{ "metadata": { "total_size": 14820847264 }, "weight_map": { "model.blocks.0.feed_forward.linear_fc1.weight": "model-00001-of-00003.safetensors", "model.blocks.0.feed_forward.linear_fc1_lora_A_list.0.weight": "model-00001-of-00003.safetensors", "model.blocks.0.feed_forward.linear_fc1_lora_A_list.1.weight": "model-00001-of-00003.safetensors", "model.blocks.0.feed_forward.linear_fc1_lora_A_list.10.weight": "model-00001-of-00003.safetensors", "model.blocks.0.feed_forward.linear_fc1_lora_A_list.11.weight": "model-00001-of-00003.safetensors", "model.blocks.0.feed_forward.linear_fc1_lora_A_list.12.weight": "model-00001-of-00003.safetensors", "model.blocks.0.feed_forward.linear_fc1_lora_A_list.2.weight": "model-00001-of-00003.safetensors", "model.blocks.0.feed_forward.linear_fc1_lora_A_list.3.weight": "model-00001-of-00003.safetensors", "model.blocks.0.feed_forward.linear_fc1_lora_A_list.4.weight": "model-00001-of-00003.safetensors", "model.blocks.0.feed_forward.linear_fc1_lora_A_list.5.weight": "model-00001-of-00003.safetensors", "model.blocks.0.feed_forward.linear_fc1_lora_A_list.6.weight": "model-00001-of-00003.safetensors", "model.blocks.0.feed_forward.linear_fc1_lora_A_list.7.weight": "model-00001-of-00003.safetensors", "model.blocks.0.feed_forward.linear_fc1_lora_A_list.8.weight": "model-00001-of-00003.safetensors", "model.blocks.0.feed_forward.linear_fc1_lora_A_list.9.weight": "model-00001-of-00003.safetensors", "model.blocks.0.feed_forward.linear_fc1_lora_B_list.0.weight": "model-00001-of-00003.safetensors", "model.blocks.0.feed_forward.linear_fc1_lora_B_list.1.weight": "model-00001-of-00003.safetensors", "model.blocks.0.feed_forward.linear_fc1_lora_B_list.10.weight": "model-00001-of-00003.safetensors", "model.blocks.0.feed_forward.linear_fc1_lora_B_list.11.weight": "model-00001-of-00003.safetensors", "model.blocks.0.feed_forward.linear_fc1_lora_B_list.12.weight": "model-00001-of-00003.safetensors", "model.blocks.0.feed_forward.linear_fc1_lora_B_list.2.weight": "model-00001-of-00003.safetensors", "model.blocks.0.feed_forward.linear_fc1_lora_B_list.3.weight": "model-00001-of-00003.safetensors", "model.blocks.0.feed_forward.linear_fc1_lora_B_list.4.weight": "model-00001-of-00003.safetensors", "model.blocks.0.feed_forward.linear_fc1_lora_B_list.5.weight": "model-00001-of-00003.safetensors", "model.blocks.0.feed_forward.linear_fc1_lora_B_list.6.weight": "model-00001-of-00003.safetensors", "model.blocks.0.feed_forward.linear_fc1_lora_B_list.7.weight": "model-00001-of-00003.safetensors", "model.blocks.0.feed_forward.linear_fc1_lora_B_list.8.weight": "model-00001-of-00003.safetensors", "model.blocks.0.feed_forward.linear_fc1_lora_B_list.9.weight": "model-00001-of-00003.safetensors", "model.blocks.0.feed_forward.linear_fc2.weight": "model-00001-of-00003.safetensors", "model.blocks.0.input_layernorm.weight": "model-00001-of-00003.safetensors", "model.blocks.0.pre_ff_layernorm.weight": "model-00001-of-00003.safetensors", "model.blocks.0.self_attn.k_proj.weight": "model-00001-of-00003.safetensors", "model.blocks.0.self_attn.o_proj.weight": "model-00001-of-00003.safetensors", "model.blocks.0.self_attn.q_proj.weight": "model-00001-of-00003.safetensors", "model.blocks.0.self_attn.v_proj.weight": "model-00001-of-00003.safetensors", "model.blocks.1.feed_forward.linear_fc1.weight": "model-00001-of-00003.safetensors", "model.blocks.1.feed_forward.linear_fc1_lora_A_list.0.weight": "model-00001-of-00003.safetensors", "model.blocks.1.feed_forward.linear_fc1_lora_A_list.1.weight": "model-00001-of-00003.safetensors", "model.blocks.1.feed_forward.linear_fc1_lora_A_list.10.weight": "model-00001-of-00003.safetensors", "model.blocks.1.feed_forward.linear_fc1_lora_A_list.11.weight": "model-00001-of-00003.safetensors", "model.blocks.1.feed_forward.linear_fc1_lora_A_list.12.weight": "model-00001-of-00003.safetensors", "model.blocks.1.feed_forward.linear_fc1_lora_A_list.2.weight": "model-00001-of-00003.safetensors", "model.blocks.1.feed_forward.linear_fc1_lora_A_list.3.weight": "model-00001-of-00003.safetensors", "model.blocks.1.feed_forward.linear_fc1_lora_A_list.4.weight": "model-00001-of-00003.safetensors", "model.blocks.1.feed_forward.linear_fc1_lora_A_list.5.weight": "model-00001-of-00003.safetensors", "model.blocks.1.feed_forward.linear_fc1_lora_A_list.6.weight": "model-00001-of-00003.safetensors", "model.blocks.1.feed_forward.linear_fc1_lora_A_list.7.weight": "model-00001-of-00003.safetensors", "model.blocks.1.feed_forward.linear_fc1_lora_A_list.8.weight": "model-00001-of-00003.safetensors", "model.blocks.1.feed_forward.linear_fc1_lora_A_list.9.weight": "model-00001-of-00003.safetensors", "model.blocks.1.feed_forward.linear_fc1_lora_B_list.0.weight": "model-00001-of-00003.safetensors", "model.blocks.1.feed_forward.linear_fc1_lora_B_list.1.weight": "model-00001-of-00003.safetensors", "model.blocks.1.feed_forward.linear_fc1_lora_B_list.10.weight": "model-00001-of-00003.safetensors", "model.blocks.1.feed_forward.linear_fc1_lora_B_list.11.weight": "model-00001-of-00003.safetensors", "model.blocks.1.feed_forward.linear_fc1_lora_B_list.12.weight": "model-00001-of-00003.safetensors", "model.blocks.1.feed_forward.linear_fc1_lora_B_list.2.weight": "model-00001-of-00003.safetensors", "model.blocks.1.feed_forward.linear_fc1_lora_B_list.3.weight": "model-00001-of-00003.safetensors", "model.blocks.1.feed_forward.linear_fc1_lora_B_list.4.weight": "model-00001-of-00003.safetensors", "model.blocks.1.feed_forward.linear_fc1_lora_B_list.5.weight": "model-00001-of-00003.safetensors", "model.blocks.1.feed_forward.linear_fc1_lora_B_list.6.weight": "model-00001-of-00003.safetensors", "model.blocks.1.feed_forward.linear_fc1_lora_B_list.7.weight": "model-00001-of-00003.safetensors", "model.blocks.1.feed_forward.linear_fc1_lora_B_list.8.weight": "model-00001-of-00003.safetensors", "model.blocks.1.feed_forward.linear_fc1_lora_B_list.9.weight": "model-00001-of-00003.safetensors", "model.blocks.1.feed_forward.linear_fc2.weight": "model-00001-of-00003.safetensors", "model.blocks.1.input_layernorm.weight": "model-00001-of-00003.safetensors", "model.blocks.1.pre_ff_layernorm.weight": "model-00001-of-00003.safetensors", "model.blocks.1.self_attn.k_proj.weight": "model-00001-of-00003.safetensors", "model.blocks.1.self_attn.o_proj.weight": "model-00001-of-00003.safetensors", "model.blocks.1.self_attn.q_proj.weight": "model-00001-of-00003.safetensors", "model.blocks.1.self_attn.v_proj.weight": "model-00001-of-00003.safetensors", "model.embed_tokens.weight": "model-00001-of-00003.safetensors", "model.final_layernorm.weight": "model-00003-of-00003.safetensors", "model.linear_layers.0.weight": "model-00003-of-00003.safetensors", "model.linear_layers.1.weight": "model-00003-of-00003.safetensors", "model.linear_layers.10.weight": "model-00003-of-00003.safetensors", "model.linear_layers.11.weight": "model-00003-of-00003.safetensors", "model.linear_layers.12.weight": "model-00003-of-00003.safetensors", "model.linear_layers.2.weight": "model-00003-of-00003.safetensors", "model.linear_layers.3.weight": "model-00003-of-00003.safetensors", "model.linear_layers.4.weight": "model-00003-of-00003.safetensors", "model.linear_layers.5.weight": "model-00003-of-00003.safetensors", "model.linear_layers.6.weight": "model-00003-of-00003.safetensors", "model.linear_layers.7.weight": "model-00003-of-00003.safetensors", "model.linear_layers.8.weight": "model-00003-of-00003.safetensors", "model.linear_layers.9.