I am impressed by how that model can navigate web pages and analyze data.
Unfortunately, when that model generates a report, it makes even more spelling mistakes than I do.
On the other hand, Bielik-11B-v3.0-Instruct is not as smart and capable as the Holo model - there is no way I would use it as the main “brain” of my local agent.
But guess what? There is a way to make those two models cooperate in a way that each acts in the area it absolutely dominates!
Hermes agent, powered by Holo-3.1-35B-A3B, uses various browsers to navigate the realms of the World Wide Web and gather data of interest.
Once it obtains the required information, it invokes a separate, standalone worker powered by Bielik-11B-v3.0-Instruct to execute spell and grammar correction.
As a result, I get a “perfectly gathered” and “perfectly written” report.
By the way, all of the above executes on my local hardware 24/7 at no cost.
Lessons learned: There is no need for a universally perfect model as long as you can organize a team that delivers expected results.
👉 Define your business goal, then together we will define your tooling.