mergekit/_data/architectures/qwen.json

{
    "model_type": "qwen",
    "architectures": [
        "QWenLMHeadModel"
    ],
    "pre_weights": [
        {
            "name": "transformer.wte.weight",
            "is_embed": true
        }
    ],
    "post_weights": [
        {
            "name": "transformer.ln_f.weight"
        },
        {
            "name": "lm_head.weight",
            "is_embed": true
        }
    ],
    "num_layers_config_key": "num_hidden_layers",
    "layer_templates": {
        "weights": [
            {
                "name": "transformer.h.${layer_index}.attn.c_attn.bias"
            },
            {
                "name": "transformer.h.${layer_index}.attn.c_attn.weight"
            },
            {
                "name": "transformer.h.${layer_index}.attn.c_proj.weight"
            },
            {
                "name": "transformer.h.${layer_index}.ln_1.weight"
            },
            {
                "name": "transformer.h.${layer_index}.ln_2.weight"
            },
            {
                "name": "transformer.h.${layer_index}.mlp.c_proj.weight"
            },
            {
                "name": "transformer.h.${layer_index}.mlp.w1.weight"
            },
            {
                "name": "transformer.h.${layer_index}.mlp.w2.weight"
            }
        ]
    }
}