{ "architectures": [ "Glm4MoeLiteForCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "bos_token_id": 0, "dtype": "bfloat16", "eos_token_id": [ 154820, 154827, 154829 ], "first_k_dense_replace": 1, "head_dim": 64, "hidden_act": "silu", "hidden_size": 2048, "initializer_range": 0.02, "intermediate_size": 10240, "kv_lora_rank": 512, "max_position_embeddings": 202752, "mlp_layer_types": [ "dense", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse", "sparse" ], "model_type": "glm4_moe_lite", "moe_intermediate_size": 1536, "n_group": 1, "n_routed_experts": 64, "n_shared_experts": 1, "norm_topk_prob": true, "num_attention_heads": 20, "num_experts_per_tok": 4, "num_hidden_layers": 47, "num_key_value_heads": 20, "num_nextn_predict_layers": 1, "pad_token_id": 154820, "partial_rotary_factor": 1.0, "pretraining_tp": 1, "q_lora_rank": 768, "qk_head_dim": 256, "qk_nope_head_dim": 192, "qk_rope_head_dim": 64, "quantization_config": { "config_groups": { "group_0": { "format": "float-quantized", "input_activations": { "actorder": null, "block_structure": null, "dynamic": true, "group_size": null, "num_bits": 8, "observer": null, "observer_kwargs": {}, "scale_dtype": null, "strategy": "token", "symmetric": true, "type": "float", "zp_dtype": null }, "output_activations": null, "targets": [ "Linear" ], "weights": { "actorder": null, "block_structure": null, "dynamic": false, "group_size": null, "num_bits": 8, "observer": "minmax", "observer_kwargs": {}, "scale_dtype": null, "strategy": "channel", "symmetric": true, "type": "float", "zp_dtype": null } } }, "format": "float-quantized", "global_compression_ratio": null, "ignore": [ "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.1.mlp.experts.0.gate_proj", "model.layers.1.mlp.experts.0.up_proj", "model.layers.1.mlp.experts.0.down_proj", "model.layers.1.mlp.experts.1.gate_proj", "model.layers.1.mlp.experts.1.up_proj", "model.layers.1.mlp.experts.1.down_proj", "model.layers.1.mlp.experts.2.gate_proj", "model.layers.1.mlp.experts.2.up_proj", "model.layers.1.mlp.experts.2.down_proj", "model.layers.1.mlp.experts.3.gate_proj", "model.layers.1.mlp.experts.3.up_proj", "model.layers.1.mlp.experts.3.down_proj", "model.layers.1.mlp.experts.4.gate_proj", "model.layers.1.mlp.experts.4.up_proj", "model.layers.1.mlp.experts.4.down_proj", "model.layers.1.mlp.experts.5.gate_proj", "model.layers.1.mlp.experts.5.up_proj", "model.layers.1.mlp.experts.5.down_proj", "model.layers.1.mlp.experts.6.gate_proj", "model.layers.1.mlp.experts.6.up_proj", "model.layers.1.mlp.experts.6.down_proj", "model.layers.1.mlp.experts.7.gate_proj", "model.layers.1.mlp.experts.7.up_proj", "model.layers.1.mlp.experts.7.down_proj", "model.layers.1.mlp.experts.8.gate_proj", "model.layers.1.mlp.experts.8.up_proj", "model.layers.1.mlp.experts.8.down_proj", "model.layers.1.mlp.experts.9.gate_proj", "model.layers.1.mlp.experts.9.up_proj", "model.layers.1.mlp.experts.9.down_proj", "model.layers.1.mlp.experts.10.gate_proj", "model.layers.1.mlp.experts.10.up_proj", "model.layers.1.mlp.experts.10.down_proj", "model.layers.1.mlp.experts.11.gate_proj", "model.layers.1.mlp.experts.11.up_proj", "model.layers.1.mlp.experts.11.down_proj", "model.layers.1.mlp.experts.12.gate_proj", "model.layers.1.mlp.experts.12.up_proj", "model.layers.1.mlp.experts.12.down_proj", "model.layers.1.mlp.experts.13.gate_proj", "model.layers.1.mlp.experts.13.up_proj", "model.layers.1.mlp.experts.13.down