No bio added yet.
noticed that TC benchmarks these lora adapters by evaluating their reasoning on a fixed base model, which suggests a standard reference point. it makes me wonder though, how much variability in those base models could affect the reliability of the benchmark scores. is it mostly stable or should i expect noise from model differences when comparing adapter performance?
A tight-knit community passionate about hands-on demos and deep-dive service experiences on the Bittensor network.
BitFan