DeepSeek-R1-0528-MXFP4 / config.json
haoyang-amd's picture
update config.json
2c901b8 verified
raw
history blame
55.8 kB
{
"architectures": [
"DeepseekV3ForCausalLM"
],
"attention_bias": false,
"attention_dropout": 0.0,
"auto_map": {
"AutoConfig": "configuration_deepseek.DeepseekV3Config",
"AutoModel": "modeling_deepseek.DeepseekV3Model",
"AutoModelForCausalLM": "modeling_deepseek.DeepseekV3ForCausalLM"
},
"bos_token_id": 0,
"dtype": "bfloat16",
"eos_token_id": 1,
"ep_size": 1,
"first_k_dense_replace": 3,
"hidden_act": "silu",
"hidden_size": 7168,
"initializer_range": 0.02,
"intermediate_size": 18432,
"kv_lora_rank": 512,
"max_position_embeddings": 163840,
"model_type": "deepseek_v3",
"moe_intermediate_size": 2048,
"moe_layer_freq": 1,
"n_group": 8,
"n_routed_experts": 256,
"n_shared_experts": 1,
"norm_topk_prob": true,
"num_attention_heads": 128,
"num_experts_per_tok": 8,
"num_hidden_layers": 61,
"num_key_value_heads": 128,
"num_nextn_predict_layers": 1,
"pad_token_id": 2,
"q_lora_rank": 1536,
"qk_nope_head_dim": 128,
"qk_rope_head_dim": 64,
"quantization_config": {
"algo_config": null,
"exclude": [
"model.layers.0.self_attn.q_a_proj",
"model.layers.0.self_attn.q_b_proj",
"model.layers.0.self_attn.kv_a_proj_with_mqa",
"model.layers.0.self_attn.kv_b_proj",
"model.layers.0.self_attn.o_proj",
"model.layers.1.self_attn.q_a_proj",
"model.layers.1.self_attn.q_b_proj",
"model.layers.1.self_attn.kv_a_proj_with_mqa",
"model.layers.1.self_attn.kv_b_proj",
"model.layers.1.self_attn.o_proj",
"model.layers.2.self_attn.q_a_proj",
"model.layers.2.self_attn.q_b_proj",
"model.layers.2.self_attn.kv_a_proj_with_mqa",
"model.layers.2.self_attn.kv_b_proj",
"model.layers.2.self_attn.o_proj",
"model.layers.3.self_attn.q_a_proj",
"model.layers.3.self_attn.q_b_proj",
"model.layers.3.self_attn.kv_a_proj_with_mqa",
"model.layers.3.self_attn.kv_b_proj",
"model.layers.3.self_attn.o_proj",
"model.layers.4.self_attn.q_a_proj",
"model.layers.4.self_attn.q_b_proj",
"model.layers.4.self_attn.kv_a_proj_with_mqa",
"model.layers.4.self_attn.kv_b_proj",
"model.layers.4.self_attn.o_proj",
"model.layers.5.self_attn.q_a_proj",
"model.layers.5.self_attn.q_b_proj",
"model.layers.5.self_attn.kv_a_proj_with_mqa",
"model.layers.5.self_attn.kv_b_proj",
"model.layers.5.self_attn.o_proj",
"model.layers.6.self_attn.q_a_proj",
"model.layers.6.self_attn.q_b_proj",
"model.layers.6.self_attn.kv_a_proj_with_mqa",
"model.layers.6.self_attn.kv_b_proj",
"model.layers.6.self_attn.o_proj",
"model.layers.7.self_attn.q_a_proj",
"model.layers.7.self_attn.q_b_proj",
"model.layers.7.self_attn.kv_a_proj_with_mqa",
"model.layers.7.self_attn.kv_b_proj",
"model.layers.7.self_attn.o_proj",
"model.layers.8.self_attn.q_a_proj",
"model.layers.8.self_attn.q_b_proj",
"model.layers.8.self_attn.kv_a_proj_with_mqa",
"model.layers.8.self_attn.kv_b_proj",
"model.layers.8.self_attn.o_proj",
"model.layers.9.self_attn.q_a_proj",
"model.layers.9.self_attn.q_b_proj",
"model.layers.9.self_attn.kv_a_proj_with_mqa",
"model.layers.9.self_attn.kv_b_proj",
"model.layers.9.self_attn.o_proj",
"model.layers.10.self_attn.q_a_proj",
"model.layers.10.self_attn.q_b_proj",
"model.layers.10.self_attn.kv_a_proj_with_mqa",
"model.layers.10.self_attn.kv_b_proj",
"model.layers.10.self_attn.o_proj",
"model.layers.11.self_attn.q_a_proj",
"model.layers.11.self_attn.q_b_proj",
"model.layers.11.self_attn.kv_a_proj_with_mqa",
"model.layers.11.self_attn.kv_b_proj",
"model.layers.11.self_attn.o_proj",
"model.layers.12.self_attn.q_a_proj",
"model.layers.12.self_attn.q_b_proj",
"model.layers.12.self_attn.kv_a_proj_with_mqa",
"model.layers.12.self_attn.kv_b_proj",
"model.layers.12.self_attn.o_proj",
"model.layers.13.self_attn.q_a_proj",
"model.layers.13.self_attn.q_b_proj",
"model.layers.13.self_attn.kv_a_proj_with_mqa",
"model.layers.13.self_attn.kv_b_proj",
"model.layers.13.self_attn.o_proj",
"model.layers.14.self_attn.q_a_proj",
"model.layers.14.self_attn.q_b_proj",
"model.layers.14.self_attn.kv_a_proj_with_mqa",
"model.layers.14.self_attn.kv_b_proj",
"model.layers.14.self_attn.o_proj",
"model.layers.15.self_attn.q_a_proj",
"model.layers.15.self_attn.q_b_proj",
"model.layers.15.self_attn.kv_a_proj_with_mqa",
"model.layers.15.self_attn.kv_b_proj",
"model.layers.15.self_attn.o_proj",
"model.layers.16.self_attn.q_a_proj",
"model.layers.16.self_attn.q_b_proj",
"model.layers.16.self_attn.kv_a_proj_with_mqa",
"model.layers.16.self_attn.kv_b_proj",
"model.layers.16.self_attn.o_proj",
"model.layers.17.self_attn.q_a_proj",
"model.layers.17.self_attn.q_b_proj",
"model.layers.17.self_attn.kv_a_proj_with_mqa",
"model.layers.17.self_attn.kv_b_proj",
"model.layers.17.self_attn.o_proj",
"model.layers.18.self_attn.q_a_proj",
"model.layers.18.self_attn.q_b_proj",
"model.layers.18.self_attn.kv_a_proj_with_mqa",
"model.layers.18.self_attn.kv_b_proj",
"model.layers.18.self_attn.o_proj",
"model.layers.19.self_attn.q_a_proj",
"model.layers.19.self_attn.q_b_proj",
"model.layers.19.self_attn.kv_a_proj_with_mqa",
"model.layers.19.self_attn.kv_b_proj",
"model.layers.19.self_attn.o_proj",
"model.layers.20.self_attn.q_a_proj",
"model.layers.20.self_attn.q_b_proj",
"model.layers.20.self_attn.kv_a_proj_with_mqa",
"model.layers.20.self_attn.kv_b_proj",
"model.layers.20.self_attn.o_proj",
"model.layers.21.self_attn.q_a_proj",
"model.layers.21.self_attn.q_b_proj",
"model.layers.21.self_attn.kv_a_proj_with_mqa",
"model.layers.21.self_attn.kv_b_proj",
"model.layers.21.self_attn.o_proj",
"model.layers.22.self_attn.q_a_proj",
"model.layers.22.self_attn.q_b_proj",
"model.layers.22.self_attn.kv_a_proj_with_mqa",
"model.layers.22.self_attn.kv_b_proj",
"model.layers.22.self_attn.o_proj",
"model.layers.23.self_attn.q_a_proj",
"model.layers.23.self_attn.q_b_proj",
"model.layers.23.self_attn.kv_a_proj_with_mqa",
"model.layers.23.self_attn.kv_b_proj",
"model.layers.23.self_attn.o_proj",
"model.layers.24.self_attn.q_a_proj",
"model.layers.24.self_attn.q_b_proj",
"model.layers.24.self_attn.kv_a_proj_with_mqa",
"model.layers.24.self_attn.kv_b_proj",
"model.layers.24.self_attn.o_proj",
"model.layers.25.self_attn.q_a_proj",
"model.layers.25.self_attn.q_b_proj",