_proj", "model.layers.1.mlp.experts.14.gate_proj", "model.layers.1.mlp.experts.14.up_proj", "model.layers.1.mlp.experts.14.down_proj", "model.layers.1.mlp.experts.15.gate_proj", "model.layers.1.mlp.experts.15.up_proj", "model.layers.1.mlp.experts.15.down_proj", "model.layers.1.mlp.experts.16.gate_proj", "model.layers.1.mlp.experts.16.up_proj", "model.layers.1.mlp.experts.16.down_proj", "model.layers.1.mlp.experts.17.gate_proj", "model.layers.1.mlp.experts.17.up_proj", "model.layers.1.mlp.experts.17.down_proj", "model.layers.1.mlp.experts.18.gate_proj", "model.layers.1.mlp.experts.18.up_proj", "model.layers.1.mlp.experts.18.down_proj", "model.layers.1.mlp.experts.19.gate_proj", "model.layers.1.mlp.experts.19.up_proj", "model.layers.1.mlp.experts.19.down_proj", "model.layers.1.mlp.experts.20.gate_proj", "model.layers.1.mlp.experts.20.up_proj", "model.layers.1.mlp.experts.20.down_proj", "model.layers.1.mlp.experts.21.gate_proj", "model.layers.1.mlp.experts.21.up_proj", "model.layers.1.mlp.experts.21.down_proj", "model.layers.1.mlp.experts.22.gate_proj", "model.layers.1.mlp.experts.22.up_proj", "model.layers.1.mlp.experts.22.down_proj", "model.layers.1.mlp.experts.23.gate_proj", "model.layers.1.mlp.experts.23.up_proj", "model.layers.1.mlp.experts.23.down_proj", "model.layers.1.mlp.experts.24.gate_proj", "model.layers.1.mlp.experts.24.up_proj", "model.layers.1.mlp.experts.24.down_proj", "model.layers.1.mlp.experts.25.gate_proj", "model.layers.1.mlp.experts.25.up_proj", "model.layers.1.mlp.experts.25.down_proj", "model.layers.1.mlp.experts.26.gate_proj", "model.layers.1.mlp.experts.26.up_proj", "model.layers.1.mlp.experts.26.down_proj", "model.layers.1.mlp.experts.27.gate_proj", "model.layers.1.mlp.experts.27.up_proj", "model.layers.1.mlp.experts.27.down_proj", "model.layers.1.mlp.experts.28.gate_proj", "model.layers.1.mlp.experts.28.up_proj", "model.layers.1.mlp.experts.28.down_proj", "model.layers.1.mlp.experts.29.gate_proj", "model.layers.1.mlp.experts.29.up_proj", "model.layers.1.mlp.experts.29.down_proj", "model.layers.1.mlp.experts.30.gate_proj", "model.layers.1.mlp.experts.30.up_proj", "model.layers.1.mlp.experts.30.down_proj", "model.layers.1.mlp.experts.31.gate_proj", "model.layers.1.mlp.experts.31.up_proj", "model.layers.1.mlp.experts.31.down_proj", "model.layers.1.mlp.experts.32.gate_proj", "model.layers.1.mlp.experts.32.up_proj", "model.layers.1.mlp.experts.32.down_proj", "model.layers.1.mlp.experts.33.gate_proj", "model.layers.1.mlp.experts.33.up_proj", "model.layers.1.mlp.experts.33.down_proj", "model.layers.1.mlp.experts.34.gate_proj", "model.layers.1.mlp.experts.34.up_proj", "model.layers.1.mlp.experts.34.down_proj", "model.layers.1.mlp.experts.35.gate_proj", "model.layers.1.mlp.experts.35.up_proj", "model.layers.1.mlp.experts.35.down_proj", "model.layers.1.mlp.experts.36.gate_proj", "model.layers.1.mlp.experts.36.up_proj", "model.layers.1.mlp.experts.36.down_proj", "model.layers.1.mlp.experts.37.gate_proj", "model.layers.1.mlp.experts.37.up_proj", "model.layers.1.mlp.experts.37.down_proj", "model.layers.1.mlp.experts.38.gate_proj", "model.layers.1.mlp.experts.38.up_proj", "model.layers.1.mlp.experts.38.down_proj", "model.layers.1.mlp.experts.39.gate_proj", "model.layers.1.mlp.experts.39.up_proj", "model.layers.1.mlp.experts.39.down_proj", "model.layers.1.mlp.experts.40.gate_proj", "model.layers.1.mlp.experts.40.up_proj", "model.layers.1.mlp.experts.40.down