Faster assisted generation support for Intel Gaudi

📰 Hugging Face Blog

Hugging Face optimizes assisted decoding for Intel Gaudi, reducing latency and costs in text generation tasks

advanced Published 4 Jun 2024
Action Steps
  1. Understand the importance of inference optimizations for text generation
  2. Explore assisted decoding as a method for speeding up text generation
  3. Optimize assisted decoding for Intel Gaudi using Hugging Face's adaptations
Who Needs to Know This

AI engineers and data scientists can benefit from this optimization to improve the efficiency of their text generation models, while product managers can consider the cost savings and improved user experience

Key Insight

💡 Assisted decoding can significantly reduce latency and costs in text generation tasks, making it an essential optimization technique for AI implementations

Share This
🚀 Faster text generation with Intel Gaudi! 🚀

Key Takeaways

Hugging Face optimizes assisted decoding for Intel Gaudi, reducing latency and costs in text generation tasks

Full Article

Published Time: 2024-06-04T00:00:00.401Z

# Faster assisted generation support for Intel Gaudi

[![Image 1: Hugging Face's logo](https://huggingface.co/front/assets/huggingface_logo-noborder.svg)Hugging Face](https://huggingface.co/)

* [Models](https://huggingface.co/models)
* [Datasets](https://huggingface.co/datasets)
* [Spaces](https://huggingface.co/spaces)
* [Buckets new](https://huggingface.co/storage)
* [Docs](https://huggingface.co/docs)
* [Enterprise](https://huggingface.co/enterprise)
* [Pricing](https://huggingface.co/pricing)
*
*
* * *

* [Log In](https://huggingface.co/login)
* [Sign Up](https://huggingface.co/join)

[Back to Articles](https://huggingface.co/blog)

# [](https://huggingface.co/blog/assisted-generation-support-gaudi#faster-assisted-generation-support-for-intel-gaudi) Faster assisted generation support for Intel Gaudi

Published June 4, 2024

[Update on GitHub](https://github.com/huggingface/blog/blob/main/assisted-generation-support-gaudi.md)

[- [x] Upvote 3](https://huggingface.co/login?next=%2Fblog%2Fassisted-generation-support-gaudi)
* [![Image 2](https://cdn-avatars.huggingface.co/v1/production/uploads/6340651b388c3fa40f9a5bc0/vM3rB17pUNT11MUhYqfFY.png)](https://huggingface.co/adamm-hf "adamm-hf")
* [![Image 3](https://huggingface.co/avatars/b2af8fd022114887c9328686cbf128e7.svg)](https://huggingface.co/benzakenelad "benzakenelad")
* [![Image 4](https://cdn-avatars.huggingface.co/v1/production/uploads/656af5e99c8778992f669423/hYrkG50PRPpCA22jI4BUQ.jpeg)](https://huggingface.co/Alan009 "Alan009")

[![Image 5: Haim Barad's avatar](https://cdn-avatars.huggingface.co/v1/production/uploads/1649762381314-noauth.jpeg)](https://huggingface.co/haimbarad)

[Haim Barad haimbarad Follow](https://huggingface.co/haimbarad)

[![Image 6: Intel's avatar](https://cdn-avatars.huggingface.co/v1/production/uploads/1616186257611-60104afcc75e19ac1738fe70.png)](https://huggingface.co/Intel "Intel")[Intel](https://huggingface.co/Intel)

[![Image 7: Neha Raste's avatar](https://huggingface.co/avatars/0ac87abe33228a6f789033f4980ea154.svg)](https://huggingface.co/neharaste)

[Neha Raste neharaste Follow](https://huggingface.co/neharaste)

[![Image 8: Intel's avatar](https://cdn-avatars.huggingface.co/v1/production/uploads/1616186257611-60104afcc75e19ac1738fe70.png)](https://huggingface.co/Intel "Intel")[Intel](https://huggingface.co/Intel)

[![Image 9: Tien Pei Chou's avatar](https://cdn-avatars.huggingface.co/v1/production/uploads/6306da2a41bf1fbadaf33a4e/JTPZxhOU27GEOzQVIsG53.jpeg)](https://huggingface.co/joeychou)

[Tien Pei Chou joeychou Follow](https://huggingface.co/joeychou)

[![Image 10: Intel's avatar](https://cdn-avatars.huggingface.co/v1/production/uploads/1616186257611-60104afcc75e19ac1738fe70.png)](https://huggingface.co/Intel "Intel")[Intel](https://huggingface.co/Intel)

* [Speculative Sampling - Assisted Decoding](https://huggingface.co/blog/assisted-generation-support-gaudi#speculative-sampling---assisted-decoding "Speculative Sampling - Assisted Decoding")

* [Usage & Experiments](https://huggingface.co/blog/assisted-generation-support-gaudi#usage--experiments "Usage & Experiments")

* [Conclusion](https://huggingface.co/blog/assisted-generation-support-gaudi#conclusion "Conclusion")

As model sizes grow, Generative AI implementations require significant inference resources. This not only increases the cost per generation, but also increases the power consumption used to serve such requests.
Inference optimizations for text generation are essential for reducing latency, infrastructure costs, and power consumption. This can lead to an improved user experience and increased efficiency in text generation tasks.

Assisted decoding is a popular method for speeding up text generation. We adapted and optimized it for Intel Gaudi, which delivers similar performance as Nvidia H100 GPUs as shown in [a previous post](https://huggingface.co/blog/bridgetower), while its price
Read full article → ← Back to Reads

Related Videos

5 Levels of AI Agents - From Simple LLM Calls to Multi-Agent Systems
5 Levels of AI Agents - From Simple LLM Calls to Multi-Agent Systems
Dave Ebbelaar (LLM Eng)
Kimi K3: The Free AI That Just Beat Claude at Coding (Ranked #1)
Kimi K3: The Free AI That Just Beat Claude at Coding (Ranked #1)
AI Andy
GLM-5.2 Is INSANE – Is it The BEST New Open Source Model?
GLM-5.2 Is INSANE – Is it The BEST New Open Source Model?
AI Andy
Watch Fable 5 Burn 2.7M Tokens On My Broken AI Video Editor
Watch Fable 5 Burn 2.7M Tokens On My Broken AI Video Editor
AI Andy
EVERY Loop From Matthew Berman's New Loop Library! (Copy & Paste!)
EVERY Loop From Matthew Berman's New Loop Library! (Copy & Paste!)
AI Andy
Ollama + OpenWebUI: Run LLM's Locally For FREE!!
Ollama + OpenWebUI: Run LLM's Locally For FREE!!
Thomas Janssen