CausalMix: Data Mixture as Causal Inference for Language Model Training
CausalMix addresses limitations in LLM data mixing by formulating mixture optimization as a causal inference problem, enabling dynamic adaptation to shifting data distributions wit…
Hugging Face · Daily Papers
·Zinan Tang, Yukun Zhang
·
·▲ 11 upvotes
Este artigo está em destaque na seleção diária de papers do Hugging Face, curada pela comunidade de pesquisa em IA.
Autores: Zinan Tang, Yukun Zhang, Shaomian Zheng, Zhuoshi Pan, Qizhi Pei, Dingnan Jin
- 11 upvotes da comunidade
- Temas: data mixing, causal inference, conditional average treatment effect, causal modeling, confounding biases, data pool
Resumo
Resumo original (em inglês), extraído do paper:
CausalMix addresses limitations in LLM data mixing by formulating mixture optimization as a causal inference problem, enabling dynamic adaptation to shifting data distributions without costly retraining.Onde ler
// relacionados
Leia também
Editorial
Claude Sonnet 5: a Anthropic aposta que o modelo do meio faz o trabalho do topo
Blog
Google’s AI buildout drove 37% increase in electricity use in 2025
Blog
OpenAI reportedly offers the Trump administration a five percent stake in the company
Blog