Training procedure

The following bitsandbytes quantization config was used during training:

• quant_method: bitsandbytes
• load_in_8bit: False
• load_in_4bit: True
• 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: nf4
• bnb_4bit_use_double_quant: True
• bnb_4bit_compute_dtype: bfloat16

Framework versions

• PEFT 0.5.0.dev0

Citation

@misc {shuvam_mandal_2023, author = { {shuvam mandal} }, title = { falcon-med-FT-v1.114 (Revision 87e4f8f) }, year = 2023, url = { https://huggingface.co/shuvom/falcon-med-FT-v1.114 }, doi = { 10.57967/hf/1012 }, publisher = { Hugging Face } }