config.json

{
  "architectures": [
    "VideoLlama3VisionModel"
  ],
  "attention_dropout": 0.1,
  "dtype": "float32",
  "hidden_act": "gelu_pytorch_tanh",
  "hidden_size": 64,
  "initializer_range": 0.02,
  "intermediate_size": 37,
  "layer_norm_eps": 1e-06,
  "model_type": "video_llama_3_vision",
  "num_attention_heads": 4,
  "num_channels": 3,
  "num_hidden_layers": 2,
  "patch_size": 2,
  "transformers_version": "5.16.0.dev0"
}