"model.layers.25.self_attn.kv_a_proj_with_mqa",
"model.layers.25.self_attn.kv_b_proj",
"model.layers.25.self_attn.o_proj",
"model.layers.26.self_attn.q_a_proj",
"model.layers.26.self_attn.q_b_proj",
"model.layers.26.self_attn.kv_a_proj_with_mqa",
"model.layers.26.self_attn.kv_b_proj",
"model.layers.26.self_attn.o_proj",
"model.layers.27.self_attn.q_a_proj",
"model.layers.27.self_attn.q_b_proj",
"model.layers.27.self_attn.kv_a_proj_with_mqa",
"model.layers.27.self_attn.kv_b_proj",
"model.layers.27.self_attn.o_proj",
"model.layers.28.self_attn.q_a_proj",
"model.layers.28.self_attn.q_b_proj",
"model.layers.28.self_attn.kv_a_proj_with_mqa",
"model.layers.28.self_attn.kv_b_proj",
"model.layers.28.self_attn.o_proj",
"model.layers.29.self_attn.q_a_proj",
"model.layers.29.self_attn.q_b_proj",
"model.layers.29.self_attn.kv_a_proj_with_mqa",
"model.layers.29.self_attn.kv_b_proj",
"model.layers.29.self_attn.o_proj",
"model.layers.30.self_attn.q_a_proj",
"model.layers.30.self_attn.q_b_proj",
"model.layers.30.self_attn.kv_a_proj_with_mqa",
"model.layers.30.self_attn.kv_b_proj",
"model.layers.30.self_attn.o_proj",
"model.layers.31.self_attn.q_a_proj",
"model.layers.31.self_attn.q_b_proj",
"model.layers.31.self_attn.kv_a_proj_with_mqa",
"model.layers.31.self_attn.kv_b_proj",
"model.layers.31.self_attn.o_proj",
"model.layers.32.self_attn.q_a_proj",
"model.layers.32.self_attn.q_b_proj",
"model.layers.32.self_attn.kv_a_proj_with_mqa",
"model.layers.32.self_attn.kv_b_proj",
"model.layers.32.self_attn.o_proj",
"model.layers.33.self_attn.q_a_proj",
"model.layers.33.self_attn.q_b_proj",
"model.layers.33.self_attn.kv_a_proj_with_mqa",
"model.layers.33.self_attn.kv_b_proj",
"model.layers.33.self_attn.o_proj",
"model.layers.34.self_attn.q_a_proj",
"model.layers.34.self_attn.q_b_proj",
"model.layers.34.self_attn.kv_a_proj_with_mqa",
"model.layers.34.self_attn.kv_b_proj",
"model.layers.34.self_attn.o_proj",
"model.layers.35.self_attn.q_a_proj",
"model.layers.35.self_attn.q_b_proj",
"model.layers.35.self_attn.kv_a_proj_with_mqa",
"model.layers.35.self_attn.kv_b_proj",
"model.layers.35.self_attn.o_proj",
"model.layers.36.self_attn.q_a_proj",
"model.layers.36.self_attn.q_b_proj",
"model.layers.36.self_attn.kv_a_proj_with_mqa",
"model.layers.36.self_attn.kv_b_proj",
"model.layers.36.self_attn.o_proj",
"model.layers.37.self_attn.q_a_proj",
"model.layers.37.self_attn.q_b_proj",
"model.layers.37.self_attn.kv_a_proj_with_mqa",
"model.layers.37.self_attn.kv_b_proj",
"model.layers.37.self_attn.o_proj",
"model.layers.38.self_attn.q_a_proj",
"model.layers.38.self_attn.q_b_proj",
"model.layers.38.self_attn.kv_a_proj_with_mqa",
"model.layers.38.self_attn.kv_b_proj",
"model.layers.38.self_attn.o_proj",
"model.layers.39.self_attn.q_a_proj",
"model.layers.39.self_attn.q_b_proj",
"model.layers.39.self_attn.kv_a_proj_with_mqa",
"model.layers.39.self_attn.kv_b_proj",
"model.layers.39.self_attn.o_proj",
"model.layers.40.self_attn.q_a_proj",
"model.layers.40.self_attn.q_b_proj",
"model.layers.40.self_attn.kv_a_proj_with_mqa",
"model.layers.40.self_attn.kv_b_proj",
"model.layers.40.self_attn.o_proj",
"model.layers.41.self_attn.q_a_proj",
"model.layers.41.self_attn.q_b_proj",
"model.layers.41.self_attn.kv_a_proj_with_mqa",
"model.layers.41.self_attn.kv_b_proj",
"model.layers.41.self_attn.o_proj",
"model.layers.42.self_attn.q_a_proj",
"model.layers.42.self_attn.q_b_proj",
"model.layers.42.self_attn.kv_a_proj_with_mqa",
"model.layers.42.self_attn.kv_b_proj",
"model.layers.42.self_attn.o_proj",
"model.layers.43.self_attn.q_a_proj",
"model.layers.43.self_attn.q_b_proj",
"model.layers.43.self_attn.kv_a_proj_with_mqa",
"model.layers.43.self_attn.kv_b_proj",
"model.layers.43.self_attn.o_proj",
"model.layers.44.self_attn.q_a_proj",
"model.layers.44.self_attn.q_b_proj",
"model.layers.44.self_attn.kv_a_proj_with_mqa",
"model.layers.44.self_attn.kv_b_proj",
"model.layers.44.self_attn.o_proj",
"model.layers.45.self_attn.q_a_proj",
"model.layers.45.self_attn.q_b_proj",
"model.layers.45.self_attn.kv_a_proj_with_mqa",
"model.layers.45.self_attn.kv_b_proj",
"model.layers.45.self_attn.o_proj",
"model.layers.46.self_attn.q_a_proj",
"model.layers.46.self_attn.q_b_proj",
"model.layers.46.self_attn.kv_a_proj_with_mqa",
"model.layers.46.self_attn.kv_b_proj",
"model.layers.46.self_attn.o_proj",
"model.layers.47.self_attn.q_a_proj",
"model.layers.47.self_attn.q_b_proj",
"model.layers.47.self_attn.kv_a_proj_with_mqa",
"model.layers.47.self_attn.kv_b_proj",
"model.layers.47.self_attn.o_proj",
"model.layers.48.self_attn.q_a_proj",
"model.layers.48.self_attn.q_b_proj",
"model.layers.48.self_attn.kv_a_proj_with_mqa",
"model.layers.48.self_attn.kv_b_proj",
"model.layers.48.self_attn.o_proj",
"model.layers.49.self_attn.q_a_proj",
"model.layers.49.self_attn.q_b_proj",
"model.layers.49.self_attn.kv_a_proj_with_mqa",
"model.layers.49.self_attn.kv_b_proj",
"model.layers.49.self_attn.o_proj",
"model.layers.50.self_attn.q_a_proj",
"model.layers.50.self_attn.q_b_proj",
"model.layers.50.self_attn.kv_a_proj_with_mqa",
"model.layers.50.self_attn.kv_b_proj",
"model.layers.50.self_attn.o_proj",
"model.layers.51.self_attn.q_a_proj",
"model.layers.51.self_attn.q_b_proj",
"model.layers.51.self_attn.kv_a_proj_with_mqa",
"model.layers.51.self_attn.kv_b_proj",
"model.layers.51.self_attn.o_proj",
"model.layers.52.self_attn.q_a_proj",
"model.layers.52.self_attn.q_b_proj",
"model.layers.52.self_attn.kv_a_proj_with_mqa",
"model.layers.52.self_attn.kv_b_proj",
"model.layers.52.self_attn.o_proj",
"model.layers.53.self_attn.q_a_proj",
"model.layers.53.self_attn.q_b_proj",
"model.layers.53.self_attn.kv_a_proj_with_mqa",
"model.layers.53.self_attn.kv_b_proj",
"model.layers.53.self_attn.o_proj",
"model.layers.54.self_attn.q_a_proj",
"model.layers.54.self_attn.q_b_proj",
"model.layers.54.self_attn.kv_a_proj_with_mqa",
"model.layers.54.self_attn.kv_b_proj",
"model.layers.54.self_attn.o_proj",
"model.layers.55.self_attn.q_a_proj",
"model.layers.55.self_attn.q_b_proj",
"model.layers.55.self_attn.kv_a_proj_with_mqa",
"model.layers.55.self_attn.kv_b_proj",
"model.layers.55.self_attn.o_proj",
"model.layers.56.self_attn.q_a_proj",
