Metrics

PPL	arc_easy	arc_challenge	piqa	winogrande	hellaswag	mmlu	QA Avg
2794.56	26.14 ± 0.90	20.56 ± 1.18	51.85 ± 1.17	49.72 ± 1.41	25.77 ± 0.44	-	34.81

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_1.58bit_ternary_range_scale

Base model

Finetuned

(53)

this model