{
  "alpha_pattern": {},
  "auto_mapping": null,
  "base_model_name_or_path": "/path/to/your/VITA_ckpt",
  "bias": "none",
  "fan_in_fan_out": false,
  "inference_mode": true,
  "init_lora_weights": true,
  "layers_pattern": null,
  "layers_to_transform": null,
  "loftq_config": {},
  "lora_alpha": 16,
  "lora_dropout": 0.05,
  "megatron_config": null,
  "megatron_core": "megatron.core",
  "modules_to_save": null,
  "peft_type": "LORA",
  "r": 64,
  "rank_pattern": {},
  "revision": null,
  "target_modules": [
    "model.layers.18.block_sparse_moe.experts.6.w3",
    "model.layers.18.block_sparse_moe.experts.3.w1",
    "model.layers.19.block_sparse_moe.experts.4.w1",
    "model.layers.30.block_sparse_moe.experts.4.w2",
    "model.audio_encoder.encoder.enc.1.encoders.15.self_attn.linear_q",
    "model.layers.0.block_sparse_moe.experts.3.w3",
    "model.layers.12.self_attn.v_proj",
    "model.layers.27.block_sparse_moe.experts.0.w1",
    "model.layers.8.block_sparse_moe.experts.5.w2",
    "model.audio_encoder.encoder.enc.1.encoders.5.self_attn.linear_out",
    "model.layers.2.block_sparse_moe.experts.4.w2",
    "model.layers.29.self_attn.q_proj",
    "model.layers.27.block_sparse_moe.experts.5.w1",
    "model.layers.7.self_attn.q_proj",
    "model.layers.17.block_sparse_moe.experts.1.w3",
    "model.audio_encoder.encoder.enc.1.encoders.15.self_attn.linear_k",
    "model.layers.23.block_sparse_moe.experts.7.w1",
    "model.layers.30.block_sparse_moe.experts.5.w1",
    "model.layers.20.block_sparse_moe.experts.0.w2",
    "model.layers.17.block_sparse_moe.experts.2.w2",
    "model.layers.15.block_sparse_moe.experts.5.w2",
    "model.layers.2.block_sparse_moe.experts.2.w3",
    "model.layers.22.block_sparse_moe.experts.5.w2",
    "model.layers.5.block_sparse_moe.experts.2.w3",
    "model.layers.6.block_sparse_moe.experts.3.w3",
    "model.layers.26.self_attn.o_proj",
    "model.layers.10.block_sparse_moe.experts.2.w2",
    "model.layers.31.block_sparse_moe.gate",
    "model.layers.15.block_sparse_moe.experts.3.w3",
    "model.layers.21.block_sparse_moe.experts.7.w1",
    "model.layers.2.block_sparse_moe.experts.6.w2",
    "model.layers.14.block_sparse_moe.experts.6.w3",
    "model.layers.25.self_attn.o_proj",
    "model.layers.26.block_sparse_moe.experts.0.w1",
    "model.layers.27.block_sparse_moe.experts.4.w1",
    "model.audio_encoder.encoder.enc.1.encoders.4.feed_forward.w_1",
    "model.layers.4.block_sparse_moe.experts.1.w2",
    "model.layers.14.block_sparse_moe.experts.7.w1",
    "model.layers.20.self_attn.k_proj",
    "model.audio_encoder.encoder.enc.1.encoders.23.self_attn.linear_k",
    "model.layers.10.block_sparse_moe.experts.3.w3",
    "model.layers.25.block_sparse_moe.experts.4.w3",
    "model.audio_encoder.encoder.enc.1.encoders.5.self_attn.linear_k",
    "model.layers.13.self_attn.o_proj",
    "model.layers.4.block_sparse_moe.experts.6.w3",
    "model.layers.21.block_sparse_moe.experts.3.w2",
    "model.layers.31.block_sparse_moe.experts.0.w1",
    "model.layers.3.block_sparse_moe.experts.2.w2",
    "model.layers.4.block_sparse_moe.experts.4.w1",
    "model.layers.16.block_sparse_moe.experts.0.w1",
    "model.layers.7.self_attn.k_proj",
    "model.layers.13.block_sparse_moe.experts.6.w3",
    "model.layers.27.block_sparse_moe.experts.2.w3",
    "model.layers.5.block_sparse_moe.experts.5.w3",
    "model.layers.14.block_sparse_moe.experts.5.w2",
    "model.layers.31.block_sparse_moe.experts.6.w3",
    "model.layers.22.block_sparse_moe.experts.4.w3",
    "model.layers.4.block_sparse_moe.experts.3.w3",
    "model.layers.23.self_attn.o_proj",
    "model.layers.30.block_sparse_moe.experts.4.w3",
    "model.layers.17.self_attn.q_proj",
    "model.layers.23.block_sparse_moe.experts.4.w3",
    "model.layers.18.block_sparse_moe.experts.5.w3",
    "model.layers.18.block_sparse_moe.experts.1.w1",
    "model.layers.26.block_sparse_moe.experts.6.w2",
    "model.audio_encoder.encoder.enc.1.encoders.3.self_attn.linear_k",
    "model.layers.26.block_sparse_moe.experts.0.w2",
    "model.layers.6.block_sparse_moe.experts.0.w3",
    "model.layers.12.block_sparse_moe.experts.7.w2",
    "model.layers.21.block_sparse_moe.experts.5.w2",
    "model.audio_encoder.encoder.enc.1.encoders.14.feed_forward.w_1",
    "model.layers.0.block_sparse_moe.experts.3.w1",
    "model.layers.2.block_sparse_moe.experts.0.w3",
    "model.audio_encoder.encoder.enc.1.encoders.4.self_attn.linear_q",
    "model.layers.11.block_sparse_moe.experts.5.w2",
    "model.layers.21.block_sparse_moe.gate",
    "model.layers.27.block_sparse_moe.experts.4.w3",
    "model.layers.29.self_attn.v_proj",
    "model.audio_encoder.encoder.enc.1.encoders.12.self_attn.linear_q",
    "model.layers.25.block_sparse_moe.experts.7.w3",
    "model.layers.25.block_sparse_moe.experts.7.w1",
    "model.layers.13.block_sparse_moe.experts.0.w2",
    "model.layers.6.block_sparse_moe.experts.2.w3",
    "model.layers.30.block_sparse_moe.experts.6.w2",
    "model.layers.18.block_sparse_moe.experts.4.w2",
    "model.layers.1.block_sparse_moe.experts.4.w2",
    "model.layers.1.block_sparse_moe.experts.5.w3",
    "model.layers.30.block_sparse_moe.experts.0.w1",
    "model.layers.16.self_attn.q_proj",
    "model.layers.0.block_sparse_moe.experts.1.w3",
    "model.audio_encoder.encoder.enc.1.encoders.15.self_attn.linear_out",
    "model.layers.26.block_sparse_moe.experts.5.w1",
    "model.layers.7.self_attn.o_proj",
    "model.layers.10.block_sparse_moe.experts.2.w1",
    "model.layers.29.block_sparse_moe.experts.5.w2",
    "model.layers.17.block_sparse_moe.experts.4.w2",
    "model.layers.5.block_sparse_moe.gate",
    "model.layers.26.block_sparse_moe.gate",
    "model.layers.1.block_sparse_moe.experts.4.w1",
    "model.layers.17.block_sparse_moe.experts.6.w2",
    "model.layers.1.block_sparse_moe.experts.3.w2",
    "model.layers.15.block_sparse_moe.experts.7.w3",
    "model.layers.30.block_sparse_moe.experts.2.w1",
    "model.layers.31.block_sparse_moe.experts.1.w3",
    "model.layers.0.block_sparse_moe.experts.2.w1",
    "model.layers.8.block_sparse_moe.experts.0.w1",
    "model.layers.10.self_attn.k_proj",
    "model.layers.24.block_sparse_moe.experts.2.w2",
    "model.layers.12.block_sparse_moe.gate",
    "model.layers.23.block_sparse_moe.experts.4.w1",
    "model.layers.0.block_sparse_moe.experts.4.w1",
    "model.layers.24.block_sparse_moe.experts.1.w2",
    "model.layers.6.block_sparse_moe.experts.7.w2",
    "model.layers.30.block_sparse_moe.experts.0.w2",
    "model.layers.22.block_sparse_moe.experts.0.w3",
    "model.layers.24.block_sparse_moe.experts.4.w1",
    "model.layers.26.block_sparse_moe.experts.4.w2",
    "model.layers.9.block_sparse_moe.experts.2.w1",
    "model.layers.22.block_sparse_moe.experts.2.w3",
    "model.layers.24.block_sparse_moe.experts.2.w1",
    "model.layers.8.block_sparse_moe.experts.2.w2",
    "model.layers.28.block_sparse_moe.experts.0.w3",
    "model.layers.29.block_sparse_moe.experts.5.w1",
    "model.layers.29.block_sparse_moe.experts.5.w3",
    "model.audio_encoder.encoder.enc.1.encoders.22.self_attn.linear_q",
    "model.layers.28.block_sparse_moe.experts.1.w2",
    "model.layers.0.self_attn.k_proj",
    "model.layers.23.block_sparse_moe.experts.5.w3",
    "model.layers.19.block_sparse_moe.experts.4.w3",
    "model.layers.2.self_attn.o_proj",
    "model.layers.25.self_attn.v_proj",
    "model.audio_encoder.encoder.enc.1.encoders.0.self_attn.linear_q",
