| library_name: peft | |
| base_model: NousResearch/Llama-2-7b-hf | |
| ## title | |
| a recipe classify from https://github.com/OpenPipe/OpenPipe | |
| lora fine-tune based on llama2-7b, 1hour on 2*3090 | |
| ## Training procedure | |
| The following `bitsandbytes` quantization config was used during training: | |
| - quant_method: bitsandbytes | |
| - load_in_8bit: True | |
| - load_in_4bit: False | |
| - llm_int8_threshold: 6.0 | |
| - llm_int8_skip_modules: None | |
| - llm_int8_enable_fp32_cpu_offload: False | |
| - llm_int8_has_fp16_weight: False | |
| - bnb_4bit_quant_type: fp4 | |
| - bnb_4bit_use_double_quant: False | |
| - bnb_4bit_compute_dtype: float32 | |
| ### Framework versions | |
| - PEFT 0.6.0.dev0 | |