You can deploy this model and start earning money today!

RichardErkhov/
MaLA-LM_-_lucky52-bloom-7b1-no-18-gguf

Name: MaLA-LM_-_lucky52-bloom-7b1-no-18-gguf
Availability: InStock

Views

No views yet

Quantization made by Richard Erkhov.

Github

Discord

Request more models

lucky52-bloom-7b1-no-18 - GGUF

Model creator: https://huggingface.co/MaLA-LM/
Original model: https://huggingface.co/MaLA-LM/lucky52-bloom-7b1-no-18/

Name	Quant method	Size
lucky52-bloom-7b1-no-18.Q2_K.gguf	Q2_K	3.2GB
lucky52-bloom-7b1-no-18.IQ3_XS.gguf	IQ3_XS	3.56GB
lucky52-bloom-7b1-no-18.IQ3_S.gguf	IQ3_S	3.63GB
lucky52-bloom-7b1-no-18.Q3_K_S.gguf	Q3_K_S	3.63GB
lucky52-bloom-7b1-no-18.IQ3_M.gguf	IQ3_M	3.9GB
lucky52-bloom-7b1-no-18.Q3_K.gguf	Q3_K	4.14GB
lucky52-bloom-7b1-no-18.Q3_K_M.gguf	Q3_K_M	4.14GB
lucky52-bloom-7b1-no-18.Q3_K_L.gguf	Q3_K_L	4.42GB
lucky52-bloom-7b1-no-18.IQ4_XS.gguf	IQ4_XS	4.33GB
lucky52-bloom-7b1-no-18.Q4_0.gguf	Q4_0	4.51GB
lucky52-bloom-7b1-no-18.IQ4_NL.gguf	IQ4_NL	4.53GB
lucky52-bloom-7b1-no-18.Q4_K_S.gguf	Q4_K_S	4.53GB
lucky52-bloom-7b1-no-18.Q4_K.gguf	Q4_K	4.91GB
lucky52-bloom-7b1-no-18.Q4_K_M.gguf	Q4_K_M	4.91GB
lucky52-bloom-7b1-no-18.Q4_1.gguf	Q4_1	4.92GB
lucky52-bloom-7b1-no-18.Q5_0.gguf	Q5_0	5.33GB
lucky52-bloom-7b1-no-18.Q5_K_S.gguf	Q5_K_S	5.33GB
lucky52-bloom-7b1-no-18.Q5_K.gguf	Q5_K	5.63GB
lucky52-bloom-7b1-no-18.Q5_K_M.gguf	Q5_K_M	5.63GB
lucky52-bloom-7b1-no-18.Q5_1.gguf	Q5_1	5.74GB
lucky52-bloom-7b1-no-18.Q6_K.gguf	Q6_K	6.2GB
lucky52-bloom-7b1-no-18.Q8_0.gguf	Q8_0	8.03GB

Original model description:

library_name: transformers pipeline_tag: text-generation language:

multilingual tags:
generation
question answering
instruction tuning datasets:
MBZUAI/Bactrian-X license: cc-by-nc-4.0

Model Description

This HF repository hosts instruction fine-tuned multilingual BLOOM model using the parallel instruction dataset called Bactrain-X in 52 languages. We progressively add a language during instruction fine-tuning at each time, and train 52 models in total. Then, we evaluate those models in three multilingual benchmarks.

Please refer to our paper for more details.

Base model: BLOOM 7B1
Instruction languages: English, Chinese, Afrikaans, Arabic, Azerbaijani, Bengali, Czech, German, Spanish, Estonian, Farsi, Finnish, French, Galician, Gujarati, Hebrew, Hindi, Croatian
Instruction language codes: en, zh, af, ar, az, bn, cs, de, es, et, fa, fi, fr, gl, gu, he, hi, hr
Training method: full-parameter fine-tuning.

Usage

The model checkpoint should be loaded using transformers library.

python

from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("MaLA-LM/lucky52-bloom-7b1-no-18")
model = AutoModelForCausalLM.from_pretrained("MaLA-LM/lucky52-bloom-7b1-no-18")

Citation

@inproceedings{ji2025lucky52,
      title={How Many Languages Make Good Multilingual Instruction Tuning? A Case Study on BLOOM}, 
      author={Shaoxiong Ji and Pinzhen Chen},
      year={2025},
      booktitle={Proceedings of COLING},
      url={https://arxiv.org/abs/2404.04850}, 
}

RichardErkhov/MaLA-LM_-_lucky52-bloom-7b1-no-18-gguf

Model Description

Usage

Citation

RichardErkhov/
MaLA-LM_-_lucky52-bloom-7b1-no-18-gguf