"model.layers.56.self_attn.q_b_proj",
"model.layers.56.self_attn.kv_a_proj_with_mqa",
"model.layers.56.self_attn.kv_b_proj",
"model.layers.56.self_attn.o_proj",
"model.layers.57.self_attn.q_a_proj",
"model.layers.57.self_attn.q_b_proj",
"model.layers.57.self_attn.kv_a_proj_with_mqa",
"model.layers.57.self_attn.kv_b_proj",
"model.layers.57.self_attn.o_proj",
"model.layers.58.self_attn.q_a_proj",
"model.layers.58.self_attn.q_b_proj",
"model.layers.58.self_attn.kv_a_proj_with_mqa",
"model.layers.58.self_attn.kv_b_proj",
"model.layers.58.self_attn.o_proj",
"model.layers.59.self_attn.q_a_proj",
"model.layers.59.self_attn.q_b_proj",
"model.layers.59.self_attn.kv_a_proj_with_mqa",
"model.layers.59.self_attn.kv_b_proj",
"model.layers.59.self_attn.o_proj",
"model.layers.60.self_attn.q_a_proj",
"model.layers.60.self_attn.q_b_proj",
"model.layers.60.self_attn.kv_a_proj_with_mqa",
"model.layers.60.self_attn.kv_b_proj",
"model.layers.60.self_attn.o_proj",
"model.layers.61.embed_tokens",
"model.layers.61.eh_proj",
"model.layers.61.self_attn.q_a_proj",
"model.layers.61.self_attn.q_b_proj",
"model.layers.61.self_attn.kv_a_proj_with_mqa",
"model.layers.61.self_attn.kv_b_proj",
"model.layers.61.self_attn.o_proj",
"model.layers.61.mlp.experts.0.gate_proj",
"model.layers.61.mlp.experts.0.up_proj",
"model.layers.61.mlp.experts.0.down_proj",
"model.layers.61.mlp.experts.1.gate_proj",
"model.layers.61.mlp.experts.1.up_proj",
"model.layers.61.mlp.experts.1.down_proj",
"model.layers.61.mlp.experts.2.gate_proj",
"model.layers.61.mlp.experts.2.up_proj",
"model.layers.61.mlp.experts.2.down_proj",
"model.layers.61.mlp.experts.3.gate_proj",
"model.layers.61.mlp.experts.3.up_proj",
"model.layers.61.mlp.experts.3.down_proj",
"model.layers.61.mlp.experts.4.gate_proj",
"model.layers.61.mlp.experts.4.up_proj",
"model.layers.61.mlp.experts.4.down_proj",
"model.layers.61.mlp.experts.5.gate_proj",
"model.layers.61.mlp.experts.5.up_proj",
"model.layers.61.mlp.experts.5.down_proj",
"model.layers.61.mlp.experts.6.gate_proj",
"model.layers.61.mlp.experts.6.up_proj",
"model.layers.61.mlp.experts.6.down_proj",
"model.layers.61.mlp.experts.7.gate_proj",
"model.layers.61.mlp.experts.7.up_proj",
"model.layers.61.mlp.experts.7.down_proj",
"model.layers.61.mlp.experts.8.gate_proj",
"model.layers.61.mlp.experts.8.up_proj",
"model.layers.61.mlp.experts.8.down_proj",
"model.layers.61.mlp.experts.9.gate_proj",
"model.layers.61.mlp.experts.9.up_proj",
"model.layers.61.mlp.experts.9.down_proj",
"model.layers.61.mlp.experts.10.gate_proj",
"model.layers.61.mlp.experts.10.up_proj",
"model.layers.61.mlp.experts.10.down_proj",
"model.layers.61.mlp.experts.11.gate_proj",
"model.layers.61.mlp.experts.11.up_proj",
"model.layers.61.mlp.experts.11.down_proj",
"model.layers.61.mlp.experts.12.gate_proj",
"model.layers.61.mlp.experts.12.up_proj",
"model.layers.61.mlp.experts.12.down_proj",
"model.layers.61.mlp.experts.13.gate_proj",
"model.layers.61.mlp.experts.13.up_proj",
"model.layers.61.mlp.experts.13.down_proj",
"model.layers.61.mlp.experts.14.gate_proj",
"model.layers.61.mlp.experts.14.up_proj",
"model.layers.61.mlp.experts.14.down_proj",
"model.layers.61.mlp.experts.15.gate_proj",
"model.layers.61.mlp.experts.15.up_proj",
"model.layers.61.mlp.experts.15.down_proj",
"model.layers.61.mlp.experts.16.gate_proj",
"model.layers.61.mlp.experts.16.up_proj",
"model.layers.61.mlp.experts.16.down_proj",
"model.layers.61.mlp.experts.17.gate_proj",
"model.layers.61.mlp.experts.17.up_proj",
"model.layers.61.mlp.experts.17.down_proj",
"model.layers.61.mlp.experts.18.gate_proj",
"model.layers.61.mlp.experts.18.up_proj",
"model.layers.61.mlp.experts.18.down_proj",
"model.layers.61.mlp.experts.19.gate_proj",
"model.layers.61.mlp.experts.19.up_proj",
"model.layers.61.mlp.experts.19.down_proj",
"model.layers.61.mlp.experts.20.gate_proj",
"model.layers.61.mlp.experts.20.up_proj",
"model.layers.61.mlp.experts.20.down_proj",
"model.layers.61.mlp.experts.21.gate_proj",
"model.layers.61.mlp.experts.21.up_proj",
"model.layers.61.mlp.experts.21.down_proj",
"model.layers.61.mlp.experts.22.gate_proj",
"model.layers.61.mlp.experts.22.up_proj",
"model.layers.61.mlp.experts.22.down_proj",
"model.layers.61.mlp.experts.23.gate_proj",
"model.layers.61.mlp.experts.23.up_proj",
"model.layers.61.mlp.experts.23.down_proj",
"model.layers.61.mlp.experts.24.gate_proj",
"model.layers.61.mlp.experts.24.up_proj",
"model.layers.61.mlp.experts.24.down_proj",
"model.layers.61.mlp.experts.25.gate_proj",
"model.layers.61.mlp.experts.25.up_proj",
"model.layers.61.mlp.experts.25.down_proj",
"model.layers.61.mlp.experts.26.gate_proj",
"model.layers.61.mlp.experts.26.up_proj",
"model.layers.61.mlp.experts.26.down_proj",
"model.layers.61.mlp.experts.27.gate_proj",
"model.layers.61.mlp.experts.27.up_proj",
"model.layers.61.mlp.experts.27.down_proj",
"model.layers.61.mlp.experts.28.gate_proj",
"model.layers.61.mlp.experts.28.up_proj",
"model.layers.61.mlp.experts.28.down_proj",
"model.layers.61.mlp.experts.29.gate_proj",
"model.layers.61.mlp.experts.29.up_proj",
"model.layers.61.mlp.experts.29.down_proj",
"model.layers.61.mlp.experts.30.gate_proj",
"model.layers.61.mlp.experts.30.up_proj",
"model.layers.61.mlp.experts.30.down_proj",
"model.layers.61.mlp.experts.31.gate_proj",
"model.layers.61.mlp.experts.31.up_proj",
"model.layers.61.mlp.experts.31.down_proj",
"model.layers.61.mlp.experts.32.gate_proj",
"model.layers.61.mlp.experts.32.up_proj",
"model.layers.61.mlp.experts.32.down_proj",
"model.layers.61.mlp.experts.33.gate_proj",
"model.layers.61.mlp.experts.33.up_proj",
"model.layers.61.mlp.experts.33.down_proj",
"model.layers.61.mlp.experts.34.gate_proj",
"model.layers.61.mlp.experts.34.up_proj",
"model.layers.61.mlp.experts.34.down_proj",
"model.layers.61.mlp.experts.35.gate_proj",
"model.layers.61.mlp.experts.35.up_proj",
"model.layers.61.mlp.experts.35.down_proj",
"model.layers.61.mlp.experts.36.gate_proj",
"model.layers.61.mlp.experts.36.up_proj",
"model.layers.61.mlp.experts.36.down_proj",
"model.layers.61.mlp.experts.37.gate_proj",
"model.layers.61.mlp.experts.37.up_proj",
"model.layers.61.mlp.experts.37.down_proj",
