Training Transformers with 4-bit Integers

Arxiv Papers
23/06/2023 44 min

Listen "Training Transformers with 4-bit Integers"

Descargar episodio Ver en sitio original

Episode Synopsis

The paper proposes a training method for transformers using int4 arithmetic, achieving competitive accuracy on various tasks, and can be implemented on current GPUs, speeding up training by up to 35.1%.



https://arxiv.org/abs//2306.11987



YouTube: https://www.youtube.com/@ArxivPapers



PODCASTS:


Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1692476016

Spotify: https://podcasters.spotify.com/pod/show/arxiv-papers

More episodes of the podcast Arxiv Papers

[QA] On the Theoretical Limitations of Embedding-Based Retrieval 01/09/2025
On the Theoretical Limitations of Embedding-Based Retrieval 01/09/2025
[QA] Beyond GPT-5: Making LLMs Cheaper and Better via Performance–Efficiency Optimized Routing 22/08/2025
Beyond GPT-5: Making LLMs Cheaper and Better via Performance–Efficiency Optimized Routing 22/08/2025
[QA] Measuring the environmental impact of delivering AI at Google Scale 22/08/2025
Measuring the environmental impact of delivering AI at Google Scale 22/08/2025
[QA] Deep Think with Confidence 22/08/2025
Deep Think with Confidence 22/08/2025
[QA] Intern-S1: A Scientific     Multimodal Foundation Model   22/08/2025
Intern-S1: A Scientific     Multimodal Foundation Model   22/08/2025
© 2003-2026 ZARZA | AHEAD OF OUR TIME
privacy policy
In God we trust In God we trust
PreSales Questions? Click Here
or call our toll-free line at +1-888-5002068 Miami, Florida, USA