Request: DOI

#18
by alesvigne - opened

I want to try this model

i wouldn't. i trained this a lot, and it seemed very static, and resistant to updating memory. my training set may not have been appropriate, but I've moved onto Qwen-2.5, and it trains well in contrast to distilbert2, understands context better, and responses are meaty and relevant.

Sign up or log in to comment