config.json

{
  "architectures": [
    "HanseForCausalLM"
  ],
  "auto_map": {
    "AutoConfig": "configuration_hanse.HanseConfig",
    "AutoModelForCausalLM": "modeling_hanse.HanseForCausalLM"
  },
  "conv_kernel_size": 7,
  "ffn_hidden_size": 1792,
  "hidden_size": 640,
  "layer_pattern": [
    "A",
    "A",
    "C",
    "A",
    "A",
    "A",
    "C",
    "A",
    "A",
    "A",
    "C",
    "A",
    "A",
    "A"
  ],
  "max_position_embeddings": 2048,
  "max_seq_len": 2048,
  "model_type": "hanse",
  "norm_eps": 1e-06,
  "num_attention_heads": 10,
  "num_hidden_layers": 14,
  "num_key_value_heads": 2,
  "num_kv_heads": 2,
  "num_layers": 14,
  "num_query_heads": 10,
  "qk_norm": true,
  "rope_theta": 10000.0,
  "tie_word_embeddings": true,
  "torch_dtype": "bfloat16",
  "transformers_version": "4.57.6",
  "use_cache": false,
  "vocab_size": 24576
}