metadata
library_name: peft
datasets:
- mohamedemam/Arabic-samsum-dialogsum
- samsum
language:
- ar
tags:
- boomz
- text2text
- bloom
- summarization
- messenge summary
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
model
this model bloomz trained on arabic dateset for summarizition of messeges
- PEFT 0.6.0.dev0