screenshot-llama-380m / config.json
princeton-nlp's picture
update
f27e32a
{
"_commit_hash": null,
"add_input_mlp": false,
"add_output_mlp": false,
"architectures": [
"LlamaForScreenshot"
],
"attention_bias": false,
"bos_token_id": 1,
"eos_token_id": 2,
"hidden_act": "silu",
"hidden_size": 1024,
"image_mode": "RGB",
"img_begin_token_id": 32000,
"img_end_token_id": 32001,
"initializer_range": 0.02,
"intermediate_size": 2816,
"max_position_embeddings": 4096,
"model_type": "screenshot-llama",
"newline_token_id": 13,
"norm_pix_loss": true,
"num_attention_heads": 16,
"num_hidden_layers": 24,
"num_key_value_heads": 16,
"patch_embed_size": 768,
"patch_token_id": 32002,
"pixel_decoder_config": null,
"pretraining_tp": 1,
"rms_norm_eps": 1e-05,
"rope_scaling": null,
"rope_theta": 10000.0,
"tie_word_embeddings": false,
"torch_dtype": "float32",
"transformers_version": null,
"use_cache": true,
"vocab_size": 32003
}