"model.layers.61.mlp.experts.38.gate_proj",
"model.layers.61.mlp.experts.38.up_proj",
"model.layers.61.mlp.experts.38.down_proj",
"model.layers.61.mlp.experts.39.gate_proj",
"model.layers.61.mlp.experts.39.up_proj",
"model.layers.61.mlp.experts.39.down_proj",
"model.layers.61.mlp.experts.40.gate_proj",
"model.layers.61.mlp.experts.40.up_proj",
"model.layers.61.mlp.experts.40.down_proj",
"model.layers.61.mlp.experts.41.gate_proj",
"model.layers.61.mlp.experts.41.up_proj",
"model.layers.61.mlp.experts.41.down_proj",
"model.layers.61.mlp.experts.42.gate_proj",
"model.layers.61.mlp.experts.42.up_proj",
"model.layers.61.mlp.experts.42.down_proj",
"model.layers.61.mlp.experts.43.gate_proj",
"model.layers.61.mlp.experts.43.up_proj",
"model.layers.61.mlp.experts.43.down_proj",
"model.layers.61.mlp.experts.44.gate_proj",
"model.layers.61.mlp.experts.44.up_proj",
"model.layers.61.mlp.experts.44.down_proj",
"model.layers.61.mlp.experts.45.gate_proj",
"model.layers.61.mlp.experts.45.up_proj",
"model.layers.61.mlp.experts.45.down_proj",
"model.layers.61.mlp.experts.46.gate_proj",
"model.layers.61.mlp.experts.46.up_proj",
"model.layers.61.mlp.experts.46.down_proj",
"model.layers.61.mlp.experts.47.gate_proj",
"model.layers.61.mlp.experts.47.up_proj",
"model.layers.61.mlp.experts.47.down_proj",
"model.layers.61.mlp.experts.48.gate_proj",
"model.layers.61.mlp.experts.48.up_proj",
"model.layers.61.mlp.experts.48.down_proj",
"model.layers.61.mlp.experts.49.gate_proj",
"model.layers.61.mlp.experts.49.up_proj",
"model.layers.61.mlp.experts.49.down_proj",
"model.layers.61.mlp.experts.50.gate_proj",
"model.layers.61.mlp.experts.50.up_proj",
"model.layers.61.mlp.experts.50.down_proj",
"model.layers.61.mlp.experts.51.gate_proj",
"model.layers.61.mlp.experts.51.up_proj",
"model.layers.61.mlp.experts.51.down_proj",
"model.layers.61.mlp.experts.52.gate_proj",
"model.layers.61.mlp.experts.52.up_proj",
"model.layers.61.mlp.experts.52.down_proj",
"model.layers.61.mlp.experts.53.gate_proj",
"model.layers.61.mlp.experts.53.up_proj",
"model.layers.61.mlp.experts.53.down_proj",
"model.layers.61.mlp.experts.54.gate_proj",
"model.layers.61.mlp.experts.54.up_proj",
"model.layers.61.mlp.experts.54.down_proj",
"model.layers.61.mlp.experts.55.gate_proj",
"model.layers.61.mlp.experts.55.up_proj",
"model.layers.61.mlp.experts.55.down_proj",
"model.layers.61.mlp.experts.56.gate_proj",
"model.layers.61.mlp.experts.56.up_proj",
"model.layers.61.mlp.experts.56.down_proj",
"model.layers.61.mlp.experts.57.gate_proj",
"model.layers.61.mlp.experts.57.up_proj",
"model.layers.61.mlp.experts.57.down_proj",
"model.layers.61.mlp.experts.58.gate_proj",
"model.layers.61.mlp.experts.58.up_proj",
"model.layers.61.mlp.experts.58.down_proj",
"model.layers.61.mlp.experts.59.gate_proj",
"model.layers.61.mlp.experts.59.up_proj",
"model.layers.61.mlp.experts.59.down_proj",
"model.layers.61.mlp.experts.60.gate_proj",
"model.layers.61.mlp.experts.60.up_proj",
"model.layers.61.mlp.experts.60.down_proj",
"model.layers.61.mlp.experts.61.gate_proj",
"model.layers.61.mlp.experts.61.up_proj",
"model.layers.61.mlp.experts.61.down_proj",
"model.layers.61.mlp.experts.62.gate_proj",
"model.layers.61.mlp.experts.62.up_proj",
"model.layers.61.mlp.experts.62.down_proj",
"model.layers.61.mlp.experts.63.gate_proj",
"model.layers.61.mlp.experts.63.up_proj",
"model.layers.61.mlp.experts.63.down_proj",
"model.layers.61.mlp.experts.64.gate_proj",
"model.layers.61.mlp.experts.64.up_proj",
"model.layers.61.mlp.experts.64.down_proj",
"model.layers.61.mlp.experts.65.gate_proj",
"model.layers.61.mlp.experts.65.up_proj",
"model.layers.61.mlp.experts.65.down_proj",
"model.layers.61.mlp.experts.66.gate_proj",
"model.layers.61.mlp.experts.66.up_proj",
"model.layers.61.mlp.experts.66.down_proj",
"model.layers.61.mlp.experts.67.gate_proj",
"model.layers.61.mlp.experts.67.up_proj",
"model.layers.61.mlp.experts.67.down_proj",
"model.layers.61.mlp.experts.68.gate_proj",
"model.layers.61.mlp.experts.68.up_proj",
"model.layers.61.mlp.experts.68.down_proj",
"model.layers.61.mlp.experts.69.gate_proj",
"model.layers.61.mlp.experts.69.up_proj",
"model.layers.61.mlp.experts.69.down_proj",
"model.layers.61.mlp.experts.70.gate_proj",
"model.layers.61.mlp.experts.70.up_proj",
"model.layers.61.mlp.experts.70.down_proj",
"model.layers.61.mlp.experts.71.gate_proj",
"model.layers.61.mlp.experts.71.up_proj",
"model.layers.61.mlp.experts.71.down_proj",
"model.layers.61.mlp.experts.72.gate_proj",
"model.layers.61.mlp.experts.72.up_proj",
"model.layers.61.mlp.experts.72.down_proj",
"model.layers.61.mlp.experts.73.gate_proj",
"model.layers.61.mlp.experts.73.up_proj",
"model.layers.61.mlp.experts.73.down_proj",
"model.layers.61.mlp.experts.74.gate_proj",
"model.layers.61.mlp.experts.74.up_proj",
"model.layers.61.mlp.experts.74.down_proj",
"model.layers.61.mlp.experts.75.gate_proj",
"model.layers.61.mlp.experts.75.up_proj",
"model.layers.61.mlp.experts.75.down_proj",
"model.layers.61.mlp.experts.76.gate_proj",
"model.layers.61.mlp.experts.76.up_proj",
"model.layers.61.mlp.experts.76.down_proj",
"model.layers.61.mlp.experts.77.gate_proj",
"model.layers.61.mlp.experts.77.up_proj",
"model.layers.61.mlp.experts.77.down_proj",
"model.layers.61.mlp.experts.78.gate_proj",
"model.layers.61.mlp.experts.78.up_proj",
"model.layers.61.mlp.experts.78.down_proj",
"model.layers.61.mlp.experts.79.gate_proj",
"model.layers.61.mlp.experts.79.up_proj",
"model.layers.61.mlp.experts.79.down_proj",
"model.layers.61.mlp.experts.80.gate_proj",
"model.layers.61.mlp.experts.80.up_proj",
"model.layers.61.mlp.experts.80.down_proj",
"model.layers.61.mlp.experts.81.gate_proj",
"model.layers.61.mlp.experts.81.up_proj",
"model.layers.61.mlp.experts.81.down_proj",
"model.layers.61.mlp.experts.82.gate_proj",
"model.layers.61.mlp.experts.82.up_proj",
"model.layers.61.mlp.experts.82.down_proj",
"model.layers.61.mlp.experts.83.gate_proj",
"model.layers.61.mlp.experts.83.up_proj",
"model.layers.61.mlp.experts.83.down_proj",
"model.layers.61.mlp.experts.84.gate_proj",
"model.layers.61.mlp.experts.84.up_proj",
"model.layers.61.mlp.experts.84.down_proj",
