{
  "architectures": [
    "Optimus5PrimeForSequencePrediction"
  ],
  "bos_token_id": 1,
  "conv_channels": 120,
  "conv_dropout": 0.0,
  "conv_kernel_size": 8,
  "dense_dropout": 0.2,
  "dtype": "float32",
  "eos_token_id": 2,
  "head": {
    "act": null,
    "bias": true,
    "dropout": 0.0,
    "hidden_size": 40,
    "layer_norm_eps": 1e-12,
    "loss_weight": null,
    "num_labels": 1,
    "output_name": null,
    "problem_type": "regression",
    "transform": null,
    "transform_act": "gelu",
    "type": null
  },
  "hidden_act": "relu",
  "hidden_size": 40,
  "id2label": null,
  "label2id": null,
  "mask_token_id": 4,
  "model_type": "optimus5prime",
  "null_token_id": 5,
  "num_conv_layers": 3,
  "num_labels": 1,
  "pad_token_id": 0,
  "sequence_length": 50,
  "tie_word_embeddings": true,
  "transformers_version": "5.9.0",
  "unk_token_id": 3,
  "vocab_size": 5
}