Modelpublic

RichardErkhov/mesolitica_-_llama-1b-hf-32768-fpf-gguf

sourceHugging Faceupdated 2y agoView on Hugging Face

0likes284downloads

Model Card

Quantization made by Richard Erkhov.

Request more models

llama-1b-hf-32768-fpf - GGUF

—Model creator: https://huggingface.co/mesolitica/
—Original model: https://huggingface.co/mesolitica/llama-1b-hf-32768-fpf/

Name	Quant method	Size
llama-1b-hf-32768-fpf.Q2_K.gguf	Q2_K	0.42GB
llama-1b-hf-32768-fpf.Q3_K_S.gguf	Q3KS	0.48GB
llama-1b-hf-32768-fpf.Q3_K.gguf	Q3_K	0.52GB
llama-1b-hf-32768-fpf.Q3_K_M.gguf	Q3KM	0.52GB
llama-1b-hf-32768-fpf.Q3_K_L.gguf	Q3KL	0.55GB
llama-1b-hf-32768-fpf.IQ4_XS.gguf	IQ4_XS	0.57GB
llama-1b-hf-32768-fpf.Q4_0.gguf	Q4_0	0.59GB
llama-1b-hf-32768-fpf.IQ4_NL.gguf	IQ4_NL	0.59GB
llama-1b-hf-32768-fpf.Q4_K_S.gguf	Q4KS	0.6GB
llama-1b-hf-32768-fpf.Q4_K.gguf	Q4_K	0.62GB
llama-1b-hf-32768-fpf.Q4_K_M.gguf	Q4KM	0.62GB
llama-1b-hf-32768-fpf.Q4_1.gguf	Q4_1	0.65GB
llama-1b-hf-32768-fpf.Q5_0.gguf	Q5_0	0.7GB
llama-1b-hf-32768-fpf.Q5_K_S.gguf	Q5KS	0.7GB
llama-1b-hf-32768-fpf.Q5_K.gguf	Q5_K	0.72GB
llama-1b-hf-32768-fpf.Q5_K_M.gguf	Q5KM	0.72GB
llama-1b-hf-32768-fpf.Q5_1.gguf	Q5_1	0.76GB
llama-1b-hf-32768-fpf.Q6_K.gguf	Q6_K	0.82GB
llama-1b-hf-32768-fpf.Q8_0.gguf	Q8_0	1.06GB

Original model description: --- language:

—ms ---

Full Parameter Finetuning 1B 32768 context length Llama2 on Malaysian text

1B derived from first 4 layers 7B model.

README at https://github.com/mesolitica/malaya/tree/5.1/session/llama2#1b-32768-context-length-flash-attention-2

WandB, https://wandb.ai/mesolitica/fpf-Llama-2-1b-32k-hf