config.json

{
  "architectures": [
    "CohereCompassForConditionalGeneration"
  ],
  "dtype": "float32",
  "image_token_id": 5,
  "model_type": "cohere_compass",
  "pad_token_id": 0,
  "text_config": {
    "attention_bias": false,
    "attention_dropout": 0.0,
    "bos_token_id": 2,
    "eos_token_id": 255001,
    "head_dim": 8,
    "hidden_act": "silu",
    "hidden_size": 32,
    "initializer_range": 0.02,
    "intermediate_size": 64,
    "layer_norm_eps": 1e-05,
    "layer_types": [
      "full_attention",
      "full_attention"
    ],
    "logit_scale": null,
    "max_position_embeddings": 200,
    "model_type": "cohere_compass_text",
    "num_attention_heads": 4,
    "num_hidden_layers": 2,
    "num_key_value_heads": 2,
    "pad_token_id": 0,
    "pooling": null,
    "rope_parameters": {
      "full_attention": {
        "mrope_section": [
          1,
          1,
          2
        ],
        "rope_theta": 10000,
        "rope_type": "default"
      },
      "rope_theta": 10000.0,
      "rope_type": "default"
    },
    "sliding_window": 4096,
    "tie_word_embeddings": false,
    "use_cache": true,
    "vocab_size": 255035
  },
  "tie_word_embeddings": false,
  "transformers_version": "5.16.0.dev0",
  "video_token_id": 8,
  "vision_config": {
    "deepstack_visual_indexes": [
      0
    ],
    "depth": 2,
    "hidden_act": "silu",
    "hidden_size": 32,
    "in_channels": 3,
    "initializer_range": 0.02,
    "intermediate_size": 64,
    "model_type": "cohere_compass_vision",
    "num_heads": 4,
    "num_position_embeddings": 2304,
    "out_hidden_size": 32,
    "patch_size": 16,
    "spatial_merge_size": 2,
    "temporal_patch_size": 2
  },
  "vision_end_token_id": 7,
  "vision_start_token_id": 6
}