Mistral.json

{
    "model_name_or_path": "mistralai/Mistral-7B-Instruct-v0.2",
    "peft_model_name_or_path": "McGill-NLP/LLM2Vec-Mistral-7B-Instruct-v2-mntp",
    "simcse_dropout": 0.3,
    "bidirectional": true,
    "pooling_mode": "mean",
    "dataset_name": "Wiki1M",
    "dataset_file_path": "cache/wiki1m_for_simcse.txt",
    "remove_unused_columns": false,
    "learning_rate": 3e-5,
    "loss_scale": 20,
    "per_device_train_batch_size": 128,
    "gradient_accumulation_steps": 1,
    "do_train": true,
    "disable_tqdm": false,
    "max_seq_length": 128,
    "overwrite_output_dir": true,
    "output_dir": "output/mntp-simcse/Mistral-7B-Instruct-v0.2",
    "logging_steps": 50,
    "save_steps": 200,
    "save_only_model": true,
    "stop_after_n_steps": 1000,
    "lora_r": 16,
    "gradient_checkpointing": true,
    "torch_dtype": "bfloat16",
    "attn_implementation": "flash_attention_2",
    "seed": 42
}