config.json

{
  "name_or_path":"career_guidance_model_llama",
  "architectures": [
    "LlamaForCausalLM"
  ],
  "model_type": "llama",
  "max_seq_length": 1024,
  "r": 8,
  "lora_alpha": 16,
  "lora_dropout": 0,
  "gradient_checkpointing": true,
  "load_in_4bit": true,
  "use_cache": true,
  "vocab_size": 32000,
  "tokenizer_class": "LlamaTokenizer",
  "pad_token_id": 0,
  "bos_token_id": 1,
  "eos_token_id": 2,
  "hidden_size": 4096,
  "num_attention_heads": 32,
  "num_hidden_layers": 32,
  "intermediate_size": 11008,
  "hidden_act": "silu",
  "initializer_range": 0.02,
  "torch_dtype": "float16",
  "fp16": true,
  "bf16": false,
  "attention_probs_dropout_prob": 0.1,
  "hidden_dropout_prob": 0.1,
  "learning_rate": 0.0001,
  "weight_decay": 0.05,
  "lr_scheduler_type": "cosine",
  "warmup_ratio": 0.03,
  "optimizer_type": "adamw_8bit",
  "task_specific_params": {
    "text-generation": {
      "temperature": 0.7,
      "top_p": 0.9,
      "max_new_tokens": 128
    }
  }
}