RichardErkhov/Vikhrmodels_-_Mini-cpm-vikhr-gguf

Quantization made by Richard Erkhov.

Github

Discord

Request more models

Mini-cpm-vikhr - GGUF

• Model creator: https://huggingface.co/Vikhrmodels/
• Original model: https://huggingface.co/Vikhrmodels/Mini-cpm-vikhr/

Name	Quant method	Size
Mini-cpm-vikhr.Q2_K.gguf	Q2_K	1.17GB
Mini-cpm-vikhr.Q3_K_S.gguf	Q3_K_S	1.31GB
Mini-cpm-vikhr.Q3_K.gguf	Q3_K	1.42GB
Mini-cpm-vikhr.Q3_K_M.gguf	Q3_K_M	1.42GB
Mini-cpm-vikhr.Q3_K_L.gguf	Q3_K_L	1.5GB
Mini-cpm-vikhr.IQ4_XS.gguf	IQ4_XS	1.5GB
Mini-cpm-vikhr.Q4_0.gguf	Q4_0	1.54GB
Mini-cpm-vikhr.IQ4_NL.gguf	IQ4_NL	1.56GB
Mini-cpm-vikhr.Q4_K_S.gguf	Q4_K_S	1.61GB
Mini-cpm-vikhr.Q4_K.gguf	Q4_K	1.72GB
Mini-cpm-vikhr.Q4_K_M.gguf	Q4_K_M	1.72GB
Mini-cpm-vikhr.Q4_1.gguf	Q4_1	1.69GB
Mini-cpm-vikhr.Q5_0.gguf	Q5_0	1.83GB
Mini-cpm-vikhr.Q5_K_S.gguf	Q5_K_S	1.86GB
Mini-cpm-vikhr.Q5_K.gguf	Q5_K	1.95GB
Mini-cpm-vikhr.Q5_K_M.gguf	Q5_K_M	1.95GB
Mini-cpm-vikhr.Q5_1.gguf	Q5_1	1.97GB
Mini-cpm-vikhr.Q6_K.gguf	Q6_K	2.25GB
Mini-cpm-vikhr.Q8_0.gguf	Q8_0	2.76GB

Original model description:

license: apache-2.0 datasets: - dichspace/darulm language: - ru - en - zh library_name: transformers

Task	Version	Metric	Value		Stderr
parus	0	acc	0.4825	±	0.0250
rcb	0	acc	0.3333	±	0.0226
		f1_macro	0.1675
rwsd	0	acc	0.4802	±	0.0203
mmlu	0	acc	0.3402	±	похуй+похуй я выполнил домашку, а cбер девайсес mmlu нет, блять какой же еБаный код ебал рот блять нахуй я этим занимаюсь сука???

https://wandb.ai/alexwortega/cpm_rus/runs/pzj0wcui