Metrics

PPL	arc_easy	arc_challenge	piqa	winogrande	hellaswag	mmlu	QA Avg
17899.79	26.26 ± 0.90	22.01 ± 1.21	52.01 ± 1.17	51.22 ± 1.40	25.96 ± 0.44	-	35.49

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_3x_data_3_epochs

Base model

Finetuned

(53)

this model