config.json

{
  "_attn_implementation_autoset": true,
  "adapters": [
    "lnn"
  ],
  "architectures": [
    "Qwen2ForCausalLM"
  ],
  "bos_token_id": 2,
  "chunk_size_feed_forward": 64,
  "d_model": 2048,
  "do_sample": true,
  "dropout": 0.1,
  "eos_token_id": 3,
  "hidden_size": 2048,
  "intermediate_size": 8192,
  "max_length": 512,
  "max_seq": 4096,
  "min_length": 64,
  "model_type": "symbiotic-llm",
  "n_heads": 32,
  "n_layers": 16,
  "no_repeat_ngram_size": 1,
  "num_attention_heads": 32,
  "num_key_value_heads": 32,
  "pad_token_id": 0,
  "rms_norm_eps": 1e-06,
  "rope_base": 1000000.0,
  "tie_embed": true,
  "tokenizer_class": "PreTrainedTokenizerFast",
  "top_p": 0.9,
  "torch_dtype": "float16",
  "transformers_version": "4.51.3",
  "typical_p": 0.95,
  "use_flash": false,
  "vocab_size": 37524
}