Qwen 2.5 1.5B retrain?
#12
by
tomaarsen
HF staff
- opened
Hello!
Can we expect a Qwen2.5 retrain of this model? Or perhaps also a 0.5B model? I think that's a very useful size.
Also, keep up the amazing work on the Qwen line - they are very very solid 👏
- Tom Aarsen
Thank you for your kind words and support for the Qwen line!
The gte-qwen2.5 series is already on its way 🚀🚀🚀
- DIngkun Long
Looking forward to it!
The gte-qwen2.5 series! Very welcome news!
Qwen2.5 - 0.5B would be super useful to have, the size would be more comparable with most embedding models then.
Is this converted using LLM2VEC approach? https://arxiv.org/abs/2404.05961, curious how you done this because would like to try Qwen2 Math as an embedding model...