config.json

{
  "model_type": "tachbit",
  "architectures": [
    "TachBitModel"
  ],
  "hidden_size": 768,
  "intermediate_size": 3072,
  "num_hidden_layers": 24,
  "num_attention_heads": 12,
  "num_key_value_heads": 4,
  "vocab_size": 50272,
  "max_position_embeddings": 2048,
  "attention_bias": false,
  "hidden_act": "swiglu",
  "rope_theta": 10000.0,
  "rms_norm_eps": 1e-06,
  "pre_norm": true,
  "tie_word_embeddings": true,
  "license": "mit",
  "tags": [
    "transformer",
    "llm",
    "language-modeling",
    "tachbit",
    "text-generation",
    "pytorch",
    "onnx"
  ]
}