_proj", "model.layers.1.mlp.experts.41.gate_proj", "model.layers.1.mlp.experts.41.up_proj", "model.layers.1.mlp.experts.41.down_proj", "model.layers.1.mlp.experts.42.gate_proj", "model.layers.1.mlp.experts.42.up_proj", "model.layers.1.mlp.experts.42.down_proj", "model.layers.1.mlp.experts.43.gate_proj", "model.layers.1.mlp.experts.43.up_proj", "model.layers.1.mlp.experts.43.down_proj", "model.layers.1.mlp.experts.44.gate_proj", "model.layers.1.mlp.experts.44.up_proj", "model.layers.1.mlp.experts.44.down_proj", "model.layers.1.mlp.experts.45.gate_proj", "model.layers.1.mlp.experts.45.up_proj", "model.layers.1.mlp.experts.45.down_proj", "model.layers.1.mlp.experts.46.gate_proj", "model.layers.1.mlp.experts.46.up_proj", "model.layers.1.mlp.experts.46.down_proj", "model.layers.1.mlp.experts.47.gate_proj", "model.layers.1.mlp.experts.47.up_proj", "model.layers.1.mlp.experts.47.down_proj", "model.layers.1.mlp.experts.48.gate_proj", "model.layers.1.mlp.experts.48.up_proj", "model.layers.1.mlp.experts.48.down_proj", "model.layers.1.mlp.experts.49.gate_proj", "model.layers.1.mlp.experts.49.up_proj", "model.layers.1.mlp.experts.49.down_proj", "model.layers.1.mlp.experts.50.gate_proj", "model.layers.1.mlp.experts.50.up_proj", "model.layers.1.mlp.experts.50.down_proj", "model.layers.1.mlp.experts.51.gate_proj", "model.layers.1.mlp.experts.51.up_proj", "model.layers.1.mlp.experts.51.down_proj", "model.layers.1.mlp.experts.52.gate_proj", "model.layers.1.mlp.experts.52.up_proj", "model.layers.1.mlp.experts.52.down_proj", "model.layers.1.mlp.experts.53.gate_proj", "model.layers.1.mlp.experts.53.up_proj", "model.layers.1.mlp.experts.53.down_proj", "model.layers.1.mlp.experts.54.gate_proj", "model.layers.1.mlp.experts.54.up_proj", "model.layers.1.mlp.experts.54.down_proj", "model.layers.1.mlp.experts.55.gate_proj", "model.layers.1.mlp.experts.55.up_proj", "model.layers.1.mlp.experts.55.down_proj", "model.layers.1.mlp.experts.56.gate_proj", "model.layers.1.mlp.experts.56.up_proj", "model.layers.1.mlp.experts.56.down_proj", "model.layers.1.mlp.experts.57.gate_proj", "model.layers.1.mlp.experts.57.up_proj", "model.layers.1.mlp.experts.57.down_proj", "model.layers.1.mlp.experts.58.gate_proj", "model.layers.1.mlp.experts.58.up_proj", "model.layers.1.mlp.experts.58.down_proj", "model.layers.1.mlp.experts.59.gate_proj", "model.layers.1.mlp.experts.59.up_proj", "model.layers.1.mlp.experts.59.down_proj", "model.layers.1.mlp.experts.60.gate_proj", "model.layers.1.mlp.experts.60.up_proj", "model.layers.1.mlp.experts.60.down_proj", "model.layers.1.mlp.experts.61.gate_proj", "model.layers.1.mlp.experts.61.up_proj", "model.layers.1.mlp.experts.61.down_proj", "model.layers.1.mlp.experts.62.gate_proj", "model.layers.1.mlp.experts.62.up_proj", "model.layers.1.mlp.experts.62.down_proj", "model.layers.1.mlp.experts.63.gate_proj", "model.layers.1.mlp.experts.63.up_proj", "model.layers.1.mlp.experts.63.down_proj", "model.layers.1.mlp.shared_experts.gate_proj", "model.layers.1.mlp.shared_experts.up_proj", "model.layers.1.mlp.shared_experts.down_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.39.mlp.experts.0.gate_proj", "model.layers.39.mlp.experts.0.up_proj", "model.layers.39.mlp.experts.0.down_proj", "model.layers.39.mlp.experts.1.gate_proj", "model.layers.39.mlp.experts.1.up_proj", "model.layers.39.mlp.experts.1.down_proj", "model.layers.39.mlp.experts.2.gate_proj", "model.layers.39.mlp.experts.