Uploading config.json
Browse files- config.json +80 -0
config.json
ADDED
|
@@ -0,0 +1,80 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"DeepseekV3ForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"attention_bias": false,
|
| 6 |
+
"attention_dropout": 0.0,
|
| 7 |
+
"auto_map": {
|
| 8 |
+
"AutoConfig": "configuration_deepseek.DeepseekV3Config",
|
| 9 |
+
"AutoModel": "modeling_deepseek.DeepseekV3Model",
|
| 10 |
+
"AutoModelForCausalLM": "modeling_deepseek.DeepseekV3ForCausalLM"
|
| 11 |
+
},
|
| 12 |
+
"aux_loss_alpha": 0.001,
|
| 13 |
+
"bos_token_id": 163584,
|
| 14 |
+
"eos_token_id": 163585,
|
| 15 |
+
"ep_size": 1,
|
| 16 |
+
"first_k_dense_replace": 1,
|
| 17 |
+
"head_dim": 64,
|
| 18 |
+
"hidden_act": "silu",
|
| 19 |
+
"hidden_size": 7168,
|
| 20 |
+
"initializer_range": 0.02,
|
| 21 |
+
"intermediate_size": 18432,
|
| 22 |
+
"kv_lora_rank": 512,
|
| 23 |
+
"max_position_embeddings": 131072,
|
| 24 |
+
"model_type": "deepseek_v3",
|
| 25 |
+
"moe_intermediate_size": 2048,
|
| 26 |
+
"moe_layer_freq": 1,
|
| 27 |
+
"n_group": 1,
|
| 28 |
+
"n_routed_experts": 384,
|
| 29 |
+
"n_shared_experts": 1,
|
| 30 |
+
"norm_topk_prob": true,
|
| 31 |
+
"num_attention_heads": 64,
|
| 32 |
+
"num_experts_per_tok": 8,
|
| 33 |
+
"num_hidden_layers": 61,
|
| 34 |
+
"num_key_value_heads": 64,
|
| 35 |
+
"num_nextn_predict_layers": 0,
|
| 36 |
+
"pretraining_tp": 1,
|
| 37 |
+
"q_lora_rank": 1536,
|
| 38 |
+
"qk_head_dim": 192,
|
| 39 |
+
"qk_nope_head_dim": 128,
|
| 40 |
+
"qk_rope_head_dim": 64,
|
| 41 |
+
"quantization_config": {
|
| 42 |
+
"_load_in_4bit": true,
|
| 43 |
+
"_load_in_8bit": false,
|
| 44 |
+
"bnb_4bit_compute_dtype": "float16",
|
| 45 |
+
"bnb_4bit_quant_storage": "uint8",
|
| 46 |
+
"bnb_4bit_quant_type": "nf4",
|
| 47 |
+
"bnb_4bit_use_double_quant": true,
|
| 48 |
+
"llm_int8_enable_fp32_cpu_offload": false,
|
| 49 |
+
"llm_int8_has_fp16_weight": false,
|
| 50 |
+
"llm_int8_skip_modules": null,
|
| 51 |
+
"llm_int8_threshold": 6.0,
|
| 52 |
+
"load_in_4bit": true,
|
| 53 |
+
"load_in_8bit": false,
|
| 54 |
+
"quant_method": "bitsandbytes"
|
| 55 |
+
},
|
| 56 |
+
"rms_norm_eps": 1e-06,
|
| 57 |
+
"rope_interleave": true,
|
| 58 |
+
"rope_scaling": {
|
| 59 |
+
"beta_fast": 1.0,
|
| 60 |
+
"beta_slow": 1.0,
|
| 61 |
+
"factor": 32.0,
|
| 62 |
+
"mscale": 1.0,
|
| 63 |
+
"mscale_all_dim": 1.0,
|
| 64 |
+
"original_max_position_embeddings": 4096,
|
| 65 |
+
"rope_type": "yarn",
|
| 66 |
+
"type": "yarn"
|
| 67 |
+
},
|
| 68 |
+
"rope_theta": 50000.0,
|
| 69 |
+
"routed_scaling_factor": 2.827,
|
| 70 |
+
"scoring_func": "sigmoid",
|
| 71 |
+
"seq_aux": true,
|
| 72 |
+
"tie_word_embeddings": false,
|
| 73 |
+
"topk_group": 1,
|
| 74 |
+
"topk_method": "noaux_tc",
|
| 75 |
+
"torch_dtype": "float16",
|
| 76 |
+
"transformers_version": "4.53.2",
|
| 77 |
+
"use_cache": true,
|
| 78 |
+
"v_head_dim": 128,
|
| 79 |
+
"vocab_size": 163840
|
| 80 |
+
}
|