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.0.input_layernorm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.0.mamba.A_log": "model-00001-of-00003.safetensors", "model.mamba_layers.0.mamba.D": "model-00001-of-00003.safetensors", "model.mamba_layers.0.mamba.conv1d.bias": "model-00001-of-00003.safetensors", "model.mamba_layers.0.mamba.conv1d.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.0.mamba.dt_bias": "model-00001-of-00003.safetensors", "model.mamba_layers.0.mamba.in_proj.0.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.0.mamba.norm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.0.mamba.out_proj.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.1.input_layernorm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.1.mamba.A_log": "model-00001-of-00003.safetensors", "model.mamba_layers.1.mamba.D": "model-00001-of-00003.safetensors", "model.mamba_layers.1.mamba.conv1d.bias": "model-00001-of-00003.safetensors", "model.mamba_layers.1.mamba.conv1d.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.1.mamba.dt_bias": "model-00001-of-00003.safetensors", "model.mamba_layers.1.mamba.in_proj.0.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.1.mamba.norm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.1.mamba.out_proj.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.10.input_layernorm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.10.mamba.A_log": "model-00001-of-00003.safetensors", "model.mamba_layers.10.mamba.D": "model-00001-of-00003.safetensors", "model.mamba_layers.10.mamba.conv1d.bias": "model-00001-of-00003.safetensors", "model.mamba_layers.10.mamba.conv1d.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.10.mamba.dt_bias": "model-00001-of-00003.safetensors", "model.mamba_layers.10.mamba.in_proj.0.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.10.mamba.norm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.10.mamba.out_proj.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.11.input_layernorm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.11.mamba.A_log": "model-00001-of-00003.safetensors", "model.mamba_layers.11.mamba.D": "model-00001-of-00003.safetensors", "model.mamba_layers.11.mamba.conv1d.bias": "model-00001-of-00003.safetensors", "model.mamba_layers.11.mamba.conv1d.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.11.mamba.dt_bias": "model-00001-of-00003.safetensors", "model.mamba_layers.11.mamba.in_proj.0.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.11.mamba.norm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.11.mamba.out_proj.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.12.input_layernorm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.12.mamba.A_log": "model-00001-of-00003.safetensors", "model.mamba_layers.12.mamba.D": "model-00001-of-00003.safetensors", "model.mamba_layers.12.mamba.conv1d.bias": "model-00001-of-00003.safetensors", "model.mamba_layers.12.mamba.conv1d.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.12.mamba.dt_bias": "model-00001-of-00003.safetensors", "model.mamba_layers.12.mamba.in_proj.0.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.12.mamba.norm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.12.mamba.out_proj.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.13.input_layernorm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.13.mamba.A_log": "model-00001-of-00003.safetensors", "model.mamba_layers.13.mamba.D": "model-00001-of-00003.safetensors", "model.mamba_layers.13.mamba.conv1d.bias": "model-00001-of-00003.safetensors", "model.mamba_layers.13.mamba.conv1d.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.13.mamba.dt_bias": "model-00001-of-00003.safetensors", "model.mamba_layers.13.mamba.in_proj.0.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.13.mamba.norm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.13.mamba.out_proj.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.14.input_layernorm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.14.mamba.A_log": "model-00001-of-00003.safetensors", "model.mamba_layers.14.mamba.D": "model-00001-of-00003.safetensors", "model.mamba_layers.14.mamba.conv1d.bias": "model-00001-of-00003.safetensors", "model.mamba_layers.14.mamba.conv1d.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.14.mamba.dt_bias": "model-00001-of-00003.safetensors", "model.mamba_layers.14.mamba.in_proj.0.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.14.mamba.norm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.14.mamba.out_proj.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.15.input_layernorm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.15.mamba.A_log": "model-00001-of-00003.safetensors", "model.mamba_layers.15.mamba.D": "model-00001-of-00003.safetensors", "model.mamba_layers.15.mamba.conv1d.bias": "model-00001-of-00003.safetensors", "model.mamba_layers.15.mamba.conv1d.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.15.mamba.dt_bias": "model-00001-of-00003.safetensors", "model.mamba_layers.15.mamba.in_proj.0.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.15.mamba.norm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.15.mamba.out_proj.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.16.input_layernorm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.16.mamba.A_log": "model-00001-of-00003.safetensors", "model.mamba_layers.16.mamba.D": "model-00001-of-00003.safetensors", "model.mamba_layers.16.mamba.conv1d.bias": "model-00001-of-00003.safetensors", "model.mamba_layers.16.mamba.conv1d.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.16.mamba.dt_bias": "model-00001-of-00003.safetensors", "model.mamba_layers.16.mamba.in_proj.0.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.16.mamba.norm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.16.mamba.out_proj.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.17.input_layernorm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.17.mamba.A_log": "model-00001-of-00003.safetensors", "model.mamba_layers.17.mamba.D": "model-00001-of-00003.safetensors", "model.mamba_layers.17.mamba.conv1d.bias": "model-00001-of-00003.safetensors", "model.mamba_layers.17.mamba.conv1d.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.17.mamba.dt_bias": "model-00001-of-00003.safetensors", "model.mamba_layers.17.mamba.in_proj.0.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.17.mamba.norm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.17.mamba.out_proj.