2.up_proj", "model.layers.39.mlp.experts.2.down_proj", "model.layers.39.mlp.experts.3.gate_proj", "model.layers.39.mlp.experts.3.up_proj", "model.layers.39.mlp.experts.3.down_proj", "model.layers.39.mlp.experts.4.gate_proj", "model.layers.39.mlp.experts.4.up_proj", "model.layers.39.mlp.experts.4.down_proj", "model.layers.39.mlp.experts.5.gate_proj", "model.layers.39.mlp.experts.5.up_proj", "model.layers.39.mlp.experts.5.down_proj", "model.layers.39.mlp.experts.6.gate_proj", "model.layers.39.mlp.experts.6.up_proj", "model.layers.39.mlp.experts.6.down_proj", "model.layers.39.mlp.experts.7.gate_proj", "model.layers.39.mlp.experts.7.up_proj", "model.layers.39.mlp.experts.7.down_proj", "model.layers.39.mlp.experts.8.gate_proj", "model.layers.39.mlp.experts.8.up_proj", "model.layers.39.mlp.experts.8.down_proj", "model.layers.39.mlp.experts.9.gate_proj", "model.layers.39.mlp.experts.9.up_proj", "model.layers.39.mlp.experts.9.down_proj", "model.layers.39.mlp.experts.10.gate_proj", "model.layers.39.mlp.experts.10.up_proj", "model.layers.39.mlp.experts.10.down_proj", "model.layers.39.mlp.experts.11.gate_proj", "model.layers.39.mlp.experts.11.up_proj", "model.layers.39.mlp.experts.11.down_proj", "model.layers.39.mlp.experts.12.gate_proj", "model.layers.39.mlp.experts.12.up_proj", "model.layers.39.mlp.experts.12.down_proj", "model.layers.39.mlp.experts.13.gate_proj", "model.layers.39.mlp.experts.13.up_proj", "model.layers.39.mlp.experts.13.down_proj", "model.layers.39.mlp.experts.14.gate_proj", "model.layers.39.mlp.experts.14.up_proj", "model.layers.39.mlp.experts.14.down_proj", "model.layers.39.mlp.experts.15.gate_proj", "model.layers.39.mlp.experts.15.up_proj", "model.layers.39.mlp.experts.15.down_proj", "model.layers.39.mlp.experts.16.gate_proj", "model.layers.39.mlp.experts.16.up_proj", "model.layers.39.mlp.experts.16.down_proj", "model.layers.39.mlp.experts.17.gate_proj", "model.layers.39.mlp.experts.17.up_proj", "model.layers.39.mlp.experts.17.down_proj", "model.layers.39.mlp.experts.18.gate_proj", "model.layers.39.mlp.experts.18.up_proj", "model.layers.39.mlp.experts.18.down_proj", "model.layers.39.mlp.experts.19.gate_proj", "model.layers.39.mlp.experts.19.up_proj", "model.layers.39.mlp.experts.19.down_proj", "model.layers.39.mlp.experts.20.gate_proj", "model.layers.39.mlp.experts.20.up_proj", "model.layers.39.mlp.experts.20.down_proj", "model.layers.39.mlp.experts.21.gate_proj", "model.layers.39.mlp.experts.21.up_proj", "model.layers.39.mlp.experts.21.down_proj", "model.layers.39.mlp.experts.22.gate_proj", "model.layers.39.mlp.experts.22.up_proj", "model.layers.39.mlp.experts.22.down_proj", "model.layers.39.mlp.experts.23.gate_proj", "model.layers.39.mlp.experts.23.up_proj", "model.layers.39.mlp.experts.23.down_proj", "model.layers.39.mlp.experts.24.gate_proj", "model.layers.39.mlp.experts.24.up_proj", "model.layers.39.mlp.experts.24.down_proj", "model.layers.39.mlp.experts.25.gate_proj", "model.layers.39.mlp.experts.25.up_proj", "model.layers.39.mlp.experts.25.down_proj", "model.layers.39.mlp.experts.26.gate_proj", "model.layers.39.mlp.experts.26.up_proj", "model.layers.39.mlp.experts.26.down_proj", "model.layers.39.mlp.experts.27.gate_proj", "model.layers.39.mlp.experts.27.up_proj", "model.layers.39.mlp.experts.27.down_proj", "model.layers.39.mlp.experts.28.gate_proj", "model.layers.39.mlp.experts.28.up_proj", "model.layers.39.mlp.experts.28.down_proj", "model.layers.