languagemodels
languagemodels copied to clipboard
jncraton
Explore large language models in 512MB of RAM
Please add simple assistant interactions like a 'mycroft' what is time calculate 2+18+9 what is weather etc. simple add a script/object to resolve specific data/know and using it together.
when i input Chinese, like: **lm.do("你好")** it returns ' ⁇ , ⁇ , ⁇ ' It seems that the current model does not support Chinese Is there any recommended Chinese...
We could support more translation directions with [m2m-100](https://opennmt.net/CTranslate2/guides/transformers.html#m2m-100) in cTranslate or use [easy translate](https://github.com/ikergarcia1996/Easy-Translate).
It will be wonderfull if there were Helper functions for external retrieval like: - lm.get_URL(path) for using an URL - lm_get_MD(path) for using Markdown document - lm_get_PDF(path) for using PDF...
It is currently easy to cause an out-of-memory condition when prompting a model with a very long prompt. This is an expected result of the implementation of both certain tokenizers...
Smaller [1.6B](https://huggingface.co/stabilityai/stablelm-2-zephyr-1_6b) and 3B models are available from Stability AI that appear to have good performance characteristics for their size. It may be worth adding support for these, but this...
Its a really good model for its size and it aligns with the goal of this project.
