Truly great model for text-based operations like analysing and researching

#56
by bkieser - opened

Adding this comment here because I have tried a very large number of LLMs for a research workflow using gpt researcher and by far this model has given the best results. Nothing else comes anywhere close to it, not even Llama 3 which is quite something.

Just goes to show that the benchmarks are only a guide, real world usage is quite often a very different experience.

Adding this comment here because I have tried a very large number of LLMs for a research workflow using gpt researcher and by far this model has given the best results. Nothing else comes anywhere close to it, not even Llama 3 which is quite something.

Just goes to show that the benchmarks are only a guide, real world usage is quite often a very different experience.

Agreed, I passed over this model until it started to get some buzz. It is my go-to model now and I keep coming back no matter what else I try.

Cohere For AI org

Thanks so much for sharing your experience @bkieser @BigHuggyD ! Glad you're enjoying working with the command-r :D

I realize now this is for command-r instead of command-r-plus but the observation is the same. I use command-r-plus

I realize now this is for command-r instead of command-r-plus but the observation is the same. I use command-r-plus

I don't have the firepower to run +, but plain r runs sweet as can be on both my RTX 4090 and also my twin Radeon 7900 XTX GPUs. I may try + under llamacpp CPU with offloading. A lot to be said for cheap PC RAM!

Sign up or log in to comment