config.json

{
  "architectures": [
    "LlamaForCausalLM"
  ],
  "bos_token_id": 1,
  "eos_token_id": 2,
  "hidden_act": "swish",
  "hidden_size": 5120,
  "initializer_range": 0.02,
  "intermediate_size": 13824,
  "layer_norm_eps": 1e-05,
  "max_position_embeddings": 4096,
  "model_type": "llama",
  "num_attention_heads": 40,
  "num_hidden_layers": 40,
  "pad_token_id": 0,
  "rms_norm_eps": 1e-06,
  "rotary_emb_base": 10000,
  "rotary_pct": 1,
  "tie_word_embeddings": false,
  "torch_dtype": "bfloat16",
  "use_bfloat16": true,
  "transformers_version": "4.31.0",
  "use_cache": true,
  "use_parallel_residual": false,
  "vocab_size": 39424,
  "rope_scaling": {"type": "linear", "factor": 2.0}
}