config.json

{
    "_name_or_path": "facebook/opt-125m",
    "_remove_final_layer_norm": false,
    "activation_dropout": 0.0,
    "activation_function": "relu",
    "architectures": [
      "FlamingoForCausalLM"
    ],
    "auto_map": {
      "AutoConfig": "configuration_flamingo.FlamingoConfig",
      "AutoModelForCausalLM": "modeling_flamingo.FlamingoForCausalLM",
      "AutoModelForSequenceClassification": "modeling_flamingo.FlamingoForSequenceClassification"
    },
    "attention_dropout": 0.0,
    "bos_token_id": 2,
    "cross_attn_every": 2,
    "do_layer_norm_before": true,
    "dropout": 0.1,
    "enable_bias": true,
    "eos_token_id": 2,
    "ffn_dim": 3072,
    "finetune_LM": true,
    "hidden_size": 768,
    "id_perceiver": false,
    "init_std": 0.02,
    "inp_dim": 768,
    "layer_norm_elementwise_affine": true,
    "layerdrop": 0.0,
    "max_position_embeddings": 2048,
    "media_token_id": 32768,
    "model_type": "opt",
    "num_attention_heads": 12,
    "num_hidden_layers": 12,
    "only_attend_immediate_media": true,
    "pad_token_id": 1,
    "perceiver_depth": 2,
    "perceiver_num_latents": 64,
    "prefix": "</s>",
    "torch_dtype": "float32",
    "transformers_version": "4.29.0",
    "use_cache": true,
    "vocab_size": 32778,
    "word_embed_proj_dim": 768
  }