license: mit
datasets:
- oscar-corpus/OSCAR-2301
- allenai/nllb
- Helsinki-NLP/opus-100
language:
- en
- hu
- el
- cs
- pl
- lt
- lv
base_model:
- haoranxu/X-ALMA-13B-Group5
X-ALMA-13B-Group5 — GGUF quantized variants
Original upload: X-ALMA-13B-Group5
This upload contains X-ALMA Group 5 (ALMA‑R base + merged LoRA for English (en), Hungarian (hu), Greek (el), Czech (cs), Polish (pl), Lithuanian (lt), and Latvian (lv)), quantized to GGUF format.
Files & sizes
Made with llama.cpp.
| File | Size (MB) |
|---|---|
| X-ALMA-13B-Group5.Q2_K.gguf | 4629.39 |
| X-ALMA-13B-Group5.Q3_K_L.gguf | 6608.54 |
| X-ALMA-13B-Group5.Q4_K_M.gguf | 7501.56 |
| X-ALMA-13B-Group5.Q5_K_M.gguf | 8802.34 |
Prompt format (example)
Translate this from {source_lang} to {target_lang}:
{source_lang}: {text_to_translate}
{target_lang}:
Replace the placeholders with your languages/text (e.g., using Python "..." .format(source_lang="English", target_lang="Polish", text_to_translate="I love machine translation.") or by manually editing the prompt).
Contact me if you'd like a different quant size or quantization of another model — I might be able to help.
Original description
X-ALMA builds upon ALMA-R by expanding support from 6 to 50 languages. It utilizes a plug-and-play architecture with language-specific modules, complemented by a carefully designed training recipe. This release includes the language-specific X-ALMA LoRA module and a merged model that supports the languages in Group 5: English (en), Hungarian (hu), Greek (el), Czech (cs), Polish (pl), Lithuanian (lt), and Latvian (lv).
@misc{xu2024xalmaplugplay,
title={X-ALMA: Plug & Play Modules and Adaptive Rejection for Quality Translation at Scale},
author={Haoran Xu and Kenton Murray and Philipp Koehn and Hieu Hoang and Akiko Eriguchi and Huda Khayrallah},
year={2024},
eprint={2410.03115},
archivePrefix={arXiv},
primaryClass={cs.CL},
url={https://arxiv.org/abs/2410.03115},
}