"model.layers.61.mlp.experts.85.gate_proj",
"model.layers.61.mlp.experts.85.up_proj",
"model.layers.61.mlp.experts.85.down_proj",
"model.layers.61.mlp.experts.86.gate_proj",
"model.layers.61.mlp.experts.86.up_proj",
"model.layers.61.mlp.experts.86.down_proj",
"model.layers.61.mlp.experts.87.gate_proj",
"model.layers.61.mlp.experts.87.up_proj",
"model.layers.61.mlp.experts.87.down_proj",
"model.layers.61.mlp.experts.88.gate_proj",
"model.layers.61.mlp.experts.88.up_proj",
"model.layers.61.mlp.experts.88.down_proj",
"model.layers.61.mlp.experts.89.gate_proj",
"model.layers.61.mlp.experts.89.up_proj",
"model.layers.61.mlp.experts.89.down_proj",
"model.layers.61.mlp.experts.90.gate_proj",
"model.layers.61.mlp.experts.90.up_proj",
"model.layers.61.mlp.experts.90.down_proj",
"model.layers.61.mlp.experts.91.gate_proj",
"model.layers.61.mlp.experts.91.up_proj",
"model.layers.61.mlp.experts.91.down_proj",
"model.layers.61.mlp.experts.92.gate_proj",
"model.layers.61.mlp.experts.92.up_proj",
"model.layers.61.mlp.experts.92.down_proj",
"model.layers.61.mlp.experts.93.gate_proj",
"model.layers.61.mlp.experts.93.up_proj",
"model.layers.61.mlp.experts.93.down_proj",
"model.layers.61.mlp.experts.94.gate_proj",
"model.layers.61.mlp.experts.94.up_proj",
"model.layers.61.mlp.experts.94.down_proj",
"model.layers.61.mlp.experts.95.gate_proj",
"model.layers.61.mlp.experts.95.up_proj",
"model.layers.61.mlp.experts.95.down_proj",
"model.layers.61.mlp.experts.96.gate_proj",
"model.layers.61.mlp.experts.96.up_proj",
"model.layers.61.mlp.experts.96.down_proj",
"model.layers.61.mlp.experts.97.gate_proj",
"model.layers.61.mlp.experts.97.up_proj",
"model.layers.61.mlp.experts.97.down_proj",
"model.layers.61.mlp.experts.98.gate_proj",
"model.layers.61.mlp.experts.98.up_proj",
"model.layers.61.mlp.experts.98.down_proj",
"model.layers.61.mlp.experts.99.gate_proj",
"model.layers.61.mlp.experts.99.up_proj",
"model.layers.61.mlp.experts.99.down_proj",
"model.layers.61.mlp.experts.100.gate_proj",
"model.layers.61.mlp.experts.100.up_proj",
"model.layers.61.mlp.experts.100.down_proj",
"model.layers.61.mlp.experts.101.gate_proj",
"model.layers.61.mlp.experts.101.up_proj",
"model.layers.61.mlp.experts.101.down_proj",
"model.layers.61.mlp.experts.102.gate_proj",
"model.layers.61.mlp.experts.102.up_proj",
"model.layers.61.mlp.experts.102.down_proj",
"model.layers.61.mlp.experts.103.gate_proj",
"model.layers.61.mlp.experts.103.up_proj",
"model.layers.61.mlp.experts.103.down_proj",
"model.layers.61.mlp.experts.104.gate_proj",
"model.layers.61.mlp.experts.104.up_proj",
"model.layers.61.mlp.experts.104.down_proj",
"model.layers.61.mlp.experts.105.gate_proj",
"model.layers.61.mlp.experts.105.up_proj",
"model.layers.61.mlp.experts.105.down_proj",
"model.layers.61.mlp.experts.106.gate_proj",
"model.layers.61.mlp.experts.106.up_proj",
"model.layers.61.mlp.experts.106.down_proj",
"model.layers.61.mlp.experts.107.gate_proj",
"model.layers.61.mlp.experts.107.up_proj",
"model.layers.61.mlp.experts.107.down_proj",
"model.layers.61.mlp.experts.108.gate_proj",
"model.layers.61.mlp.experts.108.up_proj",
"model.layers.61.mlp.experts.108.down_proj",
"model.layers.61.mlp.experts.109.gate_proj",
"model.layers.61.mlp.experts.109.up_proj",
"model.layers.61.mlp.experts.109.down_proj",
"model.layers.61.mlp.experts.110.gate_proj",
"model.layers.61.mlp.experts.110.up_proj",
"model.layers.61.mlp.experts.110.down_proj",
"model.layers.61.mlp.experts.111.gate_proj",
"model.layers.61.mlp.experts.111.up_proj",
"model.layers.61.mlp.experts.111.down_proj",
"model.layers.61.mlp.experts.112.gate_proj",
"model.layers.61.mlp.experts.112.up_proj",
"model.layers.61.mlp.experts.112.down_proj",
"model.layers.61.mlp.experts.113.gate_proj",
"model.layers.61.mlp.experts.113.up_proj",
"model.layers.61.mlp.experts.113.down_proj",
"model.layers.61.mlp.experts.114.gate_proj",
"model.layers.61.mlp.experts.114.up_proj",
"model.layers.61.mlp.experts.114.down_proj",
"model.layers.61.mlp.experts.115.gate_proj",
"model.layers.61.mlp.experts.115.up_proj",
"model.layers.61.mlp.experts.115.down_proj",
"model.layers.61.mlp.experts.116.gate_proj",
"model.layers.61.mlp.experts.116.up_proj",
"model.layers.61.mlp.experts.116.down_proj",
"model.layers.61.mlp.experts.117.gate_proj",
"model.layers.61.mlp.experts.117.up_proj",
"model.layers.61.mlp.experts.117.down_proj",
"model.layers.61.mlp.experts.118.gate_proj",
"model.layers.61.mlp.experts.118.up_proj",
"model.layers.61.mlp.experts.118.down_proj",
"model.layers.61.mlp.experts.119.gate_proj",
"model.layers.61.mlp.experts.119.up_proj",
"model.layers.61.mlp.experts.119.down_proj",
"model.layers.61.mlp.experts.120.gate_proj",
"model.layers.61.mlp.experts.120.up_proj",
"model.layers.61.mlp.experts.120.down_proj",
"model.layers.61.mlp.experts.121.gate_proj",
"model.layers.61.mlp.experts.121.up_proj",
"model.layers.61.mlp.experts.121.down_proj",
"model.layers.61.mlp.experts.122.gate_proj",
"model.layers.61.mlp.experts.122.up_proj",
"model.layers.61.mlp.experts.122.down_proj",
"model.layers.61.mlp.experts.123.gate_proj",
"model.layers.61.mlp.experts.123.up_proj",
"model.layers.61.mlp.experts.123.down_proj",
"model.layers.61.mlp.experts.124.gate_proj",
"model.layers.61.mlp.experts.124.up_proj",
"model.layers.61.mlp.experts.124.down_proj",
"model.layers.61.mlp.experts.125.gate_proj",
"model.layers.61.mlp.experts.125.up_proj",
"model.layers.61.mlp.experts.125.down_proj",
"model.layers.61.mlp.experts.126.gate_proj",
"model.layers.61.mlp.experts.126.up_proj",
"model.layers.61.mlp.experts.126.down_proj",
"model.layers.61.mlp.experts.127.gate_proj",
"model.layers.61.mlp.experts.127.up_proj",
"model.layers.61.mlp.experts.127.down_proj",
"model.layers.61.mlp.experts.128.gate_proj",
"model.layers.61.mlp.experts.128.up_proj",
"model.layers.61.mlp.experts.128.down_proj",
"model.layers.61.mlp.experts.129.gate_proj",
"model.layers.61.mlp.experts.129.up_proj",
"model.layers.61.mlp.experts.129.down_proj",
"model.layers.61.mlp.experts.130.gate_proj",
"model.layers.61.mlp.experts.130.up_proj",
"model.layers.61.mlp.experts.130.down_proj",
"model.layers.61.mlp.experts.131.gate_proj",
