openbmb

/

minicpm-dpo-bf16-ggml-model-q4_0

like 0

  "name": "models",
  "arch": "llama",
  "quant": "Q4_0",
  "context_length": 2048,
  "embedding_length": 2304,
  "num_layers": 40,
  "rope": {
    "dimension_count": 64
  },
  "head_count": 36,
  "head_count_kv": 36,
  "parameters": "13B"

I found it is 13B in the model info. And it is also the same case for internLM.
May I ask why there is a gap?