39.mlp.experts.29.gate_proj", "model.layers.39.mlp.experts.29.up_proj", "model.layers.39.mlp.experts.29.down_proj", "model.layers.39.mlp.experts.30.gate_proj", "model.layers.39.mlp.experts.30.up_proj", "model.layers.39.mlp.experts.30.down_proj", "model.layers.39.mlp.experts.31.gate_proj", "model.layers.39.mlp.experts.31.up_proj", "model.layers.39.mlp.experts.31.down_proj", "model.layers.39.mlp.experts.32.gate_proj", "model.layers.39.mlp.experts.32.up_proj", "model.layers.39.mlp.experts.32.down_proj", "model.layers.39.mlp.experts.33.gate_proj", "model.layers.39.mlp.experts.33.up_proj", "model.layers.39.mlp.experts.33.down_proj", "model.layers.39.mlp.experts.34.gate_proj", "model.layers.39.mlp.experts.34.up_proj", "model.layers.39.mlp.experts.34.down_proj", "model.layers.39.mlp.experts.35.gate_proj", "model.layers.39.mlp.experts.35.up_proj", "model.layers.39.mlp.experts.35.down_proj", "model.layers.39.mlp.experts.36.gate_proj", "model.layers.39.mlp.experts.36.up_proj", "model.layers.39.mlp.experts.36.down_proj", "model.layers.39.mlp.experts.37.gate_proj", "model.layers.39.mlp.experts.37.up_proj", "model.layers.39.mlp.experts.37.down_proj", "model.layers.39.mlp.experts.38.gate_proj", "model.layers.39.mlp.experts.38.up_proj", "model.layers.39.mlp.experts.38.down_proj", "model.layers.39.mlp.experts.39.gate_proj", "model.layers.39.mlp.experts.39.up_proj", "model.layers.39.mlp.experts.39.down_proj", "model.layers.39.mlp.experts.40.gate_proj", "model.layers.39.mlp.experts.40.up_proj", "model.layers.39.mlp.experts.40.down_proj", "model.layers.39.mlp.experts.41.gate_proj", "model.layers.39.mlp.experts.41.up_proj", "model.layers.39.mlp.experts.41.down_proj", "model.layers.39.mlp.experts.42.gate_proj", "model.layers.39.mlp.experts.42.up_proj", "model.layers.39.mlp.experts.42.down_proj", "model.layers.39.mlp.experts.43.gate_proj", "model.layers.39.mlp.experts.43.up_proj", "model.layers.39.mlp.experts.43.down_proj", "model.layers.39.mlp.experts.44.gate_proj", "model.layers.39.mlp.experts.44.up_proj", "model.layers.39.mlp.experts.44.down_proj", "model.layers.39.mlp.experts.45.gate_proj", "model.layers.39.mlp.experts.45.up_proj", "model.layers.39.mlp.experts.45.down_proj", "model.layers.39.mlp.experts.46.gate_proj", "model.layers.39.mlp.experts.46.up_proj", "model.layers.39.mlp.experts.46.down_proj", "model.layers.39.mlp.experts.47.gate_proj", "model.layers.39.mlp.experts.47.up_proj", "model.layers.39.mlp.experts.47.down_proj", "model.layers.39.mlp.experts.48.gate_proj", "model.layers.39.mlp.experts.48.up_proj", "model.layers.39.mlp.experts.48.down_proj", "model.layers.39.mlp.experts.49.gate_proj", "model.layers.39.mlp.experts.49.up_proj", "model.layers.39.mlp.experts.49.down_proj", "model.layers.39.mlp.experts.50.gate_proj", "model.layers.39.mlp.experts.50.up_proj", "model.layers.39.mlp.experts.50.down_proj", "model.layers.39.mlp.experts.51.gate_proj", "model.layers.39.mlp.experts.51.up_proj", "model.layers.39.mlp.experts.51.down_proj", "model.layers.39.mlp.experts.52.gate_proj", "model.layers.39.mlp.experts.52.up_proj", "model.layers.39.mlp.experts.52.down_proj", "model.layers.39.mlp.experts.53.gate_proj", "model.layers.39.mlp.experts.53.up_proj", "model.layers.39.mlp.experts.53.down_proj", "model.layers.39.mlp.experts.54.gate_proj", "model.layers.39.mlp.experts.54.up_proj", "model.layers.39.mlp.experts.54.down_proj", "model.layers.39.mlp.experts.55.gate_proj", "model.layers.39.