[CamperBob2]

Try the 27B dense model. It will likely do much better than the 35b MoE with only 3B active experts.

Also, performance on research-y questions isn't always a good indicator of how the model will do for code generation or agent orchestration.

[regularfry]

Currently sat waiting for the unsloth fixed quants to drop, but I'm on the edge of my seat for this.

[Balinares]

Wait, didn't they drop like two days ago?

[regularfry]

The 35b did but not the 27b. Looks like the latter has been updated in the last half hour.

[Balinares]

Neat! Thanks for correcting me there. I'll go and take a look.