config.json

{
  "absolute_positional_embedding_type": null,
  "add_marker_tokens": true,
  "architectures": [
    "LirDprTiteModel"
  ],
  "backbone_model_type": "tite",
  "doc_length": 256,
  "doc_pooling_strategy": "first",
  "dropout_prob": 0.1,
  "embedding_dim": null,
  "hidden_act": "gelu_pytorch_tanh",
  "hidden_sizes": [
    768,
    768,
    768,
    768,
    768,
    768,
    768,
    768,
    768,
    768,
    768,
    768
  ],
  "initializer_range": 0.02,
  "intermediate_sizes": [
    3072,
    3072,
    3072,
    3072,
    3072,
    3072,
    3072,
    3072,
    3072,
    3072,
    3072,
    3072
  ],
  "kernel_sizes": [
    null,
    null,
    null,
    2,
    2,
    2,
    2,
    2,
    2,
    2,
    2,
    2
  ],
  "layer_norm_eps": 1e-12,
  "max_position_embeddings": 512,
  "model_type": "lir-dpr",
  "norm_location": "post",
  "norm_type": "layer",
  "normalize": false,
  "num_attention_heads": [
    12,
    12,
    12,
    12,
    12,
    12,
    12,
    12,
    12,
    12,
    12,
    12
  ],
  "num_hidden_layers": 12,
  "pad_token_id": null,
  "pooling_implementation": "triton",
  "pooling_location": "intra",
  "positional_embedding_type": null,
  "projection": null,
  "query_length": 32,
  "query_pooling_strategy": "first",
  "relative_positional_embedding_type": "rotary",
  "rope_implementation": "eager",
  "rotary_interleaved": true,
  "save_step": 10006,
  "similarity_function": "dot",
  "sparsification": null,
  "strides": [
    null,
    null,
    null,
    2,
    2,
    2,
    2,
    2,
    2,
    2,
    2,
    2
  ],
  "torch_dtype": "float32",
  "transformers_version": "4.52.4",
  "vocab_size": 30528
}