mlp.experts.55.up_proj", "model.layers.39.mlp.experts.55.down_proj", "model.layers.39.mlp.experts.56.gate_proj", "model.layers.39.mlp.experts.56.up_proj", "model.layers.39.mlp.experts.56.down_proj", "model.layers.39.mlp.experts.57.gate_proj", "model.layers.39.mlp.experts.57.up_proj", "model.layers.39.mlp.experts.57.down_proj", "model.layers.39.mlp.experts.58.gate_proj", "model.layers.39.mlp.experts.58.up_proj", "model.layers.39.mlp.experts.58.down_proj", "model.layers.39.mlp.experts.59.gate_proj", "model.layers.39.mlp.experts.59.up_proj", "model.layers.39.mlp.experts.59.down_proj", "model.layers.39.mlp.experts.60.gate_proj", "model.layers.39.mlp.experts.60.up_proj", "model.layers.39.mlp.experts.60.down_proj", "model.layers.39.mlp.experts.61.gate_proj", "model.layers.39.mlp.experts.61.up_proj", "model.layers.39.mlp.experts.61.down_proj", "model.layers.39.mlp.experts.62.gate_proj", "model.layers.39.mlp.experts.62.up_proj", "model.layers.39.mlp.experts.62.down_proj", "model.layers.39.mlp.experts.63.gate_proj", "model.layers.39.mlp.experts.63.up_proj", "model.layers.39.mlp.experts.63.down_proj", "model.layers.39.mlp.shared_experts.gate_proj", "model.layers.39.mlp.shared_experts.up_proj", "model.layers.39.mlp.shared_experts.down_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.46.mlp.experts.0.gate_proj", "model.layers.46.mlp.experts.0.up_proj", "model.layers.46.mlp.experts.0.down_proj", "model.layers.46.mlp.experts.1.gate_proj", "model.layers.46.mlp.experts.1.up_proj", "model.layers.46.mlp.experts.1.down_proj", "model.layers.46.mlp.experts.2.gate_proj", "model.layers.46.mlp.experts.2.up_proj", "model.layers.46.mlp.experts.2.down_proj", "model.layers.46.mlp.experts.3.gate_proj", "model.layers.46.mlp.experts.3.up_proj", "model.layers.46.mlp.experts.3.down_proj", "model.layers.46.mlp.experts.4.gate_proj", "model.layers.46.mlp.experts.4.up_proj", "model.layers.46.mlp.experts.4.down_proj", "model.layers.46.mlp.experts.5.gate_proj", "model.layers.46.mlp.experts.5.up_proj", "model.layers.46.mlp.experts.5.down_proj", "model.layers.46.mlp.experts.6.gate_proj", "model.layers.46.mlp.experts.6.up_proj", "model.layers.46.mlp.experts.6.down_proj", "model.layers.46.mlp.experts.7.gate_proj", "model.layers.46.mlp.experts.7.up_proj", "model.layers.46.mlp.experts.7.down_proj", "model.layers.46.mlp.experts.8.gate_proj", "model.layers.46.mlp.experts.8.up_proj", "model.layers.46.mlp.experts.8.down_proj", "model.layers.46.mlp.experts.9.gate_proj", "model.layers.46.mlp.experts.9.up_proj", "model.layers.46.mlp.experts.9.down_proj", "model.layers.46.mlp.experts.10.gate_proj", "model.layers.46.mlp.experts.10.up_proj", "model.layers.46.mlp.experts.10.down_proj", "model.layers.46.mlp.experts.11.gate_proj", "model.layers.46.mlp.experts.11.up_proj", "model.layers.46.mlp.experts.11.down_proj", "model.layers.46.mlp.experts.12.gate_proj", "model.layers.46.mlp.experts.12.up_proj", "model.layers.46.mlp.experts.12.down_proj", "model.layers.46.mlp.experts.13.gate_proj", "model.layers.46.mlp.experts.13.up_proj", "model.layers.46.mlp.experts.13.down_proj", "model.layers.46.mlp.experts.14.gate_proj", "model.layers.46.mlp.experts.14.up_proj", "model.layers.46.mlp.experts.14.down_proj", "model.layers.46.mlp.experts.15.gate_proj", "model.layers.46.mlp.experts.15.up_proj", "model.layers.46.mlp.experts.15.down_proj", "model.layers.46.mlp.experts.16.gate_proj", "model.layers.46.mlp.experts.16.up_proj", "model.layers.46.mlp.experts.16.down_proj", "model.layers.46.mlp.experts.17.gate_proj", "model.layers.46.mlp.experts.17.up_proj", "model.layers.46.mlp.experts.17.down_proj", "model.layers.46.mlp.experts.18.gate_proj", "model.layers.46.mlp.experts.18.up_proj", "model.layers.46.mlp.experts.18.down_proj", "model.layers.46.mlp.experts.19.gate_proj", "model.layers.46.mlp.experts.19.up_proj", "model.layers.46.mlp.experts.19.down_proj", "model.layers.46.mlp.experts.20.gate_proj", "model.layers.46.mlp.experts.20.up_proj", "model.layers.46.mlp.experts.20.down_proj", "model.layers.46.mlp.experts.21.gate_proj", "model.layers.46.mlp.experts.21.up_proj", "model.layers.46.mlp.experts.21.down_proj", "model.layers.46.mlp.experts.22.gate_proj", "model.layers.46.mlp.experts.22.up_proj", "model.layers.46.mlp.experts.22.down_proj", "model.layers.46.mlp.experts.23.gate_proj", "model.layers.46.mlp.experts.23.up_proj", "model.layers.46.mlp.experts.23.down_proj", "model.layers.46.mlp.experts.24.gate_proj", "model.layers.46.mlp.experts.24.up_proj", "model.layers.46.mlp.experts.24.down_proj", "model.layers.46.mlp.experts.25.gate_proj", "model.layers.46.mlp.experts.25.up_proj", "model.layers.46.mlp.experts.25.down_proj", "model.layers.46.mlp.experts.26.gate_proj", "model.layers.46.mlp.experts.26.up_proj", "model.layers.46.mlp.experts.26.down_proj", "model.layers.46.mlp.experts.27.gate_proj", "model.layers.46.mlp.experts.27.up_proj", "model.layers.46.mlp.experts.27.down_proj", "model.layers.46.mlp.experts.28.gate_proj", "model.layers.46.mlp.experts.28.up_proj", "model.layers.46.mlp.experts.28.down_proj", "model.layers.46.mlp.experts.29.gate_proj", "model.layers.46.mlp.experts.29.up_proj", "model.layers.46.mlp.experts.29.down_proj", "model.layers.46.mlp.experts.30.gate_proj", "model.layers.46.mlp.experts.30.up_proj", "model.layers.46.mlp.experts.30.down_proj", "model.layers.46.mlp.experts.31.gate_proj", "model.layers.46.mlp.experts.31.up_proj", "model.layers.46.mlp.experts.31.down_proj", "model.layers.46.mlp.experts.32.gate_proj", "model.layers.46.mlp.experts.32.up_proj", "model.layers.46.mlp.experts.32.down_proj", "model.layers.46.mlp.experts.33.gate_proj", "model.layers.46.mlp.experts.33.up_proj", "model.layers.46.mlp.experts.33.down_proj", "model.layers.46.mlp.experts.34.gate_proj", "model.layers.46.mlp.experts.34.up_proj", "model.layers.46.mlp.experts.34.down_proj", "model.layers.46.mlp.experts.35.gate_proj", "model.layers.46.mlp.experts.35.up_proj", "model.layers.46.mlp.experts.35.down_proj", "model.layers.46.mlp.experts.36.gate_proj", "model.layers.46.mlp.experts.36.up_proj", "model.layers.46.mlp.experts.36.down_proj", "model.layers.46.mlp.experts.37.gate_proj", "model.layers.46.mlp.experts.37.up_proj", "model.layers.46.mlp.experts.37.down_proj", "model.layers.46.mlp.experts.38.gate_proj", "model.layers.46.mlp.experts.38.up_proj", "model.layers.46.mlp.experts.38.down_proj", "model.layers.46.mlp.experts.39.gate_proj", "model.layers.46.mlp.experts.39.up_proj", "model.layers.46.mlp.experts.39.down_proj", "model.layers.46.mlp.experts.40.gate_proj", "model.layers.46.mlp.experts.40.up_proj", "model.layers.46.mlp.experts.40.down_proj", "model.layers.46.mlp.experts.41.gate_proj", "model.layers.46.mlp.experts.41.up_proj", "model.layers.46.mlp.experts.41.down_proj", "model.layers.46.mlp.experts.42.gate_proj", "model.layers.46.mlp.experts.42.up_proj", "model.layers.46.mlp.experts.42.down_proj", "model.layers