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.18.input_layernorm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.18.mamba.A_log": "model-00001-of-00003.safetensors", "model.mamba_layers.18.mamba.D": "model-00001-of-00003.safetensors", "model.mamba_layers.18.mamba.conv1d.bias": "model-00001-of-00003.safetensors", "model.mamba_layers.18.mamba.conv1d.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.18.mamba.dt_bias": "model-00001-of-00003.safetensors", "model.mamba_layers.18.mamba.in_proj.0.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.18.mamba.norm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.18.mamba.out_proj.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.19.input_layernorm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.19.mamba.A_log": "model-00001-of-00003.safetensors", "model.mamba_layers.19.mamba.D": "model-00001-of-00003.safetensors", "model.mamba_layers.19.mamba.conv1d.bias": "model-00001-of-00003.safetensors", "model.mamba_layers.19.mamba.conv1d.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.19.mamba.dt_bias": "model-00001-of-00003.safetensors", "model.mamba_layers.19.mamba.in_proj.0.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.19.mamba.norm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.19.mamba.out_proj.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.2.input_layernorm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.2.mamba.A_log": "model-00001-of-00003.safetensors", "model.mamba_layers.2.mamba.D": "model-00001-of-00003.safetensors", "model.mamba_layers.2.mamba.conv1d.bias": "model-00001-of-00003.safetensors", "model.mamba_layers.2.mamba.conv1d.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.2.mamba.dt_bias": "model-00001-of-00003.safetensors", "model.mamba_layers.2.mamba.in_proj.0.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.2.mamba.norm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.2.mamba.out_proj.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.20.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.20.mamba.A_log": "model-00001-of-00003.safetensors", "model.mamba_layers.20.mamba.D": "model-00001-of-00003.safetensors", "model.mamba_layers.20.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.20.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.20.mamba.dt_bias": "model-00001-of-00003.safetensors", "model.mamba_layers.20.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.20.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.20.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.21.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.21.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.21.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.21.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.21.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.21.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.21.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.21.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.21.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.22.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.22.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.22.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.22.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.22.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.22.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.22.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.22.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.22.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.23.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.23.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.23.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.23.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.23.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.23.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.23.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.23.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.23.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.24.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.24.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.24.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.24.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.24.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.24.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.24.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.24.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.24.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.25.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.25.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.25.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.25.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.25.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.25.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.25.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.25.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.25.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.26.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.26.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.26.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.26.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.26.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.26.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.26.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.26.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.26.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.27.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.27.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.27.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.27.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.27.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.27.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.27.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.27.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.27.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.28.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.28.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.28.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.28.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.28.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.28.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.28.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.28.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.28.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.29.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.29.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.29.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.29.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.29.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.29.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.29.