lorax
lorax copied to clipboard
Multi-LoRA inference server that scales to 1000s of fine-tuned LLMs
### Feature request Perhaps it goes without saying that DeepSeek is the hot model at the moment. These models excel at reasoning, multilingual tasks and code generation. ### Motivation DeepSeek...
### Feature request Lorax should support various decoding strategies like Beam search. ### Motivation It would be invaluable when the user wants to generate multiple responses and at the same...
### Model description Are there any plans to add Qwen2VL model. I see that there was a past feature request for adding VLMs (#179) Are there any plans for this?...
THIS SHOULD BE CLOSED
Hi there! To add LGAI-EXAONE-3.5 model support to Lorax. I tested the result against the vLLM counterpart + an LoRA adapter and seemingly they were the same. **Update on the...
Hi, I am trying to confirm the version of the running server after running the docker container - `docker run ghcr.io/predibase/lorax:main` After I exec into the docker container and run...
### Model description Hi there! I would be grateful if we could have `LGAI-EXAONE/EXAONE-3.5-*` model [family](https://huggingface.co/collections/LGAI-EXAONE/exaone-35-674d0e1bb3dcd2ab6f39dbb4) supported by Lorax. ### Open source status - [x] The model implementation is available...