model/config.json

{
  "action_latent_dim": null,
  "action_seq_len": 128,
  "action_start_token_id": 4378,
  "action_vocab_size": 4375,
  "architectures": [
    "PolicyQwen3ForConditionalGeneration"
  ],
  "attention_bias": false,
  "attention_dropout": 0.0,
  "dtype": "bfloat16",
  "eos_token_id": 4379,
  "frame_stride": 1,
  "head_dim": 64,
  "hidden_act": "silu",
  "hidden_size": 256,
  "image_token_id": 4377,
  "initializer_range": 0.02,
  "intermediate_size": 1024,
  "layer_types": [
    "full_attention",
    "full_attention",
    "full_attention",
    "full_attention",
    "full_attention",
    "full_attention",
    "full_attention",
    "full_attention",
    "full_attention",
    "full_attention",
    "full_attention",
    "full_attention"
  ],
  "max_position_embeddings": 2048,
  "max_window_layers": 28,
  "model_type": "policy_qwen3",
  "num_attention_heads": 4,
  "num_frames": 2,
  "num_hidden_layers": 12,
  "num_key_value_heads": 4,
  "pad_token_id": 0,
  "policy_image_height": 64,
  "policy_image_width": 64,
  "predict_tokens": true,
  "rms_norm_eps": 1e-06,
  "rope_scaling": null,
  "rope_theta": 10000.0,
  "sliding_window": null,
  "tie_word_embeddings": true,
  "transformers_version": "4.57.1",
  "use_cache": false,
  "use_sliding_window": false,
  "vision_end_token_id": 4376,
  "vision_in_channels": 3,
  "vision_merge_size": 1,
  "vision_patch_size": 4,
  "vision_start_token_id": 4375,
  "vocab_size": 4380
}