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.29.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.29.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.3.input_layernorm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.3.mamba.A_log": "model-00001-of-00003.safetensors", "model.mamba_layers.3.mamba.D": "model-00001-of-00003.safetensors", "model.mamba_layers.3.mamba.conv1d.bias": "model-00001-of-00003.safetensors", "model.mamba_layers.3.mamba.conv1d.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.3.mamba.dt_bias": "model-00001-of-00003.safetensors", "model.mamba_layers.3.mamba.in_proj.0.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.3.mamba.norm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.3.mamba.out_proj.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.30.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.30.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.30.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.30.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.30.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.30.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.30.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.30.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.30.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.31.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.31.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.31.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.31.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.31.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.31.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.31.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.31.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.31.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.32.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.32.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.32.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.32.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.32.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.32.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.32.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.32.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.32.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.33.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.33.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.33.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.33.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.33.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.33.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.33.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.33.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.33.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.34.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.34.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.34.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.34.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.34.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.34.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.34.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.34.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.34.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.35.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.35.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.35.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.35.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.35.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.35.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.35.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.35.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.35.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.36.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.36.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.36.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.36.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.36.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.36.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.36.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.36.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.36.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.37.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.37.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.37.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.37.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.37.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.37.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.37.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.37.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.37.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.38.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.38.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.38.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.38.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.38.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.38.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.38.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.38.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.38.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.39.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.39.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.39.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.39.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.39.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.39.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.39.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.39.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.39.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.4.input_layernorm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.4.mamba.A_log": "model-00001-of-00003.safetensors", "model.mamba_layers.4.mamba.D": "model-00001-of-00003.safetensors", "model.mamba_layers.4.mamba.conv1d.bias": "model-00001-of-00003.safetensors", "model.mamba_layers.4.mamba.conv1d.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.4.mamba.dt_bias": "model-00001-of-00003.safetensors", "model.mamba_layers.4.mamba.in_proj.0.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.4.mamba.norm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.4.mamba.out_proj.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.40.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.40.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.40.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.40.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.40.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.40.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.40.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.40.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.40.