Information

SuperPlatty-30B is a merge of garage-bAInd/Platypus-30B and kaiokendev/SuperCOT-LoRA

Model Details

Trained by: Platypus-30B trained by Cole Hunter & Ariel Lee; SuperCOT-LoRA trained by kaiokendev.
Model type: SuperPlatty-30B is an auto-regressive language model based on the LLaMA transformer architecture.
Language(s): English
License for base weights: License for the base LLaMA model's weights is Meta's non-commercial bespoke license.

Hyperparameter	Value
$n_\text{parameters}$	33B
$d_\text{model}$	6656
$n_\text{layers}$	60
$n_\text{heads}$	52

Reproducing Evaluation Results

Install LM Evaluation Harness:

git clone https://github.com/EleutherAI/lm-evaluation-harness
cd lm-evaluation-harness
pip install -e .

Each task was evaluated on a single A100 80GB GPU.

ARC:

python main.py --model hf-causal-experimental --model_args pretrained=garage-bAIdnd/SuperPlatty-30B --tasks arc_challenge --batch_size 1 --no_cache --write_out --output_path results/SuperPlatty-30B/arc_challenge_25shot.json --device cuda --num_fewshot 25

HellaSwag:

python main.py --model hf-causal-experimental --model_args pretrained=garage-bAIdnd/SuperPlatty-30B --tasks hellaswag --batch_size 1 --no_cache --write_out --output_path results/SuperPlatty-30B/hellaswag_10shot.json --device cuda --num_fewshot 10

MMLU:

python main.py --model hf-causal-experimental --model_args pretrained=garage-bAIdnd/SuperPlatty-30B --tasks hendrycksTest-* --batch_size 1 --no_cache --write_out --output_path results/SuperPlatty-30B/mmlu_5shot.json --device cuda --num_fewshot 5

TruthfulQA:

python main.py --model hf-causal-experimental --model_args pretrained=garage-bAIdnd/SuperPlatty-30B --tasks truthfulqa_mc --batch_size 1 --no_cache --write_out --output_path results/SuperPlatty-30B/truthfulqa_0shot.json --device cuda

Limitations and bias

The base LLaMA model is trained on various data, some of which may contain offensive, harmful, and biased content that can lead to toxic behavior. See Section 5.1 of the LLaMA paper. We have not performed any studies to determine how fine-tuning on the aforementioned datasets affect the model's behavior and toxicity. Do not treat chat responses from this model as a substitute for human judgment or as a source of truth. Please use responsibly.

Citations

@article{touvron2023llama,
  title={LLaMA: Open and Efficient Foundation Language Models},
  author={Touvron, Hugo and Lavril, Thibaut and Izacard, Gautier and Martinet, Xavier and Lachaux, Marie-Anne and Lacroix, Timoth{\'e}e and Rozi{\`e}re, Baptiste and Goyal, Naman and Hambro, Eric and Azhar, Faisal and Rodriguez, Aurelien and Joulin, Armand and Grave, Edouard and Lample, Guillaume},
  journal={arXiv preprint arXiv:2302.13971},
  year={2023}
}
@article{hu2021lora,
  title={LoRA: Low-Rank Adaptation of Large Language Models},
  author={Hu, Edward J. and Shen, Yelong and Wallis, Phillip and Allen-Zhu, Zeyuan and Li, Yuanzhi and Wang, Shean and Chen, Weizhu},
  journal={CoRR},
  year={2021}
}

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	57.89
ARC (25-shot)	65.78
HellaSwag (10-shot)	83.95
MMLU (5-shot)	62.57
TruthfulQA (0-shot)	53.52
Winogrande (5-shot)	80.35
GSM8K (5-shot)	9.63
DROP (3-shot)	49.44