totally-not-an-llm/EverythingLM-13b-peft

Trained for 5 epochs on https://huggingface.co/datasets/totally-not-an-llm/EverythingLM-data. Merged model can be found here: https://huggingface.co/totally-not-an-llm/EverythingLM-13b-16k.

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