RichardErkhov/CarrotAI_-_Llama-3.2-Rabbit-Ko-1B-Instruct-awq

Quantization made by Richard Erkhov.

Github

Discord

Request more models

Llama-3.2-Rabbit-Ko-1B-Instruct - AWQ

• Model creator: https://huggingface.co/CarrotAI/
• Original model: https://huggingface.co/CarrotAI/Llama-3.2-Rabbit-Ko-1B-Instruct/

Original model description:

license: llama3.2 datasets: - CarrotAI/ko-instruction-dataset - CarrotAI/Magpie-Ko-Pro-AIR language: - ko - en base_model: - meta-llama/Llama-3.2-1B-Instruct pipeline_tag: text-generation

Score

Tasks	Version	Filter	n-shot	Metric		Value		Stderr
kobest_boolq	1	none	5	acc	↑	0.5726	±	0.0132
		none	5	f1	↑	0.5725	±	N/A
kobest_copa	1	none	5	acc	↑	0.5200	±	0.0158
		none	5	f1	↑	0.5189	±	N/A
kobest_hellaswag	1	none	5	acc	↑	0.3640	±	0.0215
		none	5	acc_norm	↑	0.4380	±	0.0222
		none	5	f1	↑	0.3592	±	N/A
kobest_sentineg	1	none	5	acc	↑	0.5642	±	0.0249
		none	5	f1	↑	0.5554	±	N/A
kobest_wic	1	none	5	acc	↑	0.5087	±	0.0141
		none	5	f1	↑	0.4979	±	N/A

Tasks	Version	Filter	n-shot	Metric		Value		Stderr
gsm8k	3	flexible-extract	5	exact_match	↑	0.2995	±	0.0126
		strict-match	5	exact_match	↑	0.2987	±	0.0126

@article{Llama3.2RabbitKo1BInstruct,
  title={CarrotAI/Llama-3.2-Rabbit-Ko-1B-Instruct Card},
  author={CarrotAI (L, GEUN)},
  year={2024},
  url = {https://huggingface.co/CarrotAI/Llama-3.2-Rabbit-Ko-1B-Instruct}
}