"model.layers.61.mlp.experts.131.up_proj",
"model.layers.61.mlp.experts.131.down_proj",
"model.layers.61.mlp.experts.132.gate_proj",
"model.layers.61.mlp.experts.132.up_proj",
"model.layers.61.mlp.experts.132.down_proj",
"model.layers.61.mlp.experts.133.gate_proj",
"model.layers.61.mlp.experts.133.up_proj",
"model.layers.61.mlp.experts.133.down_proj",
"model.layers.61.mlp.experts.134.gate_proj",
"model.layers.61.mlp.experts.134.up_proj",
"model.layers.61.mlp.experts.134.down_proj",
"model.layers.61.mlp.experts.135.gate_proj",
"model.layers.61.mlp.experts.135.up_proj",
"model.layers.61.mlp.experts.135.down_proj",
"model.layers.61.mlp.experts.136.gate_proj",
"model.layers.61.mlp.experts.136.up_proj",
"model.layers.61.mlp.experts.136.down_proj",
"model.layers.61.mlp.experts.137.gate_proj",
"model.layers.61.mlp.experts.137.up_proj",
"model.layers.61.mlp.experts.137.down_proj",
"model.layers.61.mlp.experts.138.gate_proj",
"model.layers.61.mlp.experts.138.up_proj",
"model.layers.61.mlp.experts.138.down_proj",
"model.layers.61.mlp.experts.139.gate_proj",
"model.layers.61.mlp.experts.139.up_proj",
"model.layers.61.mlp.experts.139.down_proj",
"model.layers.61.mlp.experts.140.gate_proj",
"model.layers.61.mlp.experts.140.up_proj",
"model.layers.61.mlp.experts.140.down_proj",
"model.layers.61.mlp.experts.141.gate_proj",
"model.layers.61.mlp.experts.141.up_proj",
"model.layers.61.mlp.experts.141.down_proj",
"model.layers.61.mlp.experts.142.gate_proj",
"model.layers.61.mlp.experts.142.up_proj",
"model.layers.61.mlp.experts.142.down_proj",
"model.layers.61.mlp.experts.143.gate_proj",
"model.layers.61.mlp.experts.143.up_proj",
"model.layers.61.mlp.experts.143.down_proj",
"model.layers.61.mlp.experts.144.gate_proj",
"model.layers.61.mlp.experts.144.up_proj",
"model.layers.61.mlp.experts.144.down_proj",
"model.layers.61.mlp.experts.145.gate_proj",
"model.layers.61.mlp.experts.145.up_proj",
"model.layers.61.mlp.experts.145.down_proj",
"model.layers.61.mlp.experts.146.gate_proj",
"model.layers.61.mlp.experts.146.up_proj",
"model.layers.61.mlp.experts.146.down_proj",
"model.layers.61.mlp.experts.147.gate_proj",
"model.layers.61.mlp.experts.147.up_proj",
"model.layers.61.mlp.experts.147.down_proj",
"model.layers.61.mlp.experts.148.gate_proj",
"model.layers.61.mlp.experts.148.up_proj",
"model.layers.61.mlp.experts.148.down_proj",
"model.layers.61.mlp.experts.149.gate_proj",
"model.layers.61.mlp.experts.149.up_proj",
"model.layers.61.mlp.experts.149.down_proj",
"model.layers.61.mlp.experts.150.gate_proj",
"model.layers.61.mlp.experts.150.up_proj",
"model.layers.61.mlp.experts.150.down_proj",
"model.layers.61.mlp.experts.151.gate_proj",
"model.layers.61.mlp.experts.151.up_proj",
"model.layers.61.mlp.experts.151.down_proj",
"model.layers.61.mlp.experts.152.gate_proj",
"model.layers.61.mlp.experts.152.up_proj",
"model.layers.61.mlp.experts.152.down_proj",
"model.layers.61.mlp.experts.153.gate_proj",
"model.layers.61.mlp.experts.153.up_proj",
"model.layers.61.mlp.experts.153.down_proj",
"model.layers.61.mlp.experts.154.gate_proj",
"model.layers.61.mlp.experts.154.up_proj",
"model.layers.61.mlp.experts.154.down_proj",
"model.layers.61.mlp.experts.155.gate_proj",
"model.layers.61.mlp.experts.155.up_proj",
"model.layers.61.mlp.experts.155.down_proj",
"model.layers.61.mlp.experts.156.gate_proj",
"model.layers.61.mlp.experts.156.up_proj",
"model.layers.61.mlp.experts.156.down_proj",
"model.layers.61.mlp.experts.157.gate_proj",
"model.layers.61.mlp.experts.157.up_proj",
"model.layers.61.mlp.experts.157.down_proj",
"model.layers.61.mlp.experts.158.gate_proj",
"model.layers.61.mlp.experts.158.up_proj",
"model.layers.61.mlp.experts.158.down_proj",
"model.layers.61.mlp.experts.159.gate_proj",
"model.layers.61.mlp.experts.159.up_proj",
"model.layers.61.mlp.experts.159.down_proj",
"model.layers.61.mlp.experts.160.gate_proj",
"model.layers.61.mlp.experts.160.up_proj",
"model.layers.61.mlp.experts.160.down_proj",
"model.layers.61.mlp.experts.161.gate_proj",
"model.layers.61.mlp.experts.161.up_proj",
"model.layers.61.mlp.experts.161.down_proj",
"model.layers.61.mlp.experts.162.gate_proj",
"model.layers.61.mlp.experts.162.up_proj",
"model.layers.61.mlp.experts.162.down_proj",
"model.layers.61.mlp.experts.163.gate_proj",
"model.layers.61.mlp.experts.163.up_proj",
"model.layers.61.mlp.experts.163.down_proj",
"model.layers.61.mlp.experts.164.gate_proj",
"model.layers.61.mlp.experts.164.up_proj",
"model.layers.61.mlp.experts.164.down_proj",
"model.layers.61.mlp.experts.165.gate_proj",
"model.layers.61.mlp.experts.165.up_proj",
"model.layers.61.mlp.experts.165.down_proj",
"model.layers.61.mlp.experts.166.gate_proj",
"model.layers.61.mlp.experts.166.up_proj",
"model.layers.61.mlp.experts.166.down_proj",
"model.layers.61.mlp.experts.167.gate_proj",
"model.layers.61.mlp.experts.167.up_proj",
"model.layers.61.mlp.experts.167.down_proj",
"model.layers.61.mlp.experts.168.gate_proj",
"model.layers.61.mlp.experts.168.up_proj",
"model.layers.61.mlp.experts.168.down_proj",
"model.layers.61.mlp.experts.169.gate_proj",
"model.layers.61.mlp.experts.169.up_proj",
"model.layers.61.mlp.experts.169.down_proj",
"model.layers.61.mlp.experts.170.gate_proj",
"model.layers.61.mlp.experts.170.up_proj",
"model.layers.61.mlp.experts.170.down_proj",
"model.layers.61.mlp.experts.171.gate_proj",
"model.layers.61.mlp.experts.171.up_proj",
"model.layers.61.mlp.experts.171.down_proj",
"model.layers.61.mlp.experts.172.gate_proj",
"model.layers.61.mlp.experts.172.up_proj",
"model.layers.61.mlp.experts.172.down_proj",
"model.layers.61.mlp.experts.173.gate_proj",
"model.layers.61.mlp.experts.173.up_proj",
"model.layers.61.mlp.experts.173.down_proj",
"model.layers.61.mlp.experts.174.gate_proj",
"model.layers.61.mlp.experts.174.up_proj",
"model.layers.61.mlp.experts.174.down_proj",
"model.layers.61.mlp.experts.175.gate_proj",
"model.layers.61.mlp.experts.175.up_proj",
"model.layers.61.mlp.experts.175.down_proj",
"model.layers.61.mlp.experts.176.gate_proj",
"model.layers.61.mlp.experts.176.up_proj",
"model.layers.61.mlp.experts.176.down_proj",
"model.layers.61.mlp.experts.177.gate_proj",
