Going into local models? Find the best model for you ↓

When most people try using a local model for the first time, they either:

a) Find a hyped up big model, that's 70b parameters. Wait 30 mins just to end up with an error message.

b) Find a small model with some fancy benchmarks, run it, and stay with that forever.

One of the hardest parts about Local AI is finding a model that's made for you. Not everyone has a cluster of 4 5090's to run models on, most have less, some might have even more.

Everyone has a different use case, a content company doesn't need a fine tuned coding model, they need a general model that works good across tool use and all the general use cases.

Now the UI of this tool is quite confusing, but I recommend just giving Codex or Claude your use cases and then telling it to run the program and navigate through it.

It's called LLMFit, updated frequently and it has thousands of models.

Here's the link: https://github.com/AlexsJones/llmfit

0 comments

Going into local models? Find the best model for you ↓