private-llm-config.json

{
  "model_type": "llama",
  "quantization": "w4a16g128asym",
  "model_config": {
    "hidden_size": 2048,
    "intermediate_size": 8192,
    "num_attention_heads": 32,
    "num_hidden_layers": 16,
    "rms_norm_eps": 1e-05,
    "vocab_size": 128256,
    "rope_theta": 500000.0,
    "tie_word_embeddings": true,
    "rope_scaling": {
      "factor": 32.0,
      "high_freq_factor": 4.0,
      "low_freq_factor": 1.0,
      "original_max_position_embeddings": 8192,
      "rope_type": "llama3"
    },
    "context_window_size": 4096,
    "prefill_chunk_size": 128,
    "num_key_value_heads": 8,
    "head_dim": 64,
    "tensor_parallel_shards": 1,
    "pipeline_parallel_stages": 1,
    "max_batch_size": 80
  },
  "vocab_size": 128256,
  "context_window_size": 4096,
  "sliding_window_size": -1,
  "prefill_chunk_size": 128,
  "attention_sink_size": -1,
  "tensor_parallel_shards": 1,
  "mean_gen_len": 128,
  "max_gen_len": 512,
  "shift_fill_factor": 0.3,
  "temperature": 0.6,
  "presence_penalty": 0.0,
  "frequency_penalty": 0.0,
  "repetition_penalty": 1.0,
  "top_p": 0.9,
  "conv_template": "llama-3",
  "pad_token_id": 0,
  "bos_token_id": 128000,
  "eos_token_id": [
    128001,
    128008,
    128009
  ],
  "tokenizer_files": [
    "tokenizer.json",
    "tokenizer_config.json"
  ],
  "version": "0.1.0"
}