    "model.audio_encoder.encoder.enc.1.encoders.11.self_attn.linear_v",
    "model.layers.27.block_sparse_moe.experts.3.w3",
    "model.layers.22.block_sparse_moe.experts.5.w3",
    "model.layers.8.block_sparse_moe.experts.7.w1",
    "model.layers.4.block_sparse_moe.experts.5.w1",
    "model.audio_encoder.encoder.enc.1.encoders.17.feed_forward.w_2",
    "model.layers.9.self_attn.q_proj",
    "model.layers.13.block_sparse_moe.gate",
    "model.layers.15.block_sparse_moe.experts.3.w2",
    "model.layers.0.block_sparse_moe.experts.1.w1",
    "model.layers.27.block_sparse_moe.experts.0.w3",
    "model.audio_encoder.encoder.enc.1.encoders.23.self_attn.linear_q",
    "model.audio_encoder.encoder.enc.1.encoders.10.self_attn.linear_pos",
    "model.layers.12.block_sparse_moe.experts.5.w1",
    "model.audio_encoder.encoder.enc.1.encoders.2.feed_forward.w_2",
    "model.layers.8.block_sparse_moe.experts.1.w1",
    "model.layers.7.block_sparse_moe.experts.1.w2",
    "model.layers.19.block_sparse_moe.experts.5.w2",
    "model.layers.8.block_sparse_moe.gate",
    "model.layers.15.block_sparse_moe.experts.5.w1",
    "model.layers.22.self_attn.v_proj",
    "model.layers.23.block_sparse_moe.gate",
    "model.layers.4.block_sparse_moe.experts.3.w2",
    "model.audio_encoder.encoder.enc.1.encoders.20.self_attn.linear_k",
    "model.layers.13.block_sparse_moe.experts.2.w3",
    "model.layers.15.block_sparse_moe.experts.3.w1",
    "model.layers.21.block_sparse_moe.experts.1.w2",
    "model.audio_encoder.encoder.enc.1.encoders.21.feed_forward.w_2",
    "model.audio_encoder.encoder.enc.1.encoders.16.self_attn.linear_out",
    "model.audio_encoder.encoder.enc.1.encoders.8.feed_forward.w_2",
    "model.layers.1.block_sparse_moe.experts.3.w1",
    "model.layers.1.block_sparse_moe.experts.7.w3",
    "model.layers.26.block_sparse_moe.experts.3.w2",
    "model.layers.31.block_sparse_moe.experts.7.w2",
    "model.audio_encoder.encoder.enc.1.encoders.19.self_attn.linear_v",
    "model.layers.1.block_sparse_moe.experts.6.w3",
    "model.layers.6.block_sparse_moe.experts.1.w1",
    "model.layers.18.block_sparse_moe.experts.5.w1",
    "model.layers.3.self_attn.k_proj",
    "model.layers.2.block_sparse_moe.experts.4.w1",
    "model.layers.30.block_sparse_moe.experts.4.w1",
    "model.layers.30.self_attn.v_proj",
    "model.layers.20.self_attn.o_proj",
    "model.layers.22.block_sparse_moe.experts.6.w3",
    "model.layers.0.block_sparse_moe.experts.4.w3",
    "model.layers.15.block_sparse_moe.experts.6.w3",
    "model.layers.24.block_sparse_moe.experts.3.w1",
    "model.layers.2.self_attn.q_proj",
    "model.layers.0.block_sparse_moe.experts.5.w3",
    "model.layers.10.block_sparse_moe.gate",
    "model.layers.14.block_sparse_moe.experts.5.w3",
    "model.layers.19.block_sparse_moe.experts.7.w2",
    "model.layers.28.block_sparse_moe.experts.6.w1",
    "model.layers.27.block_sparse_moe.experts.6.w1",
    "model.audio_encoder.encoder.enc.1.encoders.10.self_attn.linear_out",
    "model.layers.31.block_sparse_moe.experts.6.w1",
    "model.layers.17.block_sparse_moe.experts.0.w1",
    "model.layers.28.block_sparse_moe.experts.7.w2",
    "model.layers.16.block_sparse_moe.experts.6.w1",
    "model.mm_projector.0",
    "model.layers.3.block_sparse_moe.experts.1.w3",
    "model.layers.0.block_sparse_moe.experts.7.w1",
    "model.layers.13.block_sparse_moe.experts.3.w1",
    "model.layers.9.block_sparse_moe.experts.3.w3",
    "model.layers.21.self_attn.q_proj",
    "model.layers.27.block_sparse_moe.experts.3.w2",
    "model.layers.29.block_sparse_moe.experts.2.w2",
    "model.audio_encoder.encoder.enc.1.encoders.20.self_attn.linear_v",
    "model.layers.1.block_sparse_moe.experts.1.w3",
    "model.layers.14.block_sparse_moe.experts.2.w3",
    "model.layers.15.block_sparse_moe.experts.0.w3",
    "model.audio_encoder.encoder.enc.1.encoders.14.self_attn.linear_out",
    "model.layers.25.block_sparse_moe.experts.5.w2",
    "model.layers.15.block_sparse_moe.experts.1.w1",
    "model.layers.12.block_sparse_moe.experts.1.w2",
    "model.layers.20.block_sparse_moe.experts.1.w1",
    "model.layers.22.block_sparse_moe.experts.7.w3",
    "model.layers.2.block_sparse_moe.experts.4.w3",
    "model.layers.26.block_sparse_moe.experts.7.w1",
    "model.layers.18.block_sparse_moe.experts.3.w2",
    "model.layers.12.block_sparse_moe.experts.4.w1",
    "model.audio_encoder.encoder.enc.1.encoders.5.feed_forward.w_1",
    "model.layers.17.block_sparse_moe.experts.3.w2",
    "model.layers.21.self_attn.k_proj",
    "model.audio_encoder.encoder.enc.1.encoders.15.self_attn.linear_v",
    "model.layers.15.block_sparse_moe.experts.0.w1",
    "model.layers.20.block_sparse_moe.experts.7.w3",
    "model.layers.3.block_sparse_moe.experts.7.w2",
    "model.layers.27.block_sparse_moe.experts.2.w1",
    "model.layers.14.block_sparse_moe.experts.4.w3",
    "model.layers.17.block_sparse_moe.experts.5.w3",
    "model.layers.10.block_sparse_moe.experts.5.w2",
    "model.layers.7.block_sparse_moe.experts.7.w2",
    "model.layers.5.block_sparse_moe.experts.4.w3",
    "model.layers.5.block_sparse_moe.experts.5.w2",
    "model.layers.7.block_sparse_moe.experts.3.w3",
    "model.layers.10.block_sparse_moe.experts.5.w3",
    "model.layers.10.block_sparse_moe.experts.5.w1",
    "model.layers.14.block_sparse_moe.experts.1.w2",
    "model.layers.25.block_sparse_moe.experts.6.w1",
    "model.audio_encoder.encoder.enc.1.encoders.3.feed_forward.w_2",
    "model.audio_encoder.encoder.enc.1.encoders.13.feed_forward.w_1",
    "model.layers.10.block_sparse_moe.experts.6.w2",
    "model.layers.9.block_sparse_moe.experts.5.w1",
    "model.layers.8.block_sparse_moe.experts.6.w2",
    "model.audio_encoder.encoder.enc.1.encoders.11.self_attn.linear_q",
    "model.layers.24.block_sparse_moe.experts.7.w2",
    "model.layers.17.block_sparse_moe.experts.0.w2",
    "model.layers.22.block_sparse_moe.experts.7.w1",
    "model.layers.22.block_sparse_moe.experts.5.w1",
    "model.layers.3.block_sparse_moe.experts.5.w2",
    "model.layers.1.block_sparse_moe.experts.5.w1",
    "model.layers.11.block_sparse_moe.experts.2.w3",
    "model.layers.24.self_attn.o_proj",
    "model.layers.14.block_sparse_moe.experts.0.w3",
    "model.layers.18.block_sparse_moe.experts.5.w2",
    "model.layers.1.block_sparse_moe.experts.2.w3",
    "model.layers.28.block_sparse_moe.gate",
    "model.layers.20.block_sparse_moe.experts.3.w1",
    "model.layers.11.block_sparse_moe.experts.6.w2",
    "model.layers.6.block_sparse_moe.experts.7.w1",
    "model.layers.6.block_sparse_moe.experts.1.w2",
    "model.layers.20.block_sparse_moe.experts.5.w1",
    "model.audio_encoder.encoder.enc.1.encoders.1.self_attn.linear_v",
    "model.audio_encoder.encoder.enc.1.encoders.13.feed_forward.w_2",
    "model.layers.22.block_sparse_moe.experts.3.w1",
    "model.layers.21.block_sparse_moe.experts.5.w1",
    "model.layers.28.block_sparse_moe.experts.4.w1",
    "model.audio_encoder.encoder.enc.1.encoders.1.self_attn.linear_q",
    "model.layers.23.self_attn.q_proj",
    "model.audio_encoder.encoder.enc.1.encoders.2.self_attn.linear_q",
    "model.layers.31.block_sparse_moe.experts.3.w1",
    "model.layers.31.block_sparse_moe.experts.3.w2",
    "model.layers.4.block_sparse_moe.experts.5.w2",
    "model.layers.18.block_sparse_moe.experts.6.w2",
    "model.layers.27.block_sparse_moe.experts.2.w2",
    "model.layers.1.block_sparse_moe.experts.5.w2",
    "model.layers.12.block_sparse_moe.experts.2.w2",
