LLM.int8(): 8-bit Matrix Multiplication for Transformers at Scale

15 Aug 2022  ·  Tim Dettmers, Mike Lewis, Younes Belkada, Luke Zettlemoyer ·

Large language models have been widely adopted but require significant GPU memory for inference. We develop a procedure for Int8 matrix multiplication for feed-forward and attention projection layers in transformers, which cut the memory needed for inference by half while retaining full precision performance. With our method, a 175B parameter 16/32-bit checkpoint can be loaded, converted to Int8, and used immediately without performance degradation. This is made possible by understanding and working around properties of highly systematic emergent features in transformer language models that dominate attention and transformer predictive performance. To cope with these features, we develop a two-part quantization procedure, LLM.int8(). We first use vector-wise quantization with separate normalization constants for each inner product in the matrix multiplication, to quantize most of the features. However, for the emergent outliers, we also include a new mixed-precision decomposition scheme, which isolates the outlier feature dimensions into a 16-bit matrix multiplication while still more than 99.9% of values are multiplied in 8-bit. Using LLM.int8(), we show empirically it is possible to perform inference in LLMs with up to 175B parameters without any performance degradation. This result makes such models much more accessible, for example making it possible to use OPT-175B/BLOOM on a single server with consumer GPUs.

PDF Abstract

Results from the Paper


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Language Modelling C4 Zeropoint LLM.int8 (vector-wise + decomp, 13B) Perplexity 12.45 # 2
Linguistic Acceptability CoLA Vector-wise Accuracy 68.6% # 8
Semantic Textual Similarity MRPC Vector-wise Accuracy 91.0% # 6
Natural Language Inference MultiNLI Vector-wise Matched 90.2 # 9
Natural Language Inference QNLI Vector-wise Accuracy 94.7% # 12
Natural Language Inference RTE Vector-wise Accuracy 85.4% # 13
Sentiment Analysis SST-2 Binary classification Vector-wise Accuracy 96.4 # 16
Semantic Textual Similarity STS Benchmark Vector-wise Pearson Correlation 0.919 # 9

Methods


No methods listed for this paper. Add relevant methods here