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.41.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.41.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.41.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.41.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.41.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.41.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.41.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.41.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.41.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.42.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.42.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.42.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.42.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.42.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.42.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.42.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.42.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.42.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.43.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.43.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.43.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.43.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.43.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.43.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.43.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.43.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.43.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.44.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.44.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.44.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.44.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.44.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.44.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.44.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.44.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.44.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.45.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.45.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.45.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.45.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.45.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.45.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.45.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.45.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.45.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.46.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.46.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.46.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.46.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.46.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.46.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.46.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.46.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.46.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.47.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.47.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.47.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.47.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.47.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.47.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.47.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.47.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.47.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.48.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.48.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.48.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.48.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.48.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.48.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.48.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.48.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.48.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.49.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.49.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.49.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.49.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.49.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.49.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.49.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.49.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.49.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.5.input_layernorm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.5.mamba.A_log": "model-00001-of-00003.safetensors", "model.mamba_layers.5.mamba.D": "model-00001-of-00003.safetensors", "model.mamba_layers.5.mamba.conv1d.bias": "model-00001-of-00003.safetensors", "model.mamba_layers.5.mamba.conv1d.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.5.mamba.dt_bias": "model-00001-of-00003.safetensors", "model.mamba_layers.5.mamba.in_proj.0.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.5.mamba.norm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.5.mamba.out_proj.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.50.input_layernorm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.50.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.50.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.50.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.50.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.50.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.50.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.50.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.50.mamba.out_proj.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.51.input_layernorm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.51.mamba.A_log": "model-00002-of-00003.safetensors", "model.mamba_layers.51.mamba.D": "model-00002-of-00003.safetensors", "model.mamba_layers.51.mamba.conv1d.bias": "model-00002-of-00003.safetensors", "model.mamba_layers.51.mamba.conv1d.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.51.mamba.dt_bias": "model-00002-of-00003.safetensors", "model.mamba_layers.51.mamba.in_proj.0.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.51.mamba.norm.weight": "model-00002-of-00003.safetensors", "model.mamba_layers.51.mamba.out_proj.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.52.input_layernorm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.52.mamba.A_log": "model-00003-of-00003.safetensors", "model.mamba_layers.52.mamba.D": "model-00003-of-00003.safetensors", "model.mamba_layers.52.mamba.conv1d.bias": "model-00003-of-00003.safetensors", "model.mamba_layers.52.mamba.conv1d.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.52.mamba.dt_bias": "model-00003-of-00003.safetensors", "model.mamba_layers.52.mamba.in_proj.0.