I’m currently shopping around for something a bit faster than ollama and because I could not get it to use a different context and output length, which seems to be a known and long ignored issue. Somehow everything I’ve tried so far did miss one or more critical features, like:
“Hot” model replacement, so loading and unloading models on demand
Function calling
Support of most models
OpenAI API compatibility (to work well with Open WebUI)
Are you using a tiny model (1.5B-7B parameters)? ollama pulls 4bit quant by default. It looks like vllm does not used quantized models by default so this is likely the difference. Tiny models are impacted more by quantization
I have no problems with changing num_ctx or num_predict
I’ve read about this method in the GitHub issues, but to me it seemed impractical to have different models just to change the context size, and that was the point I started looking for alternatives
You can overwrite the model by using the same name instead of creating one with a new name if it bothers you. Either way there is no duplication of the llm model file
You are not logged in. However you can subscribe from another Fediverse account, for example Lemmy or Mastodon. To do this, paste the following into the search field of your instance: !selfhosted@lemmy.world
A place to share alternatives to popular online services that can be self-hosted without giving up privacy or locking you into a service you don’t control.
Rules:
Be civil: we’re here to support and learn from one another. Insults won’t be tolerated. Flame wars are frowned upon.
No spam posting.
Posts have to be centered around self-hosting. There are other communities for discussing hardware or home computing. If it’s not obvious why your post topic revolves around selfhosting, please include details to make it clear.
Don’t duplicate the full text of your blog or github here. Just post the link for folks to click.
Submission headline should match the article title (don’t cherry-pick information from the title to fit your agenda).
No trolling.
Resources:
selfh.st Newsletter and index of selfhosted software and apps
Are you using a tiny model (1.5B-7B parameters)? ollama pulls 4bit quant by default. It looks like vllm does not used quantized models by default so this is likely the difference. Tiny models are impacted more by quantization
I have no problems with changing num_ctx or num_predict
It was multiple models, mainly 32-70B
Can you try setting the
num_ctxandnum_predictusing a Modelfile with ollama? https://github.com/ollama/ollama/blob/main/docs/modelfile.md#parameterI’ve read about this method in the GitHub issues, but to me it seemed impractical to have different models just to change the context size, and that was the point I started looking for alternatives
You can overwrite the model by using the same name instead of creating one with a new name if it bothers you. Either way there is no duplication of the llm model file