.46.mlp.experts.43.gate_proj", "model.layers.46.mlp.experts.43.up_proj", "model.layers.46.mlp.experts.43.down_proj", "model.layers.46.mlp.experts.44.gate_proj", "model.layers.46.mlp.experts.44.up_proj", "model.layers.46.mlp.experts.44.down_proj", "model.layers.46.mlp.experts.45.gate_proj", "model.layers.46.mlp.experts.45.up_proj", "model.layers.46.mlp.experts.45.down_proj", "model.layers.46.mlp.experts.46.gate_proj", "model.layers.46.mlp.experts.46.up_proj", "model.layers.46.mlp.experts.46.down_proj", "model.layers.46.mlp.experts.47.gate_proj", "model.layers.46.mlp.experts.47.up_proj", "model.layers.46.mlp.experts.47.down_proj", "model.layers.46.mlp.experts.48.gate_proj", "model.layers.46.mlp.experts.48.up_proj", "model.layers.46.mlp.experts.48.down_proj", "model.layers.46.mlp.experts.49.gate_proj", "model.layers.46.mlp.experts.49.up_proj", "model.layers.46.mlp.experts.49.down_proj", "model.layers.46.mlp.experts.50.gate_proj", "model.layers.46.mlp.experts.50.up_proj", "model.layers.46.mlp.experts.50.down_proj", "model.layers.46.mlp.experts.51.gate_proj", "model.layers.46.mlp.experts.51.up_proj", "model.layers.46.mlp.experts.51.down_proj", "model.layers.46.mlp.experts.52.gate_proj", "model.layers.46.mlp.experts.52.up_proj", "model.layers.46.mlp.experts.52.down_proj", "model.layers.46.mlp.experts.53.gate_proj", "model.layers.46.mlp.experts.53.up_proj", "model.layers.46.mlp.experts.53.down_proj", "model.layers.46.mlp.experts.54.gate_proj", "model.layers.46.mlp.experts.54.up_proj", "model.layers.46.mlp.experts.54.down_proj", "model.layers.46.mlp.experts.55.gate_proj", "model.layers.46.mlp.experts.55.up_proj", "model.layers.46.mlp.experts.55.down_proj", "model.layers.46.mlp.experts.56.gate_proj", "model.layers.46.mlp.experts.56.up_proj", "model.layers.46.mlp.experts.56.down_proj", "model.layers.46.mlp.experts.57.gate_proj", "model.layers.46.mlp.experts.57.up_proj", "model.layers.46.mlp.experts.57.down_proj", "model.layers.46.mlp.experts.58.gate_proj", "model.layers.46.mlp.experts.58.up_proj", "model.layers.46.mlp.experts.58.down_proj", "model.layers.46.mlp.experts.59.gate_proj", "model.layers.46.mlp.experts.59.up_proj", "model.layers.46.mlp.experts.59.down_proj", "model.layers.46.mlp.experts.60.gate_proj", "model.layers.46.mlp.experts.60.up_proj", "model.layers.46.mlp.experts.60.down_proj", "model.layers.46.mlp.experts.61.gate_proj", "model.layers.46.mlp.experts.61.up_proj", "model.layers.46.mlp.experts.61.down_proj", "model.layers.46.mlp.experts.62.gate_proj", "model.layers.46.mlp.experts.62.up_proj", "model.layers.46.mlp.experts.62.down_proj", "model.layers.46.mlp.experts.63.gate_proj", "model.layers.46.mlp.experts.63.up_proj", "model.layers.46.mlp.experts.63.down_proj", "model.layers.46.mlp.shared_experts.gate_proj", "model.layers.46.mlp.shared_experts.up_proj", "model.layers.46.mlp.shared_experts.down_proj", "lm_head" ], "kv_cache_scheme": null, "quant_method": "compressed-tensors", "quantization_status": "compressed", "sparsity_config": {}, "transform_config": {}, "version": "0.13.1.a20260123" }, "rms_norm_eps": 1e-05, "rope_interleave": true, "rope_parameters": { "partial_rotary_factor": 1.0, "rope_theta": 1000000, "rope_type": "default" }, "routed_scaling_factor": 1.8, "scoring_func": "sigmoid", "tie_word_embeddings": false, "topk_group": 1, "topk_method": "noaux_tc", "transformers_version": "5.0.0rc3", "use_cache": true, "v_head_dim": 256, "vocab_size": 154880 }