Pith. sign in

REVIEW 2 cited by

LightSeq: A High Performance Inference Library for Transformers

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2010.13887 v4 pith:LWI6LGAU submitted 2020-10-23 cs.MS cs.LG

classification cs.MScs.LG
keywords lightseqmodelstransformercomparedinferencelibrarytensorflowachieved
verification ladder T0 review T1 audit T2 compute T3 formal

Signed reviews

No signed human review yet.

0 comments
read the original abstract

Transformer, BERT and their variants have achieved great success in natural language processing. Since Transformer models are huge in size, serving these models is a challenge for real industrial applications. In this paper, we propose LightSeq, a highly efficient inference library for models in the Transformer family. LightSeq includes a series of GPU optimization techniques to to streamline the computation of neural layers and to reduce memory footprint. LightSeq can easily import models trained using PyTorch and Tensorflow. Experimental results on machine translation benchmarks show that LightSeq achieves up to 14x speedup compared with TensorFlow and 1.4x compared with FasterTransformer, a concurrent CUDA implementation. The code is available at https://github.com/bytedance/lightseq.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. SeaLLM: Service-Aware and Latency-Optimized Resource Sharing for Large Language Model Inference

    cs.DC 2025-04 conditional novelty 6.0 of 10

    SeaLLM shares GPU resources across multiple LLM services with a preemptive, service-characteristic-aware scheduler, search-based placement, adaptive replacement, and a merged-block unified KV cache, cutting normalized...

  2. Pie: Pooling CPU Memory for LLM Inference

    cs.LG 2024-11 conditional novelty 6.0 of 10

    Pie hides the latency of swapping KV cache layers between GPU and CPU memory behind normal computation, improving LLM serving throughput and memory efficiency on high-bandwidth interconnects.

Pith tools