config.json

{
  "architectures": [
    "YForCausalLM1_1"
  ],
  "auto_map": {
    "AutoConfig": "ymodel1_1.YConfig1_1",
    "AutoModelForCausalLM": "ymodel1_1.YForCausalLM1_1"
  },
  "bos_token_id": 1,
  "dropout": 0.1,
  "eos_token_id": 2,
  "exp": 3.0,
  "ffn_shared": 3,
  "flash_attn": true,
  "groups": 8,
  "head_dim": 64,
  "hidden_act": "gelu_pytorch_tanh",
  "hidden_size": 512,
  "intermediate_size": null,
  "max_position_embeddings": 4096,
  "model_type": "ynet",
  "num_heads": 8,
  "num_layers": 12,
  "pe_dim": 64,
  "rms_norm_eps": 1e-07,
  "rope_theta": 50000.0,
  "self_distill": true,
  "torch_dtype": "bfloat16",
  "transformers_version": "4.51.3",
  "vocab_size": 6400
}