metadata
library_name: peft
base_model: NousResearch/Llama-2-7b-chat-hf
license: apache-2.0
language:
- ar
pipeline_tag: conversational
datasets:
- HeshamHaroon/oasst1-ar-threads
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: float16
Framework versions
- PEFT 0.6.2