HackerTrans
トップ新着トレンドコメント過去質問紹介求人

mluo

no profile record

投稿

DeepScaleR: Surpassing O1-Preview with a 1.5B Model by Scaling RL

pretty-radio-b75.notion.site
19 ポイント·投稿者 mluo·昨年·0 コメント

コメント

mluo
·昨年·議論
Check out one of my prior work: https://stylus-diffusion.github.io/

This work scales up selection/routing over many models/LoRAs
mluo
·昨年·議論
For quantization, very big impact for small models, can drop at much as 10% on AIME. Our model does best on bfloat16 ;)

Come checkout our repo at: https://github.com/agentica-project/deepscaler
mluo
·昨年·議論
It's simply bc the model is small (1.5B), making it sensitive to weight perturbations
mluo
·昨年·議論
Think there are some people who made GGUFs as branches of our model, try it out!

https://huggingface.co/models?other=base_model:quantized:age...
mluo
·昨年·議論
Nice, very glad to see it works! Small models are very sensitive to the dtype :(
mluo
·昨年·議論
Try bfloat16! We have a bug where the model was saved as fp32.
mluo
·昨年·議論
We beat O1-preview and even many other 7B models over many math benchmarks, which was TEST set (not in training set at all).

If you want to make the model fully generalist, feel free to train it over coding datasets (such as RL with passing unit tests as reward).
mluo
·昨年·議論
One of the authors here....

This is not a Chinese model, btw I'm American
mluo
·昨年·議論
Hi, one of the lead authors for this work.

We recommend using Bfloat16 (not fp16), quantization for small models can really hurt performance!
mluo
·3 年前·議論
Alpaca Llama Vicuna -> Gorilla

Chad move
mluo
·3 年前·議論
With inflation in mind, wouldn't there a larger gap between Ray's sort and the previous WR holder?