    "model.layers.1.block_sparse_moe.experts.7.w1",
    "model.layers.25.self_attn.k_proj",
    "model.layers.9.self_attn.k_proj",
    "model.layers.23.block_sparse_moe.experts.0.w3",
    "model.layers.28.block_sparse_moe.experts.7.w1",
    "model.layers.24.block_sparse_moe.experts.5.w2",
    "model.layers.9.block_sparse_moe.experts.6.w1",
    "model.layers.18.self_attn.q_proj",
    "model.audio_encoder.encoder.enc.1.encoders.15.feed_forward.w_1",
    "model.layers.31.self_attn.o_proj",
    "model.layers.26.block_sparse_moe.experts.4.w1",
    "model.layers.8.block_sparse_moe.experts.7.w3",
    "model.layers.24.block_sparse_moe.experts.7.w1",
    "model.layers.21.block_sparse_moe.experts.6.w1",
    "model.layers.19.block_sparse_moe.experts.2.w2",
    "model.layers.19.block_sparse_moe.experts.6.w2",
    "model.layers.28.self_attn.k_proj",
    "model.audio_encoder.encoder.enc.1.encoders.15.feed_forward.w_2",
    "model.layers.6.block_sparse_moe.gate",
    "model.audio_encoder.encoder.enc.1.encoders.16.feed_forward.w_1",
    "model.layers.9.block_sparse_moe.gate",
    "model.audio_encoder.encoder.enc.1.encoders.1.self_attn.linear_out",
    "model.layers.12.block_sparse_moe.experts.3.w1",
    "model.layers.16.block_sparse_moe.experts.1.w1",
    "model.layers.6.block_sparse_moe.experts.3.w2",
    "model.layers.31.block_sparse_moe.experts.2.w1",
    "model.audio_encoder.encoder.enc.1.encoders.10.self_attn.linear_q",
    "model.layers.0.block_sparse_moe.experts.2.w2",
    "model.layers.23.block_sparse_moe.experts.6.w1",
    "model.layers.18.block_sparse_moe.experts.1.w2",
    "model.layers.7.block_sparse_moe.experts.6.w2",
    "model.layers.5.block_sparse_moe.experts.6.w1",
    "model.layers.3.self_attn.v_proj",
    "model.layers.10.block_sparse_moe.experts.4.w3",
    "model.audio_encoder.encoder.enc.1.encoders.12.feed_forward.w_2",
    "model.layers.22.self_attn.k_proj",
    "model.layers.7.block_sparse_moe.experts.0.w1",
    "model.layers.12.block_sparse_moe.experts.2.w3",
    "model.layers.4.block_sparse_moe.experts.3.w1",
    "model.audio_encoder.encoder.enc.1.encoders.9.self_attn.linear_pos",
    "model.layers.26.block_sparse_moe.experts.1.w2",
    "model.layers.25.block_sparse_moe.experts.1.w2",
    "model.layers.29.self_attn.o_proj",
    "model.layers.6.self_attn.k_proj",
    "model.layers.8.block_sparse_moe.experts.3.w1",
    "model.layers.12.block_sparse_moe.experts.0.w3",
    "model.layers.12.block_sparse_moe.experts.3.w2",
    "model.layers.11.self_attn.o_proj",
    "model.layers.10.block_sparse_moe.experts.6.w1",
    "model.layers.5.block_sparse_moe.experts.4.w1",
    "model.layers.19.block_sparse_moe.experts.7.w1",
    "model.layers.25.block_sparse_moe.experts.5.w3",
    "model.layers.31.block_sparse_moe.experts.1.w1",
    "model.layers.16.self_attn.k_proj",
    "model.layers.26.block_sparse_moe.experts.2.w1",
    "model.layers.2.block_sparse_moe.experts.5.w2",
    "model.layers.0.block_sparse_moe.experts.5.w1",
    "model.layers.9.block_sparse_moe.experts.0.w1",
    "model.audio_encoder.encoder.enc.1.encoders.5.self_attn.linear_pos",
    "model.audio_encoder.encoder.enc.1.encoders.0.self_attn.linear_v",
    "model.layers.10.block_sparse_moe.experts.1.w2",
    "model.layers.1.block_sparse_moe.experts.3.w3",
    "model.layers.13.block_sparse_moe.experts.7.w2",
    "model.layers.27.block_sparse_moe.experts.0.w2",
    "model.layers.21.block_sparse_moe.experts.6.w3",
    "model.layers.11.block_sparse_moe.experts.1.w2",
    "model.layers.20.self_attn.q_proj",
    "model.layers.21.block_sparse_moe.experts.7.w2",
    "model.layers.30.block_sparse_moe.experts.6.w3",
    "model.audio_encoder.encoder.enc.1.encoders.7.self_attn.linear_v",
    "model.audio_encoder.encoder.enc.1.encoders.22.feed_forward.w_2",
    "model.audio_encoder.encoder.enc.1.encoders.10.self_attn.linear_v",
    "model.layers.16.block_sparse_moe.experts.4.w3",
    "model.audio_encoder.encoder.enc.1.encoders.2.self_attn.linear_pos",
    "model.layers.0.block_sparse_moe.experts.7.w2",
    "model.audio_encoder.encoder.enc.1.encoders.21.feed_forward.w_1",
    "model.layers.19.self_attn.q_proj",
    "model.layers.13.block_sparse_moe.experts.5.w1",
    "model.layers.12.block_sparse_moe.experts.7.w1",
    "model.layers.8.block_sparse_moe.experts.3.w2",
    "model.layers.4.block_sparse_moe.experts.7.w3",
    "model.layers.14.block_sparse_moe.experts.6.w2",
    "model.layers.15.block_sparse_moe.gate",
    "model.layers.23.block_sparse_moe.experts.3.w3",
    "model.layers.30.self_attn.o_proj",
    "model.layers.5.block_sparse_moe.experts.2.w1",
    "model.layers.11.block_sparse_moe.experts.3.w1",
    "model.layers.30.block_sparse_moe.experts.1.w1",
    "model.audio_encoder.encoder.enc.1.encoders.21.self_attn.linear_pos",
    "model.layers.27.block_sparse_moe.experts.5.w2",
    "model.layers.8.block_sparse_moe.experts.4.w3",
    "model.layers.18.self_attn.v_proj",
    "model.layers.22.block_sparse_moe.experts.6.w1",
    "model.layers.19.block_sparse_moe.experts.2.w3",
    "model.layers.28.block_sparse_moe.experts.1.w3",
    "model.audio_encoder.encoder.enc.1.encoders.12.self_attn.linear_v",
    "model.layers.24.self_attn.v_proj",
    "model.layers.23.block_sparse_moe.experts.2.w1",
    "model.layers.25.self_attn.q_proj",
    "model.layers.28.block_sparse_moe.experts.2.w3",
    "model.layers.31.block_sparse_moe.experts.5.w2",
    "model.layers.14.self_attn.o_proj",
    "model.audio_encoder.encoder.enc.1.encoders.7.self_attn.linear_out",
    "model.layers.16.block_sparse_moe.experts.1.w2",
    "model.layers.25.block_sparse_moe.experts.3.w1",
    "model.layers.5.block_sparse_moe.experts.1.w2",
    "model.audio_encoder.encoder.enc.1.encoders.16.self_attn.linear_q",
    "model.layers.4.block_sparse_moe.experts.1.w1",
    "model.layers.14.block_sparse_moe.experts.2.w1",
    "model.layers.0.block_sparse_moe.experts.1.w2",
    "model.layers.23.block_sparse_moe.experts.3.w2",
    "model.layers.30.block_sparse_moe.experts.1.w2",
    "model.layers.27.block_sparse_moe.experts.7.w1",
    "model.audio_encoder.encoder.enc.1.encoders.8.self_attn.linear_out",
    "model.layers.21.block_sparse_moe.experts.0.w1",
    "model.audio_encoder.encoder.enc.1.encoders.12.feed_forward.w_1",
    "model.layers.14.block_sparse_moe.experts.6.w1",
    "model.layers.16.block_sparse_moe.experts.3.w1",
    "model.audio_encoder.encoder.enc.1.encoders.13.self_attn.linear_q",
    "model.layers.27.block_sparse_moe.gate",
    "model.layers.5.self_attn.q_proj",
    "model.layers.15.block_sparse_moe.experts.2.w1",
    "model.layers.7.block_sparse_moe.experts.6.w3",
    "model.layers.19.block_sparse_moe.experts.5.w3",
    "model.layers.0.block_sparse_moe.experts.6.w2",
    "model.layers.7.block_sparse_moe.experts.3.w1",
    "model.layers.8.block_sparse_moe.experts.1.w2",
    "model.layers.13.block_sparse_moe.experts.2.w2",
    "model.layers.22.block_sparse_moe.experts.6.w2",
    "model.audio_encoder.encoder.enc.1.encoders.11.feed_forward.w_2",
    "model.layers.6.block_sparse_moe.experts.0.w2",
    "model.audio_encoder.encoder.enc.1.encoders.6.self_attn.linear_v",
    "model.layers.7.block_sparse_moe.experts.7.w3",
    "model.audio_encoder.encoder.enc.1.encoders.18.self_attn.linear_out",
    "model.audio_encoder.encoder.enc.1.encoders.19.self_attn.linear_k",
    "model.layers.2.block_sparse_moe.experts.3.w1",
    "model.layers.24.block_sparse_moe.experts.6.w3",
    "model.layers.9.block_sparse_moe.experts.6.w3",
