Different Model Suggestion
#4
by
Bearsaerker
- opened
Thank you for your great work!
I'm currently gpu poor (only a RTX 3060) and I'd love to suggest a different model which I sadly can't train myself. The new glm-4 has a big context size and I think it would be perfect for rag, but I like the answer style and intelligence of the SPPO trained models (especially the gemma 9b) way better. SPPO is in my testing way better for rag use cases and I'd love to see that. Is there a possibility for getting a glm-4 sppo from you guys? Would be phenomenal!