Quantization made by Richard Erkhov.

GritLM-8x7B-KTO - GGUF

Model creator: https://huggingface.co/GritLM/
Original model: https://huggingface.co/GritLM/GritLM-8x7B-KTO/

Name	Quant method	Size
GritLM-8x7B-KTO.Q2_K.gguf	Q2_K	16.12GB
GritLM-8x7B-KTO.IQ3_XS.gguf	IQ3_XS	18.02GB
GritLM-8x7B-KTO.IQ3_S.gguf	IQ3_S	19.03GB
GritLM-8x7B-KTO.Q3_K_S.gguf	Q3_K_S	19.03GB
GritLM-8x7B-KTO.IQ3_M.gguf	IQ3_M	19.96GB
GritLM-8x7B-KTO.Q3_K.gguf	Q3_K	21.0GB
GritLM-8x7B-KTO.Q3_K_M.gguf	Q3_K_M	21.0GB
GritLM-8x7B-KTO.Q3_K_L.gguf	Q3_K_L	22.51GB
GritLM-8x7B-KTO.IQ4_XS.gguf	IQ4_XS	23.63GB
GritLM-8x7B-KTO.Q4_0.gguf	Q4_0	24.63GB
GritLM-8x7B-KTO.IQ4_NL.gguf	IQ4_NL	24.91GB
GritLM-8x7B-KTO.Q4_K_S.gguf	Q4_K_S	24.91GB
GritLM-8x7B-KTO.Q4_K.gguf	Q4_K	26.49GB
GritLM-8x7B-KTO.Q4_K_M.gguf	Q4_K_M	26.49GB
GritLM-8x7B-KTO.Q4_1.gguf	Q4_1	27.32GB
GritLM-8x7B-KTO.Q5_0.gguf	Q5_0	30.02GB
GritLM-8x7B-KTO.Q5_K_S.gguf	Q5_K_S	30.02GB
GritLM-8x7B-KTO.Q5_K.gguf	Q5_K	30.95GB
GritLM-8x7B-KTO.Q5_K_M.gguf	Q5_K_M	30.95GB
GritLM-8x7B-KTO.Q5_1.gguf	Q5_1	32.71GB
GritLM-8x7B-KTO.Q6_K.gguf	Q6_K	35.74GB
GritLM-8x7B-KTO.Q8_0.gguf	Q8_0	46.22GB

Original model description:

pipeline_tag: text-generation inference: true license: apache-2.0 datasets:

GritLM/tulu2

Model Summary

A KTO version of https://huggingface.co/GritLM/GritLM-8x7B

GritLM is a generative representational instruction tuned language model. It unifies text representation (embedding) and text generation into a single model achieving state-of-the-art performance on both types of tasks.

Repository: ContextualAI/gritlm
Paper: https://arxiv.org/abs/2402.09906
Logs: https://wandb.ai/muennighoff/gritlm/runs/0uui712t/overview
Script: https://github.com/ContextualAI/gritlm/blob/main/scripts/training/train_gritlm_7b.sh

Model	Description
GritLM 7B	Mistral 7B finetuned using GRIT
GritLM 8x7B	Mixtral 8x7B finetuned using GRIT

Use

The model usage is documented here.

Citation

@misc{muennighoff2024generative,
      title={Generative Representational Instruction Tuning}, 
      author={Niklas Muennighoff and Hongjin Su and Liang Wang and Nan Yang and Furu Wei and Tao Yu and Amanpreet Singh and Douwe Kiela},
      year={2024},
      eprint={2402.09906},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}