Metrics

PPL	arc_easy	arc_challenge	piqa	winogrande	hellaswag	mmlu	QA Avg
22.4281 ± 0.00	35.40 ± 0.98	21.67 ± 1.20	60.55 ± 1.14	51.93 ± 1.40	32.22 ± 0.47	-	40.36

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 Niks898/TinyLlama_v1.1_2bit_int_ce_plus_cakld_20

Base model

Finetuned

(52)

this model