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.52.mamba.norm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.52.mamba.out_proj.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.53.input_layernorm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.53.mamba.A_log": "model-00003-of-00003.safetensors", "model.mamba_layers.53.mamba.D": "model-00003-of-00003.safetensors", "model.mamba_layers.53.mamba.conv1d.bias": "model-00003-of-00003.safetensors", "model.mamba_layers.53.mamba.conv1d.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.53.mamba.dt_bias": "model-00003-of-00003.safetensors", "model.mamba_layers.53.mamba.in_proj.0.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.53.mamba.norm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.53.mamba.out_proj.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.54.input_layernorm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.54.mamba.A_log": "model-00003-of-00003.safetensors", "model.mamba_layers.54.mamba.D": "model-00003-of-00003.safetensors", "model.mamba_layers.54.mamba.conv1d.bias": "model-00003-of-00003.safetensors", "model.mamba_layers.54.mamba.conv1d.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.54.mamba.dt_bias": "model-00003-of-00003.safetensors", "model.mamba_layers.54.mamba.in_proj.0.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.54.mamba.norm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.54.mamba.out_proj.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.55.input_layernorm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.55.mamba.A_log": "model-00003-of-00003.safetensors", "model.mamba_layers.55.mamba.D": "model-00003-of-00003.safetensors", "model.mamba_layers.55.mamba.conv1d.bias": "model-00003-of-00003.safetensors", "model.mamba_layers.55.mamba.conv1d.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.55.mamba.dt_bias": "model-00003-of-00003.safetensors", "model.mamba_layers.55.mamba.in_proj.0.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.55.mamba.norm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.55.mamba.out_proj.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.56.input_layernorm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.56.mamba.A_log": "model-00003-of-00003.safetensors", "model.mamba_layers.56.mamba.D": "model-00003-of-00003.safetensors", "model.mamba_layers.56.mamba.conv1d.bias": "model-00003-of-00003.safetensors", "model.mamba_layers.56.mamba.conv1d.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.56.mamba.dt_bias": "model-00003-of-00003.safetensors", "model.mamba_layers.56.mamba.in_proj.0.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.56.mamba.norm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.56.mamba.out_proj.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.57.input_layernorm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.57.mamba.A_log": "model-00003-of-00003.safetensors", "model.mamba_layers.57.mamba.D": "model-00003-of-00003.safetensors", "model.mamba_layers.57.mamba.conv1d.bias": "model-00003-of-00003.safetensors", "model.mamba_layers.57.mamba.conv1d.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.57.mamba.dt_bias": "model-00003-of-00003.safetensors", "model.mamba_layers.57.mamba.in_proj.0.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.57.mamba.norm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.57.mamba.out_proj.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.58.input_layernorm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.58.mamba.A_log": "model-00003-of-00003.safetensors", "model.mamba_layers.58.mamba.D": "model-00003-of-00003.safetensors", "model.mamba_layers.58.mamba.conv1d.bias": "model-00003-of-00003.safetensors", "model.mamba_layers.58.mamba.conv1d.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.58.mamba.dt_bias": "model-00003-of-00003.safetensors", "model.mamba_layers.58.mamba.in_proj.0.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.58.mamba.norm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.58.mamba.out_proj.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.59.input_layernorm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.59.mamba.A_log": "model-00003-of-00003.safetensors", "model.mamba_layers.59.mamba.D": "model-00003-of-00003.safetensors", "model.mamba_layers.59.mamba.conv1d.bias": "model-00003-of-00003.safetensors", "model.mamba_layers.59.mamba.conv1d.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.59.mamba.dt_bias": "model-00003-of-00003.safetensors", "model.mamba_layers.59.mamba.in_proj.0.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.59.mamba.norm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.59.mamba.out_proj.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.6.input_layernorm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.6.mamba.A_log": "model-00001-of-00003.safetensors", "model.mamba_layers.6.mamba.D": "model-00001-of-00003.safetensors", "model.mamba_layers.6.mamba.conv1d.bias": "model-00001-of-00003.safetensors", "model.mamba_layers.6.mamba.conv1d.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.6.mamba.dt_bias": "model-00001-of-00003.safetensors", "model.mamba_layers.6.mamba.in_proj.0.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.6.mamba.norm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.6.mamba.out_proj.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.60.input_layernorm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.60.mamba.A_log": "model-00003-of-00003.safetensors", "model.mamba_layers.60.mamba.D": "model-00003-of-00003.safetensors", "model.mamba_layers.60.mamba.conv1d.bias": "model-00003-of-00003.safetensors", "model.mamba_layers.60.mamba.conv1d.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.60.mamba.dt_bias": "model-00003-of-00003.safetensors", "model.mamba_layers.60.mamba.in_proj.0.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.60.mamba.norm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.60.mamba.out_proj.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.61.input_layernorm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.61.mamba.A_log": "model-00003-of-00003.safetensors", "model.mamba_layers.61.mamba.D": "model-00003-of-00003.safetensors", "model.mamba_layers.61.mamba.conv1d.bias": "model-00003-of-00003.safetensors", "model.mamba_layers.61.mamba.conv1d.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.61.mamba.dt_bias": "model-00003-of-00003.safetensors", "model.mamba_layers.61.mamba.in_proj.0.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.61.mamba.norm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.61.mamba.out_proj.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.62.input_layernorm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.62.mamba.A_log": "model-00003-of-00003.safetensors", "model.mamba_layers.62.mamba.D": "model-00003-of-00003.safetensors", "model.mamba_layers.62.mamba.conv1d.bias": "model-00003-of-00003.safetensors", "model.mamba_layers.62.mamba.conv1d.