    "model.layers.21.block_sparse_moe.experts.1.w3",
    "model.layers.19.block_sparse_moe.experts.4.w2",
    "model.layers.8.block_sparse_moe.experts.5.w3",
    "model.audio_encoder.encoder.enc.1.encoders.2.feed_forward.w_1",
    "model.layers.12.block_sparse_moe.experts.1.w1",
    "model.audio_encoder.encoder.enc.1.encoders.12.self_attn.linear_k",
    "model.layers.23.block_sparse_moe.experts.1.w2",
    "model.layers.0.self_attn.o_proj",
    "model.layers.16.block_sparse_moe.experts.5.w3",
    "model.layers.18.block_sparse_moe.experts.4.w3",
    "model.layers.26.block_sparse_moe.experts.6.w3",
    "model.audio_encoder.encoder.enc.1.encoders.9.self_attn.linear_q",
    "model.layers.21.block_sparse_moe.experts.3.w1",
    "model.layers.11.block_sparse_moe.experts.0.w1",
    "model.layers.3.block_sparse_moe.experts.1.w1",
    "model.audio_encoder.encoder.enc.1.encoders.8.feed_forward.w_1",
    "model.layers.5.block_sparse_moe.experts.2.w2",
    "model.audio_encoder.encoder.enc.1.encoders.2.self_attn.linear_k",
    "model.layers.30.self_attn.k_proj",
    "model.layers.4.block_sparse_moe.experts.0.w3",
    "model.layers.14.block_sparse_moe.experts.3.w1",
    "model.layers.8.self_attn.v_proj",
    "model.layers.17.block_sparse_moe.experts.5.w2",
    "model.layers.22.block_sparse_moe.experts.0.w2",
    "model.layers.25.block_sparse_moe.experts.0.w3",
    "model.layers.16.block_sparse_moe.experts.5.w1",
    "model.layers.19.block_sparse_moe.experts.6.w1",
    "model.layers.11.block_sparse_moe.experts.4.w1",
    "model.layers.10.block_sparse_moe.experts.6.w3",
    "model.audio_encoder.encoder.enc.1.encoders.3.self_attn.linear_v",
    "model.layers.28.self_attn.v_proj",
    "model.layers.1.self_attn.k_proj",
    "model.layers.14.block_sparse_moe.experts.7.w2",
    "model.layers.24.block_sparse_moe.experts.7.w3",
    "model.layers.0.block_sparse_moe.experts.6.w1",
    "model.layers.10.self_attn.o_proj",
    "model.layers.21.block_sparse_moe.experts.4.w3",
    "model.layers.25.block_sparse_moe.experts.2.w2",
    "model.layers.31.block_sparse_moe.experts.7.w1",
    "model.layers.14.block_sparse_moe.experts.0.w1",
    "model.layers.31.self_attn.q_proj",
    "model.audio_encoder.encoder.enc.1.encoders.6.feed_forward.w_2",
    "model.audio_encoder.encoder.enc.1.encoders.10.feed_forward.w_1",
    "model.layers.11.block_sparse_moe.experts.6.w1",
    "model.layers.28.block_sparse_moe.experts.2.w2",
    "model.layers.12.block_sparse_moe.experts.2.w1",
    "model.layers.12.self_attn.o_proj",
    "model.layers.26.block_sparse_moe.experts.5.w3",
    "model.layers.20.self_attn.v_proj",
    "model.layers.10.block_sparse_moe.experts.1.w3",
    "model.audio_encoder.encoder.enc.1.encoders.13.self_attn.linear_out",
    "model.audio_encoder.encoder.enc.1.encoders.19.feed_forward.w_1",
    "model.audio_encoder.encoder.enc.1.encoders.3.self_attn.linear_out",
    "model.layers.0.block_sparse_moe.experts.2.w3",
    "model.layers.5.self_attn.o_proj",
    "model.audio_encoder.encoder.enc.1.encoders.3.self_attn.linear_q",
    "model.layers.8.block_sparse_moe.experts.6.w1",
    "model.layers.28.block_sparse_moe.experts.1.w1",
    "model.layers.29.block_sparse_moe.experts.7.w3",
    "model.layers.12.block_sparse_moe.experts.0.w1",
    "model.layers.6.block_sparse_moe.experts.7.w3",
    "model.layers.27.block_sparse_moe.experts.7.w2",
    "model.audio_encoder.encoder.enc.1.encoders.0.feed_forward.w_1",
    "model.layers.9.block_sparse_moe.experts.1.w3",
    "model.layers.9.block_sparse_moe.experts.6.w2",
    "model.audio_encoder.encoder.enc.1.encoders.15.self_attn.linear_pos",
    "model.layers.25.block_sparse_moe.experts.1.w3",
    "model.layers.16.block_sparse_moe.experts.2.w3",
    "model.layers.18.block_sparse_moe.experts.2.w1",
    "model.layers.18.block_sparse_moe.experts.3.w3",
    "model.layers.18.block_sparse_moe.experts.7.w1",
    "model.layers.17.block_sparse_moe.experts.7.w3",
    "model.audio_encoder.encoder.enc.1.encoders.17.self_attn.linear_k",
    "model.layers.29.block_sparse_moe.experts.6.w3",
    "model.audio_encoder.encoder.enc.1.encoders.0.self_attn.linear_out",
    "model.layers.27.block_sparse_moe.experts.6.w2",
    "model.layers.1.block_sparse_moe.experts.7.w2",
    "model.layers.6.block_sparse_moe.experts.4.w2",
    "model.layers.2.self_attn.k_proj",
    "model.layers.4.self_attn.k_proj",
    "model.layers.23.block_sparse_moe.experts.0.w2",
    "model.audio_encoder.encoder.enc.1.encoders.17.self_attn.linear_q",
    "model.audio_encoder.encoder.enc.1.encoders.21.self_attn.linear_k",
    "model.layers.1.self_attn.o_proj",
    "model.layers.5.block_sparse_moe.experts.3.w1",
    "model.layers.17.block_sparse_moe.experts.3.w1",
    "model.layers.3.block_sparse_moe.experts.7.w1",
    "model.layers.27.block_sparse_moe.experts.3.w1",
    "model.layers.1.block_sparse_moe.experts.6.w1",
    "model.layers.2.block_sparse_moe.experts.6.w1",
    "model.audio_encoder.encoder.enc.1.encoders.0.self_attn.linear_pos",
    "model.layers.14.self_attn.v_proj",
    "model.layers.16.block_sparse_moe.experts.6.w2",
    "model.layers.28.self_attn.q_proj",
    "model.layers.17.self_attn.k_proj",
    "model.layers.29.block_sparse_moe.experts.7.w2",
    "model.layers.31.block_sparse_moe.experts.0.w2",
    "model.layers.30.block_sparse_moe.experts.6.w1",
    "model.layers.11.block_sparse_moe.experts.6.w3",
    "model.layers.11.block_sparse_moe.experts.1.w1",
    "model.layers.15.block_sparse_moe.experts.6.w1",
    "model.layers.18.block_sparse_moe.experts.2.w2",
    "model.layers.13.block_sparse_moe.experts.6.w2",
    "model.layers.13.block_sparse_moe.experts.0.w3",
    "model.layers.14.block_sparse_moe.experts.7.w3",
    "model.layers.6.block_sparse_moe.experts.6.w2",
    "model.layers.31.self_attn.k_proj",
    "model.audio_encoder.encoder.enc.1.encoders.6.self_attn.linear_q",
    "model.layers.0.block_sparse_moe.experts.4.w2",
    "model.layers.8.block_sparse_moe.experts.4.w2",
    "model.layers.23.block_sparse_moe.experts.1.w1",
    "model.audio_encoder.encoder.enc.1.encoders.8.self_attn.linear_v",
    "model.layers.13.block_sparse_moe.experts.5.w3",
    "model.layers.9.self_attn.v_proj",
    "model.layers.27.self_attn.q_proj",
    "model.layers.20.block_sparse_moe.gate",
    "model.layers.8.block_sparse_moe.experts.2.w3",
    "model.layers.23.block_sparse_moe.experts.2.w3",
    "model.layers.30.block_sparse_moe.experts.0.w3",
    "model.audio_encoder.encoder.enc.1.encoders.17.self_attn.linear_v",
    "model.layers.24.self_attn.k_proj",
    "model.layers.17.block_sparse_moe.experts.2.w1",
    "model.layers.15.block_sparse_moe.experts.4.w3",
    "model.layers.6.self_attn.q_proj",
    "model.layers.2.block_sparse_moe.experts.3.w2",
    "model.layers.10.block_sparse_moe.experts.4.w2",
    "model.layers.13.block_sparse_moe.experts.1.w3",
    "model.layers.3.block_sparse_moe.experts.6.w3",
    "model.layers.30.block_sparse_moe.experts.7.w1",
    "model.layers.15.block_sparse_moe.experts.4.w2",
    "model.layers.19.block_sparse_moe.experts.0.w3",
    "model.layers.31.block_sparse_moe.experts.0.w3",
    "model.layers.4.self_attn.q_proj",
    "model.audio_encoder.encoder.enc.1.encoders.14.self_attn.linear_q",
    "model.layers.8.block_sparse_moe.experts.5.w1",
    "model.layers.19.block_sparse_moe.experts.6.w3",
    "model.layers.10.block_sparse_moe.experts.7.w1",
    "model.layers.28.block_sparse_moe.experts.3.w3",
    "model.layers.5.block_sparse_moe.experts.0.w3",
