The case for 4-bit precision: k-bit Inference Scaling Laws
Jul 24, 2023
Sprecher:innen
Tim Dettmers
Řečník · 0 sledujících
Luke Zettlemoyer
Řečník · 5 sledujících
Über
Quantization methods reduce the number of bits required to represent each parameter in a model, trading accuracy for smaller memory footprints and inference latencies. However, the final model size depends on both the number of parameters of the original model and the rate of compression. For example, a 30B 8-bit model and a 60B 4-bit model have the same number of bits but may have very different zero-shot accuracies. In this work, we study this trade-off by developing inference scaling laws of zero-shot performance in Large Language Models (LLMs) to determine the bit-precision and model size that maximizes zero-shot performance. We run more than 35,000 experiments with 16-bit inputs and k-bit parameters to examine which zero-shot quantization methods improve scaling for 3 to 8-bit precision at scales of 19M to 176B parameters across the LLM families BLOOM, OPT, NeoX/Pythia, and GPT-2. We find that it is challenging to improve the bit-level scaling trade-off, with the only improvements being the use of a small block size – splitting the parameters into small independently quantized blocks – and the quantization data type being used (e.g., Int vs Float). Overall, our findings show that 4-bit precision is almost universally optimal for total model bits and zero-shot accuracyQuantization methods reduce the number of bits required to represent each parameter in a model, trading accuracy for smaller memory footprints and inference latencies. However, the final model size depends on both the number of parameters of the original model and the rate of compression. For example, a 30B 8-bit model and a 60B 4-bit model have the same number of bits but may have very different zero-shot accuracies. In this work, we study this trade-off by developing inference scaling laws of…
Gefällt euch das Format? Vertraut auf SlidesLive, um euer nächstes Event festzuhalten!
Professionelle Aufzeichnung und Livestreaming – weltweit.
Freigeben
Empfohlene Videos
Präsentationen, deren Thema, Kategorie oder Sprecher:in ähnlich sind
Universal Physics-Informed Neural Networks:
Zhlédnout později
Lena Podina, …
Revisiting Weighted Aggregation in Federated Learning with Neural Networks
Zhlédnout později
Zexi Li, …
Provable Reset-free Reinforcement Learning by No-Regret Reduction
Zhlédnout později
Minimum Width of Leaky-ReLU Neural Networks for Uniform Universal Approximation
Zhlédnout později
Liang Li, …
Invited Talk: Jack Balkin
Zhlédnout později
Low-Variance Gradient Estimation in Unrolled Computation Graphs with ES-Single
Zhlédnout později
Paul Vicol, …