"model.layers.61.mlp.experts.177.up_proj",
"model.layers.61.mlp.experts.177.down_proj",
"model.layers.61.mlp.experts.178.gate_proj",
"model.layers.61.mlp.experts.178.up_proj",
"model.layers.61.mlp.experts.178.down_proj",
"model.layers.61.mlp.experts.179.gate_proj",
"model.layers.61.mlp.experts.179.up_proj",
"model.layers.61.mlp.experts.179.down_proj",
"model.layers.61.mlp.experts.180.gate_proj",
"model.layers.61.mlp.experts.180.up_proj",
"model.layers.61.mlp.experts.180.down_proj",
"model.layers.61.mlp.experts.181.gate_proj",
"model.layers.61.mlp.experts.181.up_proj",
"model.layers.61.mlp.experts.181.down_proj",
"model.layers.61.mlp.experts.182.gate_proj",
"model.layers.61.mlp.experts.182.up_proj",
"model.layers.61.mlp.experts.182.down_proj",
"model.layers.61.mlp.experts.183.gate_proj",
"model.layers.61.mlp.experts.183.up_proj",
"model.layers.61.mlp.experts.183.down_proj",
"model.layers.61.mlp.experts.184.gate_proj",
"model.layers.61.mlp.experts.184.up_proj",
"model.layers.61.mlp.experts.184.down_proj",
"model.layers.61.mlp.experts.185.gate_proj",
"model.layers.61.mlp.experts.185.up_proj",
"model.layers.61.mlp.experts.185.down_proj",
"model.layers.61.mlp.experts.186.gate_proj",
"model.layers.61.mlp.experts.186.up_proj",
"model.layers.61.mlp.experts.186.down_proj",
"model.layers.61.mlp.experts.187.gate_proj",
"model.layers.61.mlp.experts.187.up_proj",
"model.layers.61.mlp.experts.187.down_proj",
"model.layers.61.mlp.experts.188.gate_proj",
"model.layers.61.mlp.experts.188.up_proj",
"model.layers.61.mlp.experts.188.down_proj",
"model.layers.61.mlp.experts.189.gate_proj",
"model.layers.61.mlp.experts.189.up_proj",
"model.layers.61.mlp.experts.189.down_proj",
"model.layers.61.mlp.experts.190.gate_proj",
"model.layers.61.mlp.experts.190.up_proj",
"model.layers.61.mlp.experts.190.down_proj",
"model.layers.61.mlp.experts.191.gate_proj",
"model.layers.61.mlp.experts.191.up_proj",
"model.layers.61.mlp.experts.191.down_proj",
"model.layers.61.mlp.experts.192.gate_proj",
"model.layers.61.mlp.experts.192.up_proj",
"model.layers.61.mlp.experts.192.down_proj",
"model.layers.61.mlp.experts.193.gate_proj",
"model.layers.61.mlp.experts.193.up_proj",
"model.layers.61.mlp.experts.193.down_proj",
"model.layers.61.mlp.experts.194.gate_proj",
"model.layers.61.mlp.experts.194.up_proj",
"model.layers.61.mlp.experts.194.down_proj",
"model.layers.61.mlp.experts.195.gate_proj",
"model.layers.61.mlp.experts.195.up_proj",
"model.layers.61.mlp.experts.195.down_proj",
"model.layers.61.mlp.experts.196.gate_proj",
"model.layers.61.mlp.experts.196.up_proj",
"model.layers.61.mlp.experts.196.down_proj",
"model.layers.61.mlp.experts.197.gate_proj",
"model.layers.61.mlp.experts.197.up_proj",
"model.layers.61.mlp.experts.197.down_proj",
"model.layers.61.mlp.experts.198.gate_proj",
"model.layers.61.mlp.experts.198.up_proj",
"model.layers.61.mlp.experts.198.down_proj",
"model.layers.61.mlp.experts.199.gate_proj",
"model.layers.61.mlp.experts.199.up_proj",
"model.layers.61.mlp.experts.199.down_proj",
"model.layers.61.mlp.experts.200.gate_proj",
"model.layers.61.mlp.experts.200.up_proj",
"model.layers.61.mlp.experts.200.down_proj",
"model.layers.61.mlp.experts.201.gate_proj",
"model.layers.61.mlp.experts.201.up_proj",
"model.layers.61.mlp.experts.201.down_proj",
"model.layers.61.mlp.experts.202.gate_proj",
"model.layers.61.mlp.experts.202.up_proj",
"model.layers.61.mlp.experts.202.down_proj",
"model.layers.61.mlp.experts.203.gate_proj",
"model.layers.61.mlp.experts.203.up_proj",
"model.layers.61.mlp.experts.203.down_proj",
"model.layers.61.mlp.experts.204.gate_proj",
"model.layers.61.mlp.experts.204.up_proj",
"model.layers.61.mlp.experts.204.down_proj",
"model.layers.61.mlp.experts.205.gate_proj",
"model.layers.61.mlp.experts.205.up_proj",
"model.layers.61.mlp.experts.205.down_proj",
"model.layers.61.mlp.experts.206.gate_proj",
"model.layers.61.mlp.experts.206.up_proj",
"model.layers.61.mlp.experts.206.down_proj",
"model.layers.61.mlp.experts.207.gate_proj",
"model.layers.61.mlp.experts.207.up_proj",
"model.layers.61.mlp.experts.207.down_proj",
"model.layers.61.mlp.experts.208.gate_proj",
"model.layers.61.mlp.experts.208.up_proj",
"model.layers.61.mlp.experts.208.down_proj",
"model.layers.61.mlp.experts.209.gate_proj",
"model.layers.61.mlp.experts.209.up_proj",
"model.layers.61.mlp.experts.209.down_proj",
"model.layers.61.mlp.experts.210.gate_proj",
"model.layers.61.mlp.experts.210.up_proj",
"model.layers.61.mlp.experts.210.down_proj",
"model.layers.61.mlp.experts.211.gate_proj",
"model.layers.61.mlp.experts.211.up_proj",
"model.layers.61.mlp.experts.211.down_proj",
"model.layers.61.mlp.experts.212.gate_proj",
"model.layers.61.mlp.experts.212.up_proj",
"model.layers.61.mlp.experts.212.down_proj",
"model.layers.61.mlp.experts.213.gate_proj",
"model.layers.61.mlp.experts.213.up_proj",
"model.layers.61.mlp.experts.213.down_proj",
"model.layers.61.mlp.experts.214.gate_proj",
"model.layers.61.mlp.experts.214.up_proj",
"model.layers.61.mlp.experts.214.down_proj",
"model.layers.61.mlp.experts.215.gate_proj",
"model.layers.61.mlp.experts.215.up_proj",
"model.layers.61.mlp.experts.215.down_proj",
"model.layers.61.mlp.experts.216.gate_proj",
"model.layers.61.mlp.experts.216.up_proj",
"model.layers.61.mlp.experts.216.down_proj",
"model.layers.61.mlp.experts.217.gate_proj",
"model.layers.61.mlp.experts.217.up_proj",
"model.layers.61.mlp.experts.217.down_proj",
"model.layers.61.mlp.experts.218.gate_proj",
"model.layers.61.mlp.experts.218.up_proj",
"model.layers.61.mlp.experts.218.down_proj",
"model.layers.61.mlp.experts.219.gate_proj",
"model.layers.61.mlp.experts.219.up_proj",
"model.layers.61.mlp.experts.219.down_proj",
"model.layers.61.mlp.experts.220.gate_proj",
"model.layers.61.mlp.experts.220.up_proj",
"model.layers.61.mlp.experts.220.down_proj",
"model.layers.61.mlp.experts.221.gate_proj",
"model.layers.61.mlp.experts.221.up_proj",
"model.layers.61.mlp.experts.221.down_proj",
"model.layers.61.mlp.experts.222.gate_proj",
"model.layers.61.mlp.experts.222.up_proj",
"model.layers.61.mlp.experts.222.down_proj",
"model.layers.61.mlp.experts.223.gate_proj",