    "model.layers.31.block_sparse_moe.experts.5.w1",
    "model.audio_encoder.encoder.enc.1.encoders.17.feed_forward.w_1",
    "model.layers.1.block_sparse_moe.experts.0.w2",
    "model.layers.6.self_attn.o_proj",
    "model.layers.30.block_sparse_moe.experts.7.w3",
    "model.layers.2.block_sparse_moe.experts.0.w1",
    "model.layers.24.block_sparse_moe.experts.0.w1",
    "model.layers.0.block_sparse_moe.experts.6.w3",
    "model.layers.5.self_attn.k_proj",
    "model.layers.22.block_sparse_moe.experts.4.w1",
    "model.layers.23.block_sparse_moe.experts.7.w2",
    "model.audio_encoder.encoder.enc.1.encoders.20.self_attn.linear_out",
    "model.layers.20.block_sparse_moe.experts.0.w1",
    "model.layers.3.block_sparse_moe.experts.2.w3",
    "model.layers.3.block_sparse_moe.experts.5.w1",
    "model.layers.21.block_sparse_moe.experts.1.w1",
    "model.layers.3.block_sparse_moe.experts.6.w2",
    "model.layers.14.block_sparse_moe.experts.5.w1",
    "model.audio_encoder.encoder.enc.1.encoders.2.self_attn.linear_v",
    "model.audio_encoder.encoder.enc.1.encoders.6.self_attn.linear_k",
    "model.layers.20.block_sparse_moe.experts.3.w3",
    "model.layers.6.block_sparse_moe.experts.4.w1",
    "model.audio_encoder.adpter.project",
    "model.layers.5.block_sparse_moe.experts.6.w2",
    "model.layers.17.block_sparse_moe.experts.0.w3",
    "model.layers.12.block_sparse_moe.experts.6.w3",
    "model.layers.11.block_sparse_moe.experts.7.w3",
    "model.layers.31.self_attn.v_proj",
    "model.layers.31.block_sparse_moe.experts.4.w1",
    "model.layers.28.block_sparse_moe.experts.4.w3",
    "model.layers.29.block_sparse_moe.experts.7.w1",
    "model.layers.15.block_sparse_moe.experts.2.w2",
    "model.layers.14.self_attn.q_proj",
    "model.layers.15.block_sparse_moe.experts.0.w2",
    "model.layers.21.self_attn.o_proj",
    "model.audio_encoder.encoder.enc.1.encoders.22.feed_forward.w_1",
    "model.layers.14.block_sparse_moe.experts.3.w2",
    "model.layers.3.self_attn.o_proj",
    "model.layers.13.block_sparse_moe.experts.1.w2",
    "model.layers.10.block_sparse_moe.experts.0.w3",
    "model.layers.11.block_sparse_moe.experts.3.w2",
    "model.layers.2.block_sparse_moe.experts.6.w3",
    "model.layers.3.block_sparse_moe.experts.7.w3",
    "model.layers.24.block_sparse_moe.experts.1.w3",
    "model.audio_encoder.encoder.enc.1.encoders.6.self_attn.linear_pos",
    "model.layers.16.block_sparse_moe.experts.0.w2",
    "model.layers.20.block_sparse_moe.experts.4.w2",
    "model.layers.3.block_sparse_moe.experts.6.w1",
    "model.layers.13.block_sparse_moe.experts.6.w1",
    "model.layers.28.block_sparse_moe.experts.5.w3",
    "model.layers.21.block_sparse_moe.experts.6.w2",
    "model.audio_encoder.encoder.enc.1.encoders.4.feed_forward.w_2",
    "model.layers.29.block_sparse_moe.experts.0.w1",
    "model.layers.28.block_sparse_moe.experts.3.w2",
    "model.layers.19.block_sparse_moe.experts.5.w1",
    "model.layers.8.block_sparse_moe.experts.4.w1",
    "model.layers.24.block_sparse_moe.experts.4.w2",
    "model.layers.1.block_sparse_moe.experts.1.w1",
    "model.layers.0.block_sparse_moe.experts.7.w3",
    "model.layers.10.block_sparse_moe.experts.7.w3",
    "model.layers.3.block_sparse_moe.experts.0.w3",
    "model.layers.15.block_sparse_moe.experts.2.w3",
    "model.layers.10.block_sparse_moe.experts.1.w1",
    "model.layers.9.block_sparse_moe.experts.7.w2",
    "model.layers.19.block_sparse_moe.experts.0.w2",
    "model.layers.10.block_sparse_moe.experts.4.w1",
    "model.audio_encoder.encoder.enc.1.encoders.1.self_attn.linear_pos",
    "model.layers.4.block_sparse_moe.experts.6.w2",
    "model.layers.0.block_sparse_moe.gate",
    "model.layers.22.block_sparse_moe.experts.1.w3",
    "model.layers.5.block_sparse_moe.experts.0.w2",
    "model.layers.29.block_sparse_moe.experts.2.w1",
    "model.layers.31.block_sparse_moe.experts.2.w3",
    "model.audio_encoder.encoder.enc.1.encoders.22.self_attn.linear_v",
    "model.audio_encoder.encoder.enc.1.encoders.20.self_attn.linear_q",
    "model.audio_encoder.encoder.enc.1.encoders.19.feed_forward.w_2",
    "model.layers.10.block_sparse_moe.experts.3.w1",
    "model.layers.20.block_sparse_moe.experts.2.w3",
    "model.layers.2.block_sparse_moe.experts.1.w2",
    "model.audio_encoder.encoder.enc.1.encoders.7.self_attn.linear_k",
    "model.layers.22.block_sparse_moe.experts.3.w3",
    "model.layers.23.self_attn.k_proj",
    "model.layers.27.block_sparse_moe.experts.7.w3",
    "model.audio_encoder.encoder.enc.1.encoders.23.feed_forward.w_2",
    "model.layers.30.block_sparse_moe.experts.5.w3",
    "model.layers.23.block_sparse_moe.experts.4.w2",
    "model.layers.4.self_attn.v_proj",
    "model.layers.1.self_attn.q_proj",
    "model.layers.13.block_sparse_moe.experts.7.w3",
    "model.layers.2.block_sparse_moe.experts.3.w3",
    "model.layers.7.block_sparse_moe.gate",
    "model.audio_encoder.encoder.enc.1.encoders.14.self_attn.linear_k",
    "model.layers.11.block_sparse_moe.experts.0.w2",
    "model.audio_encoder.encoder.enc.1.encoders.16.self_attn.linear_k",
    "model.audio_encoder.encoder.enc.1.encoders.20.feed_forward.w_2",
    "model.layers.20.block_sparse_moe.experts.4.w1",
    "model.layers.9.block_sparse_moe.experts.2.w2",
    "model.layers.15.block_sparse_moe.experts.5.w3",
    "model.audio_encoder.encoder.enc.1.encoders.4.self_attn.linear_k",
    "model.layers.10.block_sparse_moe.experts.2.w3",
    "model.layers.14.block_sparse_moe.experts.2.w2",
    "model.audio_encoder.encoder.enc.1.encoders.0.feed_forward.w_2",
    "model.layers.3.block_sparse_moe.gate",
    "model.layers.12.self_attn.q_proj",
    "model.layers.26.block_sparse_moe.experts.3.w1",
    "model.layers.6.block_sparse_moe.experts.2.w1",
    "model.layers.7.block_sparse_moe.experts.2.w3",
    "model.layers.12.block_sparse_moe.experts.4.w3",
    "model.layers.9.block_sparse_moe.experts.1.w2",
    "model.layers.25.block_sparse_moe.experts.0.w1",
    "model.layers.26.block_sparse_moe.experts.2.w3",
    "model.audio_encoder.encoder.enc.1.encoders.2.self_attn.linear_out",
    "model.audio_encoder.encoder.enc.1.encoders.3.feed_forward.w_1",
    "model.layers.7.block_sparse_moe.experts.0.w3",
    "model.layers.27.self_attn.k_proj",
    "model.layers.31.block_sparse_moe.experts.5.w3",
    "model.layers.27.block_sparse_moe.experts.1.w2",
    "model.layers.28.block_sparse_moe.experts.6.w2",
    "model.layers.3.block_sparse_moe.experts.4.w2",
    "model.layers.3.block_sparse_moe.experts.4.w3",
    "model.audio_encoder.encoder.enc.1.encoders.7.self_attn.linear_pos",
    "model.layers.4.block_sparse_moe.experts.1.w3",
    "model.layers.7.block_sparse_moe.experts.1.w3",
    "model.layers.12.block_sparse_moe.experts.6.w1",
    "model.layers.24.block_sparse_moe.experts.0.w2",
    "model.layers.24.block_sparse_moe.experts.0.w3",
    "model.layers.29.block_sparse_moe.experts.3.w2",
    "model.layers.16.block_sparse_moe.experts.0.w3",
    "model.layers.11.block_sparse_moe.experts.7.w2",
    "model.audio_encoder.encoder.enc.1.encoders.11.self_attn.linear_pos",
    "model.layers.7.block_sparse_moe.experts.5.w1",
    "model.layers.22.block_sparse_moe.experts.0.w1",
    "model.layers.30.self_attn.q_proj",
    "model.layers.3.self_attn.q_proj",
    "model.audio_encoder.encoder.enc.1.encoders.11.feed_forward.w_1",
    "model.layers.16.block_sparse_moe.experts.4.w1",
    "model.audio_encoder.encoder.enc.1.encoders.9.self_attn.linear_out",
    "model.layers.5.block_sparse_moe.experts.0.w1",
