joehuangx/spatial-vicuna-7b-v1.5-LoRA

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 0.6.0.dev0

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	44.55
ARC (25-shot)	50.77
HellaSwag (10-shot)	74.63
MMLU (5-shot)	48.13
TruthfulQA (0-shot)	49.36
Winogrande (5-shot)	72.38
GSM8K (5-shot)	6.9
DROP (3-shot)	9.72