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.62.mamba.dt_bias": "model-00003-of-00003.safetensors", "model.mamba_layers.62.mamba.in_proj.0.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.62.mamba.norm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.62.mamba.out_proj.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.63.input_layernorm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.63.mamba.A_log": "model-00003-of-00003.safetensors", "model.mamba_layers.63.mamba.D": "model-00003-of-00003.safetensors", "model.mamba_layers.63.mamba.conv1d.bias": "model-00003-of-00003.safetensors", "model.mamba_layers.63.mamba.conv1d.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.63.mamba.dt_bias": "model-00003-of-00003.safetensors", "model.mamba_layers.63.mamba.in_proj.0.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.63.mamba.norm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.63.mamba.out_proj.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.64.input_layernorm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.64.mamba.A_log": "model-00003-of-00003.safetensors", "model.mamba_layers.64.mamba.D": "model-00003-of-00003.safetensors", "model.mamba_layers.64.mamba.conv1d.bias": "model-00003-of-00003.safetensors", "model.mamba_layers.64.mamba.conv1d.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.64.mamba.dt_bias": "model-00003-of-00003.safetensors", "model.mamba_layers.64.mamba.in_proj.0.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.64.mamba.norm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.64.mamba.out_proj.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.65.input_layernorm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.65.mamba.A_log": "model-00003-of-00003.safetensors", "model.mamba_layers.65.mamba.D": "model-00003-of-00003.safetensors", "model.mamba_layers.65.mamba.conv1d.bias": "model-00003-of-00003.safetensors", "model.mamba_layers.65.mamba.conv1d.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.65.mamba.dt_bias": "model-00003-of-00003.safetensors", "model.mamba_layers.65.mamba.in_proj.0.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.65.mamba.norm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.65.mamba.out_proj.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.66.input_layernorm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.66.mamba.A_log": "model-00003-of-00003.safetensors", "model.mamba_layers.66.mamba.D": "model-00003-of-00003.safetensors", "model.mamba_layers.66.mamba.conv1d.bias": "model-00003-of-00003.safetensors", "model.mamba_layers.66.mamba.conv1d.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.66.mamba.dt_bias": "model-00003-of-00003.safetensors", "model.mamba_layers.66.mamba.in_proj.0.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.66.mamba.norm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.66.mamba.out_proj.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.67.input_layernorm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.67.mamba.A_log": "model-00003-of-00003.safetensors", "model.mamba_layers.67.mamba.D": "model-00003-of-00003.safetensors", "model.mamba_layers.67.mamba.conv1d.bias": "model-00003-of-00003.safetensors", "model.mamba_layers.67.mamba.conv1d.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.67.mamba.dt_bias": "model-00003-of-00003.safetensors", "model.mamba_layers.67.mamba.in_proj.0.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.67.mamba.norm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.67.mamba.out_proj.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.68.input_layernorm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.68.mamba.A_log": "model-00003-of-00003.safetensors", "model.mamba_layers.68.mamba.D": "model-00003-of-00003.safetensors", "model.mamba_layers.68.mamba.conv1d.bias": "model-00003-of-00003.safetensors", "model.mamba_layers.68.mamba.conv1d.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.68.mamba.dt_bias": "model-00003-of-00003.safetensors", "model.mamba_layers.68.mamba.in_proj.0.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.68.mamba.norm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.68.mamba.out_proj.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.69.input_layernorm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.69.mamba.A_log": "model-00003-of-00003.safetensors", "model.mamba_layers.69.mamba.D": "model-00003-of-00003.safetensors", "model.mamba_layers.69.mamba.conv1d.bias": "model-00003-of-00003.safetensors", "model.mamba_layers.69.mamba.conv1d.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.69.mamba.dt_bias": "model-00003-of-00003.safetensors", "model.mamba_layers.69.mamba.in_proj.0.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.69.mamba.norm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.69.mamba.out_proj.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.7.input_layernorm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.7.mamba.A_log": "model-00001-of-00003.safetensors", "model.mamba_layers.7.mamba.D": "model-00001-of-00003.safetensors", "model.mamba_layers.7.mamba.conv1d.bias": "model-00001-of-00003.safetensors", "model.mamba_layers.7.mamba.conv1d.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.7.mamba.dt_bias": "model-00001-of-00003.safetensors", "model.mamba_layers.7.mamba.in_proj.0.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.7.mamba.norm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.7.mamba.out_proj.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.70.input_layernorm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.70.mamba.A_log": "model-00003-of-00003.safetensors", "model.mamba_layers.70.mamba.D": "model-00003-of-00003.safetensors", "model.mamba_layers.70.mamba.conv1d.bias": "model-00003-of-00003.safetensors", "model.mamba_layers.70.mamba.conv1d.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.70.mamba.dt_bias": "model-00003-of-00003.safetensors", "model.mamba_layers.70.mamba.in_proj.0.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.70.mamba.norm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.70.mamba.out_proj.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.71.input_layernorm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.71.mamba.A_log": "model-00003-of-00003.safetensors", "model.mamba_layers.71.mamba.D": "model-00003-of-00003.safetensors", "model.mamba_layers.71.mamba.conv1d.bias": "model-00003-of-00003.safetensors", "model.mamba_layers.71.mamba.conv1d.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.71.mamba.dt_bias": "model-00003-of-00003.safetensors", "model.mamba_layers.71.mamba.in_proj.0.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.71.mamba.norm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.71.mamba.out_proj.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.72.input_layernorm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.72.mamba.A_log": "model-00003-of-00003.safetensors", "model.mamba_layers.72.mamba.D": "model-00003-of-00003.safetensors", "model.mamba_layers.72.mamba.conv1d.bias": "model-00003-of-00003.safetensors", "model.mamba_layers.72.mamba.conv1d.