December 20, 2024
GitHub – huggingface/search-and-learn
Something worth learning painful lesson It is the power of general-purpose methods that they can continue to scale as the computation increases, even if the available computation becomes very large. Two methods that seem to arbitrarily scale in this way are search and study.
conda create -n sal python=3.10 && conda activate sal
this Recipe readme Startup commands and configuration files are included to allow replication of our results.
If you use this code library or blog post, it would be great if you could quote us:
@misc{beeching2024scalingtesttimecompute,
title={Scaling test-time compute with open models},
author={Edward Beeching and Lewis Tunstall and Sasha Rush},
url={https://huggingface.co/spaces/HuggingFaceH4/blogpost-scaling-test-time-compute},
}
Please also cite this work:
@misc{snell2024scalingllmtesttimecompute,
title={Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters},
author={Charlie Snell and Jaehoon Lee and Kelvin Xu and Aviral Kumar},
year={2024},
eprint={2408.03314},
archivePrefix={arXiv},
primaryClass={cs.LG},
url={https://arxiv.org/abs/2408.03314},
}
2024-12-16 20:35:45