
| 159 |
+
56.7077647874992,
|
| 160 |
+
57.71697544677006,
|
| 161 |
+
58.58171910802236,
|
| 162 |
+
59.31882031581807,
|
| 163 |
+
59.94433101822328,
|
| 164 |
+
60.47314411958625,
|
| 165 |
+
60.918782569507,
|
| 166 |
+
61.29331890286281,
|
| 167 |
+
61.60738599471455,
|
| 168 |
+
61.87024727431288,
|
| 169 |
+
62.089902123428836,
|
| 170 |
+
62.27320880977746,
|
| 171 |
+
62.42601274014111,
|
| 172 |
+
62.55327203194878,
|
| 173 |
+
62.65917552585329,
|
| 174 |
+
62.74725058582382,
|
| 175 |
+
62.82045955451526,
|
| 176 |
+
62.88128472678279,
|
| 177 |
+
62.931802319077946,
|
| 178 |
+
62.97374626130382,
|
| 179 |
+
63.008562806439365
|
| 180 |
+
],
|
| 181 |
+
"type": "longrope"
|
| 182 |
+
},
|
| 183 |
+
"rope_theta": 10000.0,
|
| 184 |
+
"scale_depth": 1.4,
|
| 185 |
+
"scale_emb": 12,
|
| 186 |
+
"tie_word_embeddings": false,
|
| 187 |
+
"torch_dtype": "float16",
|
| 188 |
+
"transformers_version": "4.53.2",
|
| 189 |
+
"use_cache": false,
|
| 190 |
+
"vocab_size": 73448
|
| 191 |
+
}
|