{ "format_version": 1, "source": { "m-a-p/YuE2-3B": "29b3558dd46954a0cd9021dc76d5c91864a0f1c7", "m-a-p/YuE2-Vae": "9a94e1d0ea9f8087e98f77fa88df4a4068104d2a", "orbitquant": "cd58b4ecf77f22b8c4116b3d0b7d4af258e16ba3" }, "config": { "quant_method": "orbitquant", "weight_bits": 4, "activation_bits": 4, "rotation": "rpbh", "rotation_seed": 0, "block_size": "paper", "codebook": "lloyd_max", "codebook_version": 2, "codebook_dtype": "float32", "row_norm_dtype": "bfloat16", "activation_norm_dtype": "float32", "activation_eps": 1e-10, "weight_pack_dtype": "uint8", "target_policy": "universal", "adaln_policy": "int4_rtn", "adaln_group_size": 64, "modules_to_convert": [], "modules_to_use_adaln": [], "modules_to_not_convert": [], "modules_dtype_dict": {}, "artifact_format_version": 1, "runtime_mode": "auto_fused", "activation_kernel_backend": "auto", "packed_matmul_block_m": 64, "packed_matmul_block_n": 64, "packed_matmul_block_k": 128, "packed_matmul_num_warps": 4, "weight_row_tile_size": 256, "w4a4_int8_weight_cache": false, "lowbit_boundary_protection": "auto", "lowbit_protected_blocks": 4, "lowbit_protected_bits": 4, "lowbit_interior_protection": "auto" }, "modules": [ { "name": "model.layers.0.self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.0.self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.0.self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.0.self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.0.nar_self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.0.nar_self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.0.nar_self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.0.nar_self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.0.mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.0.mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.0.mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.0.nar_mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.0.nar_mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.0.nar_mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.1.self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.1.self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.1.self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.1.self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.1.nar_self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.1.nar_self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.1.nar_self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.1.nar_self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.1.mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.1.mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.1.mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.1.nar_mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.1.nar_mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.1.nar_mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.2.self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.2.self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.2.self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.2.self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.2.nar_self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.2.nar_self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.2.nar_self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.2.nar_self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.2.mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.2.mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.2.mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.2.nar_mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.2.nar_mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.2.nar_mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.3.self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.3.self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.3.self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.3.self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.3.nar_self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.3.nar_self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.3.nar_self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.3.nar_self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.3.mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.3.mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.3.mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.3.nar_mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.3.nar_mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.3.nar_mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.4.self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.4.self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.4.self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.4.self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.4.nar_self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.4.nar_self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.4.nar_self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.4.nar_self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.4.mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.4.mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.4.mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.4.nar_mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.4.nar_mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.4.nar_mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.5.self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.5.self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.5.self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.5.self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.5.nar_self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.5.nar_self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.5.nar_self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.5.nar_self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.5.mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.5.mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.5.mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.5.nar_mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.5.nar_mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.5.nar_mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.6.self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.6.self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.6.self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.6.self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.6.nar_self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.6.nar_self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.6.nar_self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.6.nar_self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.6.mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.6.mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.6.mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.6.nar_mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.6.nar_mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.6.nar_mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.7.self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.7.self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.7.self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.7.self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.7.nar_self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.7.nar_self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.7.nar_self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.7.nar_self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.7.mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.7.mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.7.mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.7.nar_mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.7.nar_mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.7.nar_mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.8.self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.8.self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.8.self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.8.self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.8.nar_self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.8.nar_self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.8.nar_self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.8.nar_self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.8.mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.8.mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.8.mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.8.nar_mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.8.nar_mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.8.nar_mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.9.self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.9.self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.9.self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.9.self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.9.nar_self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.9.nar_self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.9.nar_self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.9.nar_self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.9.mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.9.mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.9.mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.9.nar_mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.9.nar_mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.9.nar_mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.10.self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.10.self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.10.self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.10.self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.10.nar_self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.10.nar_self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.10.nar_self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.10.nar_self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.10.mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.10.mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.10.mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.10.nar_mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.10.nar_mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.10.nar_mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.11.self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.11.self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.11.self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.11.self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.11.nar_self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.11.nar_self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.11.nar_self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.11.nar_self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.11.mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.11.mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.11.mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.11.nar_mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.11.nar_mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.11.nar_mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.12.self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.12.self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.12.self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.12.self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.12.nar_self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.12.nar_self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.12.nar_self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.12.nar_self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.12.mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.12.mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.12.mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.12.nar_mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.12.nar_mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.12.nar_mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.13.self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.13.self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.13.self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.13.self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.13.nar_self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.13.nar_self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.13.nar_self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.13.nar_self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.13.mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.13.mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.13.mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