Upload config.json with huggingface_hub
Browse files- config.json +46 -0
config.json
ADDED
|
@@ -0,0 +1,46 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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": 2,
|
| 17 |
+
"hidden_act": "silu",
|
| 18 |
+
"hidden_size": 2304,
|
| 19 |
+
"initializer_range": 0.1,
|
| 20 |
+
"intermediate_size": 5760,
|
| 21 |
+
"max_position_embeddings": 4096,
|
| 22 |
+
"model_type": "minicpm",
|
| 23 |
+
"num_attention_heads": 36,
|
| 24 |
+
"num_hidden_layers": 40,
|
| 25 |
+
"num_key_value_heads": 36,
|
| 26 |
+
"pretraining_tp": 1,
|
| 27 |
+
"quantization": {
|
| 28 |
+
"group_size": 64,
|
| 29 |
+
"bits": 4,
|
| 30 |
+
"mode": "affine"
|
| 31 |
+
},
|
| 32 |
+
"quantization_config": {
|
| 33 |
+
"group_size": 64,
|
| 34 |
+
"bits": 4,
|
| 35 |
+
"mode": "affine"
|
| 36 |
+
},
|
| 37 |
+
"rms_norm_eps": 1e-05,
|
| 38 |
+
"rope_scaling": null,
|
| 39 |
+
"rope_theta": 10000.0,
|
| 40 |
+
"scale_depth": 1.4,
|
| 41 |
+
"scale_emb": 12,
|
| 42 |
+
"torch_dtype": "bfloat16",
|
| 43 |
+
"transformers_version": "4.40.2",
|
| 44 |
+
"use_cache": true,
|
| 45 |
+
"vocab_size": 122753
|
| 46 |
+
}
|