Petals: Collaborative Inference and Fine-tuning of Large Models
This addresses the accessibility issue for researchers lacking high-end hardware by providing a flexible alternative to offloading and APIs, though it is incremental as it builds on existing collaborative and fine-tuning methods.
The authors tackled the problem of high hardware requirements for using large language models (LLMs) by proposing Petals, a collaborative system for inference and fine-tuning that runs BLOOM-176B on consumer GPUs at about 1 step per second, enabling interactive applications and access to model weights.
Many NLP tasks benefit from using large language models (LLMs) that often have more than 100 billion parameters. With the release of BLOOM-176B and OPT-175B, everyone can download pretrained models of this scale. Still, using these models requires high-end hardware unavailable to many researchers. In some cases, LLMs can be used more affordably via RAM offloading or hosted APIs. However, these techniques have innate limitations: offloading is too slow for interactive inference, while APIs are not flexible enough for research that requires access to weights, attention or logits. In this work, we propose Petals - a system for inference and fine-tuning of large models collaboratively by joining the resources of multiple parties. We demonstrate that this strategy outperforms offloading for very large models, running inference of BLOOM-176B on consumer GPUs with $\approx$ 1 step per second, which is enough for many interactive LLM applications. Unlike most inference APIs, Petals also natively exposes hidden states of served models, allowing to train and share custom model extensions based on efficient fine-tuning methods.