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.72.mamba.dt_bias": "model-00003-of-00003.safetensors", "model.mamba_layers.72.mamba.in_proj.0.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.72.mamba.norm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.72.mamba.out_proj.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.73.input_layernorm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.73.mamba.A_log": "model-00003-of-00003.safetensors", "model.mamba_layers.73.mamba.D": "model-00003-of-00003.safetensors", "model.mamba_layers.73.mamba.conv1d.bias": "model-00003-of-00003.safetensors", "model.mamba_layers.73.mamba.conv1d.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.73.mamba.dt_bias": "model-00003-of-00003.safetensors", "model.mamba_layers.73.mamba.in_proj.0.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.73.mamba.norm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.73.mamba.out_proj.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.74.input_layernorm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.74.mamba.A_log": "model-00003-of-00003.safetensors", "model.mamba_layers.74.mamba.D": "model-00003-of-00003.safetensors", "model.mamba_layers.74.mamba.conv1d.bias": "model-00003-of-00003.safetensors", "model.mamba_layers.74.mamba.conv1d.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.74.mamba.dt_bias": "model-00003-of-00003.safetensors", "model.mamba_layers.74.mamba.in_proj.0.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.74.mamba.norm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.74.mamba.out_proj.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.75.input_layernorm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.75.mamba.A_log": "model-00003-of-00003.safetensors", "model.mamba_layers.75.mamba.D": "model-00003-of-00003.safetensors", "model.mamba_layers.75.mamba.conv1d.bias": "model-00003-of-00003.safetensors", "model.mamba_layers.75.mamba.conv1d.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.75.mamba.dt_bias": "model-00003-of-00003.safetensors", "model.mamba_layers.75.mamba.in_proj.0.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.75.mamba.norm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.75.mamba.out_proj.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.76.input_layernorm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.76.mamba.A_log": "model-00003-of-00003.safetensors", "model.mamba_layers.76.mamba.D": "model-00003-of-00003.safetensors", "model.mamba_layers.76.mamba.conv1d.bias": "model-00003-of-00003.safetensors", "model.mamba_layers.76.mamba.conv1d.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.76.mamba.dt_bias": "model-00003-of-00003.safetensors", "model.mamba_layers.76.mamba.in_proj.0.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.76.mamba.norm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.76.mamba.out_proj.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.77.input_layernorm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.77.mamba.A_log": "model-00003-of-00003.safetensors", "model.mamba_layers.77.mamba.D": "model-00003-of-00003.safetensors", "model.mamba_layers.77.mamba.conv1d.bias": "model-00003-of-00003.safetensors", "model.mamba_layers.77.mamba.conv1d.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.77.mamba.dt_bias": "model-00003-of-00003.safetensors", "model.mamba_layers.77.mamba.in_proj.0.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.77.mamba.norm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.77.mamba.out_proj.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.78.input_layernorm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.78.mamba.A_log": "model-00003-of-00003.safetensors", "model.mamba_layers.78.mamba.D": "model-00003-of-00003.safetensors", "model.mamba_layers.78.mamba.conv1d.bias": "model-00003-of-00003.safetensors", "model.mamba_layers.78.mamba.conv1d.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.78.mamba.dt_bias": "model-00003-of-00003.safetensors", "model.mamba_layers.78.mamba.in_proj.0.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.78.mamba.norm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.78.mamba.out_proj.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.79.input_layernorm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.79.mamba.A_log": "model-00003-of-00003.safetensors", "model.mamba_layers.79.mamba.D": "model-00003-of-00003.safetensors", "model.mamba_layers.79.mamba.conv1d.bias": "model-00003-of-00003.safetensors", "model.mamba_layers.79.mamba.conv1d.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.79.mamba.dt_bias": "model-00003-of-00003.safetensors", "model.mamba_layers.79.mamba.in_proj.0.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.79.mamba.norm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.79.mamba.out_proj.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.8.input_layernorm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.8.mamba.A_log": "model-00001-of-00003.safetensors", "model.mamba_layers.8.mamba.D": "model-00001-of-00003.safetensors", "model.mamba_layers.8.mamba.conv1d.bias": "model-00001-of-00003.safetensors", "model.mamba_layers.8.mamba.conv1d.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.8.mamba.dt_bias": "model-00001-of-00003.safetensors", "model.mamba_layers.8.mamba.in_proj.0.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.8.mamba.norm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.8.mamba.out_proj.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.80.input_layernorm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.80.mamba.A_log": "model-00003-of-00003.safetensors", "model.mamba_layers.80.mamba.D": "model-00003-of-00003.safetensors", "model.mamba_layers.80.mamba.conv1d.bias": "model-00003-of-00003.safetensors", "model.mamba_layers.80.mamba.conv1d.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.80.mamba.dt_bias": "model-00003-of-00003.safetensors", "model.mamba_layers.80.mamba.in_proj.0.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.80.mamba.norm.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.80.mamba.out_proj.weight": "model-00003-of-00003.safetensors", "model.mamba_layers.9.input_layernorm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.9.mamba.A_log": "model-00001-of-00003.safetensors", "model.mamba_layers.9.mamba.D": "model-00001-of-00003.safetensors", "model.mamba_layers.9.mamba.conv1d.bias": "model-00001-of-00003.safetensors", "model.mamba_layers.9.mamba.conv1d.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.9.mamba.dt_bias": "model-00001-of-00003.safetensors", "model.mamba_layers.9.mamba.in_proj.0.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.9.mamba.norm.weight": "model-00001-of-00003.safetensors", "model.mamba_layers.9.mamba.out_proj.weight": "model-00001-of-00003.safetensors" } }