    "model.layers.27.block_sparse_moe.experts.4.w2",
    "model.audio_encoder.encoder.enc.1.encoders.23.feed_forward.w_1",
    "model.audio_encoder.encoder.enc.1.encoders.18.self_attn.linear_k",
    "model.audio_encoder.encoder.enc.1.encoders.11.self_attn.linear_k",
    "model.layers.3.block_sparse_moe.experts.3.w3",
    "model.layers.22.block_sparse_moe.experts.2.w1",
    "model.layers.25.block_sparse_moe.experts.3.w2",
    "model.layers.24.block_sparse_moe.experts.5.w3",
    "model.layers.30.block_sparse_moe.gate",
    "model.layers.30.block_sparse_moe.experts.5.w2",
    "model.layers.18.block_sparse_moe.experts.7.w2",
    "model.layers.21.block_sparse_moe.experts.4.w1",
    "model.audio_encoder.encoder.enc.1.encoders.5.self_attn.linear_q",
    "model.layers.29.block_sparse_moe.experts.6.w2",
    "model.layers.10.block_sparse_moe.experts.3.w2",
    "model.layers.2.block_sparse_moe.experts.0.w2",
    "model.layers.15.self_attn.q_proj",
    "model.layers.25.block_sparse_moe.experts.2.w3",
    "model.layers.6.block_sparse_moe.experts.0.w1",
    "model.layers.16.block_sparse_moe.experts.7.w3",
    "model.audio_encoder.encoder.enc.1.encoders.6.self_attn.linear_out",
    "model.audio_encoder.encoder.enc.1.encoders.20.self_attn.linear_pos",
    "model.layers.21.block_sparse_moe.experts.0.w2",
    "model.layers.22.block_sparse_moe.experts.7.w2",
    "model.layers.23.block_sparse_moe.experts.7.w3",
    "model.layers.4.block_sparse_moe.experts.4.w3",
    "model.layers.3.block_sparse_moe.experts.1.w2",
    "model.layers.7.block_sparse_moe.experts.2.w2",
    "model.layers.22.self_attn.q_proj",
    "model.audio_encoder.encoder.enc.1.encoders.9.feed_forward.w_1",
    "model.audio_encoder.encoder.enc.1.encoders.7.feed_forward.w_1",
    "model.layers.10.block_sparse_moe.experts.7.w2",
    "model.layers.24.self_attn.q_proj",
    "model.layers.4.block_sparse_moe.experts.0.w1",
    "model.layers.14.self_attn.k_proj",
    "model.layers.30.block_sparse_moe.experts.2.w2",
    "model.layers.12.block_sparse_moe.experts.3.w3",
    "model.audio_encoder.encoder.enc.1.encoders.18.self_attn.linear_pos",
    "model.layers.20.block_sparse_moe.experts.3.w2",
    "model.layers.9.block_sparse_moe.experts.4.w2",
    "model.layers.15.block_sparse_moe.experts.7.w2",
    "model.layers.28.block_sparse_moe.experts.2.w1",
    "model.layers.6.block_sparse_moe.experts.4.w3",
    "model.layers.17.block_sparse_moe.experts.2.w3",
    "model.layers.29.block_sparse_moe.experts.4.w2",
    "model.layers.22.block_sparse_moe.gate",
    "model.layers.26.self_attn.v_proj",
    "model.audio_encoder.encoder.enc.1.encoders.4.self_attn.linear_pos",
    "model.layers.6.block_sparse_moe.experts.5.w3",
    "model.layers.10.self_attn.q_proj",
    "model.layers.25.block_sparse_moe.experts.4.w2",
    "model.layers.3.block_sparse_moe.experts.0.w1",
    "model.audio_encoder.encoder.enc.1.encoders.1.self_attn.linear_k",
    "model.layers.8.block_sparse_moe.experts.6.w3",
    "model.layers.24.block_sparse_moe.experts.3.w2",
    "model.layers.5.block_sparse_moe.experts.3.w3",
    "model.layers.1.block_sparse_moe.gate",
    "model.layers.4.block_sparse_moe.experts.0.w2",
    "model.layers.12.block_sparse_moe.experts.1.w3",
    "model.audio_encoder.encoder.enc.1.encoders.8.self_attn.linear_pos",
    "model.layers.5.block_sparse_moe.experts.1.w3",
    "model.layers.28.block_sparse_moe.experts.3.w1",
    "model.layers.11.self_attn.q_proj",
    "model.layers.19.block_sparse_moe.experts.1.w1",
    "model.layers.26.block_sparse_moe.experts.7.w2",
    "model.layers.8.block_sparse_moe.experts.3.w3",
    "model.layers.2.block_sparse_moe.experts.5.w3",
    "model.layers.5.block_sparse_moe.experts.3.w2",
    "model.layers.13.block_sparse_moe.experts.2.w1",
    "model.layers.15.self_attn.v_proj",
    "model.layers.21.block_sparse_moe.experts.7.w3",
    "model.layers.25.block_sparse_moe.experts.4.w1",
    "model.layers.16.self_attn.v_proj",
    "model.layers.17.self_attn.o_proj",
    "model.layers.29.block_sparse_moe.experts.0.w3",
    "model.layers.29.block_sparse_moe.experts.6.w1",
    "model.layers.16.self_attn.o_proj",
    "model.layers.3.block_sparse_moe.experts.3.w2",
    "model.layers.19.block_sparse_moe.experts.3.w2",
    "model.layers.6.block_sparse_moe.experts.2.w2",
    "model.layers.29.block_sparse_moe.experts.1.w1",
    "model.layers.5.block_sparse_moe.experts.7.w1",
    "model.layers.25.block_sparse_moe.experts.3.w3",
    "model.layers.11.self_attn.v_proj",
    "model.layers.7.block_sparse_moe.experts.4.w1",
    "model.layers.12.block_sparse_moe.experts.5.w3",
    "model.layers.17.block_sparse_moe.experts.3.w3",
    "model.layers.19.self_attn.v_proj",
    "model.layers.20.block_sparse_moe.experts.7.w1",
    "model.layers.2.block_sparse_moe.gate",
    "model.layers.28.block_sparse_moe.experts.6.w3",
    "model.layers.2.block_sparse_moe.experts.1.w1",
    "model.layers.16.block_sparse_moe.experts.1.w3",
    "model.layers.31.block_sparse_moe.experts.2.w2",
    "model.layers.29.block_sparse_moe.experts.4.w3",
    "model.layers.9.block_sparse_moe.experts.1.w1",
    "model.layers.26.block_sparse_moe.experts.5.w2",
    "model.layers.13.block_sparse_moe.experts.0.w1",
    "model.layers.13.block_sparse_moe.experts.4.w2",
    "model.layers.1.block_sparse_moe.experts.6.w2",
    "model.layers.9.block_sparse_moe.experts.0.w3",
    "model.layers.17.block_sparse_moe.experts.6.w3",
    "model.layers.30.block_sparse_moe.experts.3.w2",
    "model.layers.7.block_sparse_moe.experts.6.w1",
    "model.layers.25.block_sparse_moe.experts.1.w1",
    "model.layers.1.block_sparse_moe.experts.4.w3",
    "model.layers.23.self_attn.v_proj",
    "model.layers.15.self_attn.o_proj",
    "model.layers.21.block_sparse_moe.experts.2.w1",
    "model.layers.12.block_sparse_moe.experts.0.w2",
    "model.layers.29.block_sparse_moe.experts.0.w2",
    "model.layers.9.block_sparse_moe.experts.3.w2",
    "model.layers.17.block_sparse_moe.experts.5.w1",
    "model.layers.27.block_sparse_moe.experts.1.w3",
    "model.layers.29.block_sparse_moe.experts.1.w3",
    "model.layers.1.block_sparse_moe.experts.1.w2",
    "model.layers.31.block_sparse_moe.experts.7.w3",
    "model.layers.19.block_sparse_moe.experts.0.w1",
    "model.layers.18.block_sparse_moe.experts.0.w3",
    "model.audio_encoder.encoder.enc.1.encoders.7.feed_forward.w_2",
    "model.audio_encoder.encoder.enc.1.encoders.18.feed_forward.w_2",
    "model.layers.0.block_sparse_moe.experts.0.w2",
    "model.layers.6.block_sparse_moe.experts.5.w2",
    "model.layers.12.block_sparse_moe.experts.4.w2",
    "model.audio_encoder.encoder.enc.1.encoders.18.feed_forward.w_1",
    "model.audio_encoder.encoder.enc.1.encoders.19.self_attn.linear_pos",
    "model.layers.23.block_sparse_moe.experts.6.w3",
    "model.layers.7.block_sparse_moe.experts.4.w2",
    "model.layers.26.block_sparse_moe.experts.1.w3",
    "model.layers.27.block_sparse_moe.experts.5.w3",
    "model.audio_encoder.encoder.enc.1.encoders.23.self_attn.linear_out",
    "model.layers.27.block_sparse_moe.experts.1.w1",
    "model.layers.16.block_sparse_moe.experts.3.w3",
    "model.layers.21.block_sparse_moe.experts.3.w3",
    "model.audio_encoder.encoder.enc.1.encoders.13.self_attn.linear_v",
    "model.layers.10.block_sparse_moe.experts.0.w2",
    "model.layers.20.block_sparse_moe.experts.6.w1",
    "model.layers.21.block_sparse_moe.experts.2.w2",
    "model.layers.2.block_sparse_moe.experts.1.w3",
    "model.layers.26.self_attn.k_proj",
