library_name: peft | |
base_model: ybelkada/falcon-7b-sharded-bf16 | |
## Training procedure | |
The following `bitsandbytes` quantization config was used during training: | |
- 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: float16 | |
### Framework versions | |
- PEFT 0.5.0.dev0 | |