Metrics

PPL	arc_easy	arc_challenge	piqa	winogrande	hellaswag	mmlu	QA Avg
5.99	74.66 ± 0.89	39.76 ± 1.43	76.88 ± 0.98	67.88 ± 1.31	55.16 ± 0.50	-	62.87

Training method based on BitDistiller Paper

Safetensors

Model size

7B params

Tensor type

BF16

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for BrownianNotion/Llama-2-7b-hf_3bit_int

Base model

Finetuned

(52)

this model