    "model.layers.7.block_sparse_moe.experts.0.w2",
    "model.layers.0.self_attn.q_proj",
    "model.layers.23.block_sparse_moe.experts.0.w1",
    "model.layers.8.self_attn.q_proj",
    "model.layers.13.self_attn.k_proj",
    "model.layers.5.self_attn.v_proj",
    "model.layers.11.self_attn.k_proj",
    "model.layers.17.block_sparse_moe.experts.4.w1",
    "model.layers.21.block_sparse_moe.experts.5.w3",
    "model.layers.31.block_sparse_moe.experts.1.w2",
    "model.layers.25.block_sparse_moe.gate",
    "model.audio_encoder.encoder.enc.1.encoders.18.self_attn.linear_q",
    "model.layers.18.block_sparse_moe.experts.0.w2",
    "model.layers.19.block_sparse_moe.experts.3.w1",
    "model.layers.30.block_sparse_moe.experts.3.w3",
    "model.layers.31.block_sparse_moe.experts.6.w2",
    "model.layers.24.block_sparse_moe.experts.3.w3",
    "model.layers.11.block_sparse_moe.experts.4.w2",
    "model.layers.14.block_sparse_moe.experts.3.w3",
    "model.layers.7.block_sparse_moe.experts.1.w1",
    "model.layers.25.block_sparse_moe.experts.5.w1",
    "model.mm_projector.2",
    "model.layers.4.block_sparse_moe.gate",
    "model.audio_encoder.encoder.enc.1.encoders.8.self_attn.linear_k",
    "model.layers.22.block_sparse_moe.experts.1.w1",
    "model.layers.15.block_sparse_moe.experts.1.w2",
    "model.layers.18.block_sparse_moe.experts.6.w1",
    "model.layers.11.block_sparse_moe.experts.4.w3",
    "model.layers.17.block_sparse_moe.experts.7.w2",
    "model.layers.22.block_sparse_moe.experts.2.w2",
    "model.layers.23.block_sparse_moe.experts.1.w3",
    "model.layers.18.self_attn.k_proj",
    "model.layers.16.block_sparse_moe.experts.6.w3",
    "model.layers.18.block_sparse_moe.experts.2.w3",
    "model.layers.1.self_attn.v_proj",
    "model.layers.20.block_sparse_moe.experts.5.w2",
    "model.layers.15.self_attn.k_proj",
    "model.layers.14.block_sparse_moe.gate",
    "model.layers.29.self_attn.k_proj",
    "model.layers.14.block_sparse_moe.experts.1.w3",
    "model.layers.11.block_sparse_moe.gate",
    "model.layers.13.block_sparse_moe.experts.4.w3",
    "model.layers.27.self_attn.o_proj",
    "model.layers.28.block_sparse_moe.experts.7.w3",
    "model.audio_encoder.encoder.enc.1.encoders.4.self_attn.linear_out",
    "model.layers.8.block_sparse_moe.experts.7.w2",
    "model.layers.12.block_sparse_moe.experts.7.w3",
    "model.layers.2.block_sparse_moe.experts.2.w1",
    "model.layers.21.block_sparse_moe.experts.4.w2",
    "model.layers.29.block_sparse_moe.gate",
    "model.layers.16.block_sparse_moe.experts.3.w2",
    "model.layers.24.block_sparse_moe.experts.6.w2",
    "model.layers.19.block_sparse_moe.experts.3.w3",
    "model.layers.0.self_attn.v_proj",
    "model.layers.1.block_sparse_moe.experts.2.w1",
    "model.layers.8.block_sparse_moe.experts.2.w1",
    "model.layers.1.block_sparse_moe.experts.0.w1",
    "model.layers.26.block_sparse_moe.experts.1.w1",
    "model.layers.3.block_sparse_moe.experts.2.w1",
    "model.layers.25.block_sparse_moe.experts.2.w1",
    "model.layers.5.block_sparse_moe.experts.7.w3",
    "model.layers.16.block_sparse_moe.experts.5.w2",
    "model.layers.25.block_sparse_moe.experts.7.w2",
    "model.layers.26.block_sparse_moe.experts.3.w3",
    "model.layers.27.block_sparse_moe.experts.6.w3",
    "model.audio_encoder.encoder.enc.1.encoders.22.self_attn.linear_pos",
    "model.layers.9.block_sparse_moe.experts.5.w3",
    "model.audio_encoder.encoder.enc.1.encoders.8.self_attn.linear_q",
    "model.layers.26.block_sparse_moe.experts.7.w3",
    "model.layers.0.block_sparse_moe.experts.0.w1",
    "model.audio_encoder.encoder.enc.1.encoders.17.self_attn.linear_out",
    "model.layers.4.block_sparse_moe.experts.2.w1",
    "model.layers.2.block_sparse_moe.experts.7.w3",
    "model.layers.15.block_sparse_moe.experts.6.w2",
    "model.layers.19.self_attn.o_proj",
    "model.layers.18.block_sparse_moe.experts.4.w1",
    "model.layers.17.block_sparse_moe.gate",
    "model.audio_encoder.encoder.enc.1.encoders.5.self_attn.linear_v",
    "model.layers.6.block_sparse_moe.experts.1.w3",
    "model.layers.15.block_sparse_moe.experts.7.w1",
    "model.audio_encoder.encoder.enc.1.encoders.9.self_attn.linear_k",
    "model.layers.4.block_sparse_moe.experts.7.w2",
    "model.layers.13.block_sparse_moe.experts.1.w1",
    "model.audio_encoder.encoder.enc.1.encoders.19.self_attn.linear_out",
    "model.layers.10.block_sparse_moe.experts.0.w1",
    "model.layers.6.block_sparse_moe.experts.3.w1",
    "model.layers.0.block_sparse_moe.experts.3.w2",
    "model.layers.30.block_sparse_moe.experts.3.w1",
    "model.layers.25.block_sparse_moe.experts.0.w2",
    "model.layers.29.block_sparse_moe.experts.2.w3",
    "model.layers.10.self_attn.v_proj",
    "model.layers.25.block_sparse_moe.experts.6.w3",
    "model.layers.11.block_sparse_moe.experts.5.w3",
    "model.layers.9.block_sparse_moe.experts.5.w2",
    "model.audio_encoder.encoder.enc.1.encoders.16.self_attn.linear_pos",
    "model.layers.22.block_sparse_moe.experts.3.w2",
    "model.layers.7.block_sparse_moe.experts.2.w1",
    "model.layers.9.block_sparse_moe.experts.3.w1",
    "model.layers.11.block_sparse_moe.experts.7.w1",
    "model.audio_encoder.encoder.enc.1.embed.0",
    "model.layers.26.block_sparse_moe.experts.4.w3",
    "model.layers.8.block_sparse_moe.experts.0.w2",
    "model.layers.29.block_sparse_moe.experts.3.w3",
    "model.layers.11.block_sparse_moe.experts.2.w1",
    "model.layers.3.block_sparse_moe.experts.5.w3",
    "model.layers.9.block_sparse_moe.experts.2.w3",
    "model.audio_encoder.encoder.enc.1.encoders.14.self_attn.linear_pos",
    "model.layers.19.block_sparse_moe.experts.1.w2",
    "model.layers.23.block_sparse_moe.experts.5.w1",
    "model.layers.14.block_sparse_moe.experts.4.w2",
    "model.layers.17.block_sparse_moe.experts.4.w3",
    "model.layers.12.block_sparse_moe.experts.6.w2",
    "model.layers.25.block_sparse_moe.experts.6.w2",
    "model.layers.7.block_sparse_moe.experts.5.w2",
    "model.audio_encoder.encoder.enc.1.encoders.13.self_attn.linear_pos",
    "model.layers.7.block_sparse_moe.experts.7.w1",
    "model.layers.20.block_sparse_moe.experts.0.w3",
    "model.layers.31.block_sparse_moe.experts.4.w2",
    "model.audio_encoder.encoder.enc.1.encoders.5.feed_forward.w_2",
    "model.layers.6.self_attn.v_proj",
    "model.layers.21.block_sparse_moe.experts.0.w3",
    "model.audio_encoder.encoder.enc.1.encoders.23.self_attn.linear_v",
    "model.layers.17.self_attn.v_proj",
    "model.layers.12.block_sparse_moe.experts.5.w2",
    "model.audio_encoder.encoder.enc.1.encoders.12.self_attn.linear_out",
    "model.layers.13.block_sparse_moe.experts.5.w2",
    "model.layers.7.self_attn.v_proj",
    "model.layers.2.block_sparse_moe.experts.5.w1",
    "model.layers.20.block_sparse_moe.experts.2.w1",
    "model.layers.16.block_sparse_moe.gate",
    "model.layers.24.block_sparse_moe.experts.5.w1",
    "model.audio_encoder.encoder.enc.1.encoders.6.feed_forward.w_1",
    "model.layers.1.block_sparse_moe.experts.2.w2",
    "model.layers.8.block_sparse_moe.experts.0.w3",
    "model.layers.11.block_sparse_moe.experts.5.w1",
    "model.layers.9.block_sparse_moe.experts.7.w3",
    "model.layers.19.block_sparse_moe.gate",
    "model.layers.20.block_sparse_moe.experts.1.w3",
    "model.layers.23.block_sparse_moe.experts.3.w1",
    "model.layers.11.block_sparse_moe.experts.0.w3",
    "model.layers.8.self_attn.o_proj",
