transformer/config.json

{
    "_class_name": "WanTransformer3DModel",
    "_diffusers_version": "0.37.1",
    "_name_or_path": "/lustre/fsw/coreai_mlperf_inference/tinyinl/models/Wan2.2-T2V-A14B-Diffusers/transformer",
    "added_kv_proj_dim": null,
    "attention_head_dim": 128,
    "cross_attn_norm": true,
    "eps": 1e-06,
    "ffn_dim": 13824,
    "freq_dim": 256,
    "image_dim": null,
    "in_channels": 16,
    "num_attention_heads": 40,
    "num_layers": 40,
    "out_channels": 16,
    "patch_size": [
        1,
        2,
        2
    ],
    "pos_embed_seq_len": null,
    "qk_norm": "rms_norm_across_heads",
    "rope_max_seq_len": 1024,
    "text_dim": 4096,
    "quantization_config": {
        "config_groups": {
            "group_0": {
                "input_activations": {
                    "dynamic": false,
                    "num_bits": 8,
                    "type": "float"
                },
                "weights": {
                    "dynamic": false,
                    "num_bits": 8,
                    "type": "float"
                },
                "targets": [
                    "Linear"
                ]
            }
        },
        "ignore": [
            "blocks.0*",
            "blocks.1.*",
            "blocks.2.*",
            "blocks.37*",
            "blocks.38*",
            "blocks.39*",
            "condition_embedder*",
            "patch_embedding",
            "proj_out"
        ],
        "quant_algo": "FP8",
        "kv_cache_scheme": {
            "dynamic": false,
            "num_bits": 8,
            "type": "float"
        },
        "producer": {
            "name": "modelopt",
            "version": "0.46.0.dev170+gf479e7890"
        },
        "quant_method": "modelopt"
    }
}