Anthonious

Would be interesting to see how a specialised model does against a model with more params, I use qwen8b in my HA and would love to downgrade to 1.7B and optimise HA... But you have no stats or comparisons