"model.layers.61.mlp.experts.223.up_proj",
"model.layers.61.mlp.experts.223.down_proj",
"model.layers.61.mlp.experts.224.gate_proj",
"model.layers.61.mlp.experts.224.up_proj",
"model.layers.61.mlp.experts.224.down_proj",
"model.layers.61.mlp.experts.225.gate_proj",
"model.layers.61.mlp.experts.225.up_proj",
"model.layers.61.mlp.experts.225.down_proj",
"model.layers.61.mlp.experts.226.gate_proj",
"model.layers.61.mlp.experts.226.up_proj",
"model.layers.61.mlp.experts.226.down_proj",
"model.layers.61.mlp.experts.227.gate_proj",
"model.layers.61.mlp.experts.227.up_proj",
"model.layers.61.mlp.experts.227.down_proj",
"model.layers.61.mlp.experts.228.gate_proj",
"model.layers.61.mlp.experts.228.up_proj",
"model.layers.61.mlp.experts.228.down_proj",
"model.layers.61.mlp.experts.229.gate_proj",
"model.layers.61.mlp.experts.229.up_proj",
"model.layers.61.mlp.experts.229.down_proj",
"model.layers.61.mlp.experts.230.gate_proj",
"model.layers.61.mlp.experts.230.up_proj",
"model.layers.61.mlp.experts.230.down_proj",
"model.layers.61.mlp.experts.231.gate_proj",
"model.layers.61.mlp.experts.231.up_proj",
"model.layers.61.mlp.experts.231.down_proj",
"model.layers.61.mlp.experts.232.gate_proj",
"model.layers.61.mlp.experts.232.up_proj",
"model.layers.61.mlp.experts.232.down_proj",
"model.layers.61.mlp.experts.233.gate_proj",
"model.layers.61.mlp.experts.233.up_proj",
"model.layers.61.mlp.experts.233.down_proj",
"model.layers.61.mlp.experts.234.gate_proj",
"model.layers.61.mlp.experts.234.up_proj",
"model.layers.61.mlp.experts.234.down_proj",
"model.layers.61.mlp.experts.235.gate_proj",
"model.layers.61.mlp.experts.235.up_proj",
"model.layers.61.mlp.experts.235.down_proj",
"model.layers.61.mlp.experts.236.gate_proj",
"model.layers.61.mlp.experts.236.up_proj",
"model.layers.61.mlp.experts.236.down_proj",
"model.layers.61.mlp.experts.237.gate_proj",
"model.layers.61.mlp.experts.237.up_proj",
"model.layers.61.mlp.experts.237.down_proj",
"model.layers.61.mlp.experts.238.gate_proj",
"model.layers.61.mlp.experts.238.up_proj",
"model.layers.61.mlp.experts.238.down_proj",
"model.layers.61.mlp.experts.239.gate_proj",
"model.layers.61.mlp.experts.239.up_proj",
"model.layers.61.mlp.experts.239.down_proj",
"model.layers.61.mlp.experts.240.gate_proj",
"model.layers.61.mlp.experts.240.up_proj",
"model.layers.61.mlp.experts.240.down_proj",
"model.layers.61.mlp.experts.241.gate_proj",
"model.layers.61.mlp.experts.241.up_proj",
"model.layers.61.mlp.experts.241.down_proj",
"model.layers.61.mlp.experts.242.gate_proj",
"model.layers.61.mlp.experts.242.up_proj",
"model.layers.61.mlp.experts.242.down_proj",
"model.layers.61.mlp.experts.243.gate_proj",
"model.layers.61.mlp.experts.243.up_proj",
"model.layers.61.mlp.experts.243.down_proj",
"model.layers.61.mlp.experts.244.gate_proj",
"model.layers.61.mlp.experts.244.up_proj",
"model.layers.61.mlp.experts.244.down_proj",
"model.layers.61.mlp.experts.245.gate_proj",
"model.layers.61.mlp.experts.245.up_proj",
"model.layers.61.mlp.experts.245.down_proj",
"model.layers.61.mlp.experts.246.gate_proj",
"model.layers.61.mlp.experts.246.up_proj",
"model.layers.61.mlp.experts.246.down_proj",
"model.layers.61.mlp.experts.247.gate_proj",
"model.layers.61.mlp.experts.247.up_proj",
"model.layers.61.mlp.experts.247.down_proj",
"model.layers.61.mlp.experts.248.gate_proj",
"model.layers.61.mlp.experts.248.up_proj",
"model.layers.61.mlp.experts.248.down_proj",
"model.layers.61.mlp.experts.249.gate_proj",
"model.layers.61.mlp.experts.249.up_proj",
"model.layers.61.mlp.experts.249.down_proj",
"model.layers.61.mlp.experts.250.gate_proj",
"model.layers.61.mlp.experts.250.up_proj",
"model.layers.61.mlp.experts.250.down_proj",
"model.layers.61.mlp.experts.251.gate_proj",
"model.layers.61.mlp.experts.251.up_proj",
"model.layers.61.mlp.experts.251.down_proj",
"model.layers.61.mlp.experts.252.gate_proj",
"model.layers.61.mlp.experts.252.up_proj",
"model.layers.61.mlp.experts.252.down_proj",
"model.layers.61.mlp.experts.253.gate_proj",
"model.layers.61.mlp.experts.253.up_proj",
"model.layers.61.mlp.experts.253.down_proj",
"model.layers.61.mlp.experts.254.gate_proj",
"model.layers.61.mlp.experts.254.up_proj",
"model.layers.61.mlp.experts.254.down_proj",
"model.layers.61.mlp.experts.255.gate_proj",
"model.layers.61.mlp.experts.255.up_proj",
"model.layers.61.mlp.experts.255.down_proj",
"model.layers.61.mlp.shared_experts.gate_proj",
"model.layers.61.mlp.shared_experts.up_proj",
"model.layers.61.mlp.shared_experts.down_proj",
"model.layers.61.shared_head.head",
"lm_head"
],
"export": {
"kv_cache_group": [],
"min_kv_scale": 0.0,
"pack_method": "reorder",
"weight_format": "real_quantized",
"weight_merge_groups": null
},
"global_quant_config": {
"bias": null,
"input_tensors": {
"ch_axis": -1,
"dtype": "fp4",
"group_size": 32,
"is_dynamic": true,
"is_scale_quant": false,
"mx_element_dtype": null,
"observer_cls": "PerBlockMXObserver",
"qscheme": "per_group",
"round_method": "half_even",
"scale_calculation_mode": "even",
"scale_format": "e8m0",
"scale_type": "float",
"symmetric": null
},
"output_tensors": null,
"target_device": null,
"weight": {
"ch_axis": -1,
"dtype": "fp4",
"group_size": 32,
"is_dynamic": false,
"is_scale_quant": false,
"mx_element_dtype": null,
"observer_cls": "PerBlockMXObserver",
"qscheme": "per_group",
"round_method": "half_even",
"scale_calculation_mode": "even",
"scale_format": "e8m0",
"scale_type": "float",
"symmetric": null
}
},
"kv_cache_post_rope": false,
"kv_cache_quant_config": {},
"layer_quant_config": {},
"layer_type_quant_config": {},
"quant_method": "quark",
"quant_mode": "eager_mode",
"softmax_quant_spec": null,
"version": "0.12+03a839631e"
},
"rms_norm_eps": 1e-06,
"rope_scaling": {
"beta_fast": 32,
"beta_slow": 1,
"factor": 40,
"mscale": 1.0,
"mscale_all_dim": 1.0,
"original_max_position_embeddings": 4096,
"type": "yarn"
},
"rope_theta": 10000,
"routed_scaling_factor": 2.5,
"scoring_func": "sigmoid",
"tie_word_embeddings": false,
"topk_group": 4,
"topk_method": "noaux_tc",
"transformers_version": "4.57.1",
"unsloth_fixed": true,
"use_cache": true,
"v_head_dim": 128,
"vocab_size": 129280
}