.13.nar_mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.13.nar_mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.13.nar_mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.14.self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.14.self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.14.self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.14.self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.14.nar_self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.14.nar_self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.14.nar_self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.14.nar_self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.14.mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.14.mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.14.mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.14.nar_mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.14.nar_mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.14.nar_mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.15.self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.15.self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.15.self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.15.self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.15.nar_self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.15.nar_self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.15.nar_self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.15.nar_self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.15.mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.15.mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.15.mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.15.nar_mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.15.nar_mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.15.nar_mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.16.self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.16.self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.16.self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.16.self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.16.nar_self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.16.nar_self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.16.nar_self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.16.nar_self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.16.mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.16.mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.16.mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.16.nar_mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.16.nar_mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.16.nar_mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.17.self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.17.self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.17.self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.17.self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.17.nar_self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.17.nar_self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.17.nar_self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.17.nar_self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.17.mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.17.mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.17.mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.17.nar_mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.17.nar_mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.17.nar_mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.18.self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.18.self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.18.self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.18.self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.18.nar_self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.18.nar_self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.18.nar_self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.18.nar_self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.18.mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.18.mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.18.mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.18.nar_mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.18.nar_mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.18.nar_mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.19.self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.19.self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.19.self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.19.self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.19.nar_self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.19.nar_self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.19.nar_self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.19.nar_self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.19.mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.19.mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.19.mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.19.nar_mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.19.nar_mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.19.nar_mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.20.self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.20.self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.20.self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.20.self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.20.nar_self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.20.nar_self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.20.nar_self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.20.nar_self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.20.mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.20.mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.20.mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.20.nar_mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.20.nar_mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.20.nar_mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.21.self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.21.self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.21.self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.21.self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.21.nar_self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.21.nar_self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.21.nar_self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.21.nar_self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.21.mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.21.mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.21.mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.21.nar_mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.21.nar_mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.21.nar_mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.22.self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.22.self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.22.self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.22.self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.22.nar_self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.22.nar_self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.22.nar_self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.22.nar_self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.22.mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.22.mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.22.mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.22.nar_mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.22.nar_mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.22.nar_mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.23.self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.23.self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.23.self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.23.self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.23.nar_self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.23.nar_self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.23.nar_self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.23.nar_self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.23.mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.23.mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.23.mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.23.nar_mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.23.nar_mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.23.nar_mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.24.self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.24.self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.24.self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.24.self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.24.nar_self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.24.nar_self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.24.nar_self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.24.nar_self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.24.mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.24.mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.24.mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.24.nar_mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.24.nar_mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.24.nar_mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.25.self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.25.self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.25.self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.25.self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.25.nar_self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.25.nar_self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.25.nar_self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.25.nar_self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.25.mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.25.mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.25.mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.25.nar_mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.25.nar_mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.25.nar_mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.26.self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.26.self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.26.self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.26.self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.26.nar_self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.26.nar_self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.26.nar_self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.26.nar_self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.26.mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.26.mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.26.mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.26.nar_mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.26.nar_mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.26.nar_mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.27.self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.27.self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.27.self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.27.self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.27.nar_self_attn.q_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.27.nar_self_attn.k_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.27.nar_self_attn.v_proj", "in_features": 2048, "out_features": 1024, "bias": false, "weight_parameters": 2097152, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.27.nar_self_attn.o_proj", "in_features": 2048, "out_features": 2048, "bias": false, "weight_parameters": 4194304, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.27.mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.27.mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.27.mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.27.nar_mlp.gate_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.27.nar_mlp.up_proj", "in_features": 2048, "out_features": 6144, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 }, { "name": "model.layers.27.nar_mlp.down_proj", "in_features": 6144, "out_features": 2048, "bias": false, "weight_parameters": 12582912, "weight_bits": 4, "activation_bits": 4 } ], "quantized_weight_parameters": 2818572288, "quantized_modules": 392, "kept_precision": [ "model.embed_tokens", "lm_head", "normalization", "vae2llm", "llm2vae", "time_embedder", "YuE2-Vae FP32" ], "elapsed_seconds": 16.647601778851822 }