GPT4 does quite well with text translation. Unfortunately, the free version, it has limits on the input/translated text.
(Google translator and DeepL translate worse)
Is there any ranking available somewhere of Local LLMs ALE used to translate texts into other languages?
What FREE language models are available with a window context limit of at least 64k tokens or more (only such are suitable for text translation)?
Unless there is some way to somehow automatically split a long text into chunks and send them to the LLM for translation.
Well, no LLama model would come even close to GPT-4 in translation…
Try https://modernmt.com form Translated
https://www.idc.com/getdoc.jsp?containerId=US48325622
Way better than gpt4. LLM gives you fluency, MMT gives you accuracy.
" ModernMT cloud service can be accessed via API (for enterprises) or via plugin (for translators), after purchasing a license here. "
Perhaps this solution is better, but paid for.
There’s the ALMA models based on LLaMA 2:
https://huggingface.co/haoranxu/ALMA-13B
I’ve tried this for translating Japanese, seems pretty good: https://huggingface.co/mmnga/webbigdata-ALMA-7B-Ja-V2-gguf
I try TheBloke/ alma 13B q5_k_m.gguf in LLM Studio
Unfortunately, when translated from English into my language, the translation is VERY bad
You can try something like Claude.ai which has long context and is free to use.
You can use a python script to load the model, split the text into chunks, and ask the model to translate per chunk, then you don’t need a model with 64K context window (which will take up a lot of memory and are not that common).
It also depends on the language you are trying to translate, it would be best to find models that has been trained in the original language, most models have a large english corpus, with many finetuned with chinese data, but there are specialty models for German/arabic/japanese, try google search or find on hugging face.