Metrics

PPL	arc_easy	arc_challenge	piqa	winogrande	hellaswag	mmlu	QA Avg
17.11	37.84 ± 1.00	21.59 ± 1.20	62.46 ± 1.13	52.41 ± 1.40	33.39 ± 0.47	-	41.54

Training method based on BitDistiller Paper

Safetensors

Model size

1B params

Tensor type

BF16

Inference Providers NEW

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

Model tree for BrownianNotion/TinyLlama_v1.1_2bit_int_3B_data

Base model

Finetuned

(54)

this model