config.json

{
  "architectures": [
    "SpTransformerModel"
  ],
  "attention_hidden_size": 256,
  "batch_norm_eps": 1e-05,
  "batch_norm_momentum": 0.1,
  "bos_token_id": null,
  "bucket_size": 64,
  "context": 4000,
  "dtype": "float32",
  "encoders": [
    {
      "hidden_size": 128
    },
    {
      "hidden_size": 64
    }
  ],
  "eos_token_id": null,
  "head": {
    "act": null,
    "bias": true,
    "dropout": 0.0,
    "hidden_size": 256,
    "layer_norm_eps": 1e-12,
    "loss_weight": null,
    "num_labels": 15,
    "output_name": null,
    "problem_type": "regression",
    "transform": null,
    "transform_act": "gelu",
    "type": null
  },
  "hidden_act": "relu",
  "hidden_size": 128,
  "id2label": null,
  "intermediate_act": "gelu",
  "intermediate_size": 1024,
  "label2id": null,
  "mask_token_id": null,
  "max_seq_len": 8192,
  "model_type": "sptransformer",
  "null_token_id": null,
  "num_attention_heads": 8,
  "num_hidden_layers": 8,
  "num_labels": 15,
  "num_local_attention_heads": 2,
  "num_splice_labels": 3,
  "num_tissues": 15,
  "output_contexts": false,
  "pad_token_id": 4,
  "problem_type": "regression",
  "tie_word_embeddings": true,
  "transformers_version": "5.7.0",
  "unk_token_id": 4,
  "vocab_size": 5
}