haoguo2000's picture
Add files using upload-large-folder tool
7e3e0c4 verified
Raw
History Blame Contribute Delete
9.03 kB
{
"architectures": [
"DeepseekV4ForCausalLM"
],
"attention_bias": false,
"attention_dropout": 0.0,
"bos_token_id": 0,
"compress_ratios": [
128,
128,
4,
128,
4,
128,
4,
128,
4,
128,
4,
128,
4,
128,
4,
128,
4,
128,
4,
128,
4,
128,
4,
128,
4,
128,
4,
128,
4,
128,
4,
128,
4,
128,
4,
128,
4,
128,
4,
128,
4,
128,
4,
128,
4,
128,
4,
128,
4,
128,
4,
128,
4,
128,
4,
128,
4,
128,
4,
128,
4,
0,
0,
0
],
"compress_rope_theta": 160000,
"dspark_block_size": 5,
"dspark_markov_rank": 512,
"dspark_noise_token_id": 128799,
"dspark_target_layer_ids": [
58,
59,
60
],
"eos_token_id": 1,
"expert_dtype": "fp4",
"hc_eps": 1e-06,
"hc_mult": 4,
"hc_sinkhorn_iters": 20,
"head_dim": 512,
"hidden_act": "silu",
"hidden_size": 7168,
"index_head_dim": 128,
"index_n_heads": 64,
"index_topk": 1024,
"initializer_range": 0.02,
"max_position_embeddings": 1048576,
"model_type": "deepseek_v4",
"moe_intermediate_size": 3072,
"n_routed_experts": 384,
"n_shared_experts": 1,
"norm_topk_prob": true,
"num_attention_heads": 128,
"num_experts_per_tok": 6,
"num_hash_layers": 3,
"num_hidden_layers": 61,
"num_key_value_heads": 1,
"num_nextn_predict_layers": 1,
"o_groups": 16,
"o_lora_rank": 1024,
"q_lora_rank": 1536,
"qk_rope_head_dim": 64,
"quantization_config": {
"activation_scheme": "dynamic",
"config_groups": {
"group_0": {
"input_activations": {
"dynamic": false,
"group_size": 16,
"num_bits": 4,
"type": "float"
},
"targets": [
"Linear"
],
"weights": {
"dynamic": false,
"group_size": 16,
"num_bits": 4,
"type": "float"
}
}
},
"fmt": "e4m3",
"group_size": 16,
"ignore": [
"*.attn.*",
"*.ffn.shared_experts.*",
"head"
],
"kv_cache_quant_algo": null,
"moe_quant_algo": "NVFP4",
"producer": {
"name": "modelopt",
"version": "dsv4-nvfp4-experts"
},
"quant_algo": "MIXED_PRECISION",
"quant_method": "fp8",
"quantized_layers": {
"layers.0.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.1.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.10.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.11.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.12.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.13.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.14.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.15.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.16.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.17.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.18.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.19.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.2.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.20.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.21.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.22.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.23.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.24.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.25.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.26.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.27.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.28.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.29.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.3.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.30.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.31.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.32.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.33.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.34.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.35.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.36.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.37.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.38.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.39.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.4.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.40.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.41.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.42.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.43.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.44.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.45.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.46.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.47.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.48.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.49.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.5.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.50.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.51.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.52.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.53.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.54.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.55.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.56.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.57.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.58.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.59.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.6.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.60.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.7.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.8.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"layers.9.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
},
"mtp.0.ffn.experts": {
"group_size": 16,
"quant_algo": "NVFP4"
}
},
"scale_fmt": "ue8m0",
"weight_block_size": [
128,
128
]
},
"rms_norm_eps": 1e-06,
"rope_scaling": {
"beta_fast": 32,
"beta_slow": 1,
"factor": 16,
"original_max_position_embeddings": 65536,
"type": "yarn"
},
"rope_theta": 10000,
"routed_scaling_factor": 2.5,
"scoring_func": "sqrtsoftplus",
"sliding_window": 128,
"swiglu_limit": 10.0,
"tie_word_embeddings": false,
"topk_method": "noaux_tc",
"torch_dtype": "bfloat16",
"transformers_version": "4.57.1",
"use_cache": true,
"vocab_size": 129280
}