    "model.audio_encoder.encoder.enc.1.encoders.16.self_attn.linear_v",
    "model.layers.28.block_sparse_moe.experts.0.w1",
    "model.layers.11.block_sparse_moe.experts.2.w2",
    "model.layers.14.block_sparse_moe.experts.1.w1",
    "model.audio_encoder.encoder.enc.1.encoders.14.self_attn.linear_v",
    "model.layers.29.block_sparse_moe.experts.4.w1",
    "model.layers.8.self_attn.k_proj",
    "model.audio_encoder.encoder.enc.1.encoders.21.self_attn.linear_out",
    "model.layers.8.block_sparse_moe.experts.1.w3",
    "model.layers.19.block_sparse_moe.experts.2.w1",
    "model.layers.2.block_sparse_moe.experts.7.w2",
    "model.layers.5.block_sparse_moe.experts.5.w1",
    "model.layers.13.block_sparse_moe.experts.3.w2",
    "model.layers.16.block_sparse_moe.experts.2.w2",
    "model.layers.19.block_sparse_moe.experts.1.w3",
    "model.layers.21.self_attn.v_proj",
    "model.layers.29.block_sparse_moe.experts.3.w1",
    "model.layers.18.block_sparse_moe.gate",
    "model.audio_encoder.encoder.enc.1.encoders.23.self_attn.linear_pos",
    "model.layers.24.block_sparse_moe.gate",
    "model.layers.17.block_sparse_moe.experts.6.w1",
    "model.audio_encoder.encoder.enc.1.encoders.21.self_attn.linear_v",
    "model.layers.24.block_sparse_moe.experts.6.w1",
    "model.audio_encoder.encoder.enc.1.encoders.21.self_attn.linear_q",
    "model.layers.19.self_attn.k_proj",
    "model.layers.9.block_sparse_moe.experts.4.w1",
    "model.layers.20.block_sparse_moe.experts.7.w2",
    "model.layers.4.block_sparse_moe.experts.2.w2",
    "model.audio_encoder.encoder.enc.1.encoders.17.self_attn.linear_pos",
    "model.layers.14.block_sparse_moe.experts.4.w1",
    "model.layers.27.self_attn.v_proj",
    "model.audio_encoder.encoder.enc.1.encoders.11.self_attn.linear_out",
    "model.audio_encoder.encoder.enc.1.encoders.16.feed_forward.w_2",
    "model.layers.4.block_sparse_moe.experts.7.w1",
    "model.layers.24.block_sparse_moe.experts.1.w1",
    "model.layers.31.block_sparse_moe.experts.4.w3",
    "model.layers.24.block_sparse_moe.experts.2.w3",
    "model.layers.17.block_sparse_moe.experts.1.w1",
    "model.layers.26.self_attn.q_proj",
    "model.layers.26.block_sparse_moe.experts.0.w3",
    "model.audio_encoder.encoder.enc.1.encoders.10.self_attn.linear_k",
    "model.layers.13.block_sparse_moe.experts.4.w1",
    "model.audio_encoder.encoder.enc.1.encoders.20.feed_forward.w_1",
    "model.layers.29.block_sparse_moe.experts.1.w2",
    "model.layers.6.block_sparse_moe.experts.6.w3",
    "model.audio_encoder.encoder.enc.1.encoders.14.feed_forward.w_2",
    "model.layers.13.block_sparse_moe.experts.3.w3",
    "model.layers.16.block_sparse_moe.experts.7.w2",
    "model.audio_encoder.encoder.enc.1.encoders.4.self_attn.linear_v",
    "model.layers.2.block_sparse_moe.experts.2.w2",
    "model.layers.4.self_attn.o_proj",
    "model.layers.11.block_sparse_moe.experts.1.w3",
    "model.layers.20.block_sparse_moe.experts.5.w3",
    "model.layers.13.self_attn.v_proj",
    "model.layers.12.self_attn.k_proj",
    "model.layers.23.block_sparse_moe.experts.6.w2",
    "model.layers.3.block_sparse_moe.experts.3.w1",
    "model.layers.7.block_sparse_moe.experts.3.w2",
    "model.layers.30.block_sparse_moe.experts.7.w2",
    "model.audio_encoder.encoder.enc.1.encoders.13.self_attn.linear_k",
    "model.layers.5.block_sparse_moe.experts.4.w2",
    "model.audio_encoder.encoder.enc.1.encoders.22.self_attn.linear_out",
    "model.layers.18.self_attn.o_proj",
    "model.layers.6.block_sparse_moe.experts.6.w1",
    "model.layers.3.block_sparse_moe.experts.4.w1",
    "model.audio_encoder.encoder.enc.0.core.out.0",
    "model.layers.1.block_sparse_moe.experts.0.w3",
    "model.layers.7.block_sparse_moe.experts.5.w3",
    "model.layers.2.self_attn.v_proj",
    "model.layers.15.block_sparse_moe.experts.1.w3",
    "model.layers.9.block_sparse_moe.experts.4.w3",
    "model.audio_encoder.encoder.enc.1.encoders.19.self_attn.linear_q",
    "model.layers.0.block_sparse_moe.experts.0.w3",
    "model.layers.9.block_sparse_moe.experts.7.w1",
    "model.layers.16.block_sparse_moe.experts.4.w2",
    "model.audio_encoder.encoder.enc.1.encoders.9.self_attn.linear_v",
    "model.audio_encoder.encoder.enc.1.encoders.3.self_attn.linear_pos",
    "model.layers.28.block_sparse_moe.experts.4.w2",
    "model.audio_encoder.encoder.enc.1.encoders.22.self_attn.linear_k",
    "model.layers.4.block_sparse_moe.experts.5.w3",
    "model.audio_encoder.encoder.enc.1.encoders.0.self_attn.linear_k",
    "model.layers.16.block_sparse_moe.experts.2.w1",
    "model.audio_encoder.encoder.enc.1.encoders.1.feed_forward.w_2",
    "model.audio_encoder.encoder.enc.1.encoders.12.self_attn.linear_pos",
    "model.layers.26.block_sparse_moe.experts.6.w1",
    "model.layers.3.block_sparse_moe.experts.0.w2",
    "model.layers.18.block_sparse_moe.experts.7.w3",
    "model.layers.5.block_sparse_moe.experts.7.w2",
    "model.layers.20.block_sparse_moe.experts.6.w3",
    "model.layers.22.self_attn.o_proj",
    "model.layers.20.block_sparse_moe.experts.4.w3",
    "model.layers.17.block_sparse_moe.experts.1.w2",
    "model.layers.23.block_sparse_moe.experts.5.w2",
    "model.layers.7.block_sparse_moe.experts.4.w3",
    "model.layers.30.block_sparse_moe.experts.1.w3",
    "model.layers.28.block_sparse_moe.experts.0.w2",
    "model.layers.15.block_sparse_moe.experts.4.w1",
    "model.audio_encoder.encoder.enc.1.encoders.18.self_attn.linear_v",
    "model.layers.16.block_sparse_moe.experts.7.w1",
    "model.layers.18.block_sparse_moe.experts.1.w3",
    "model.layers.24.block_sparse_moe.experts.4.w3",
    "model.layers.28.block_sparse_moe.experts.5.w2",
    "model.layers.22.block_sparse_moe.experts.4.w2",
    "model.layers.17.block_sparse_moe.experts.7.w1",
    "model.layers.28.block_sparse_moe.experts.5.w1",
    "model.layers.6.block_sparse_moe.experts.5.w1",
    "model.layers.20.block_sparse_moe.experts.1.w2",
    "model.layers.11.block_sparse_moe.experts.3.w3",
    "model.layers.13.block_sparse_moe.experts.7.w1",
    "model.layers.23.block_sparse_moe.experts.2.w2",
    "model.layers.4.block_sparse_moe.experts.2.w3",
    "model.audio_encoder.encoder.enc.1.encoders.1.feed_forward.w_1",
    "model.layers.5.block_sparse_moe.experts.6.w3",
    "model.layers.28.self_attn.o_proj",
    "model.layers.4.block_sparse_moe.experts.4.w2",
    "model.layers.19.block_sparse_moe.experts.7.w3",
    "model.layers.18.block_sparse_moe.experts.0.w1",
    "model.layers.20.block_sparse_moe.experts.2.w2",
    "model.layers.9.block_sparse_moe.experts.0.w2",
    "model.layers.21.block_sparse_moe.experts.2.w3",
    "model.layers.22.block_sparse_moe.experts.1.w2",
    "model.layers.14.block_sparse_moe.experts.0.w2",
    "model.layers.5.block_sparse_moe.experts.1.w1",
    "model.layers.4.block_sparse_moe.experts.6.w1",
    "model.layers.2.block_sparse_moe.experts.7.w1",
    "model.layers.20.block_sparse_moe.experts.6.w2",
    "model.layers.30.block_sparse_moe.experts.2.w3",
    "model.audio_encoder.encoder.enc.1.encoders.7.self_attn.linear_q",
    "model.audio_encoder.encoder.enc.1.encoders.10.feed_forward.w_2",
    "model.audio_encoder.encoder.enc.1.encoders.9.feed_forward.w_2",
    "model.layers.13.self_attn.q_proj",
    "model.layers.0.block_sparse_moe.experts.5.w2",
    "model.layers.26.block_sparse_moe.experts.2.w2",
    "model.layers.31.block_sparse_moe.experts.3.w3",
    "model.layers.9.self_attn.o_proj"
  ],
  "task_type": "CAUSAL_LM",
  "use_dora": false,
  "use_rslora": false
}