Papers › How to Train BERT with an Academic Budget

How to Train BERT with an Academic Budget

15 Apr 2021EMNLP 2021 11arXiv:2104.07705archive 2025-07-28

Peter Izsak, Moshe Berchansky, Omer Levy

While large language models a la BERT are used ubiquitously in NLP, pretraining them is considered a luxury that only a few well-funded industry labs can afford. How can one train such models with a more modest budget? We present a recipe for pretraining a masked language model in 24 hours using a single low-end deep learning server. We demonstrate that through a combination of software optimizations, design choices, and hyperparameter tuning, it is possible to produce models that are competitive with BERT-base on GLUE tasks at a fraction of the original pretraining cost.

Paper PDF Conference PDF Code

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

IntelLabs/academic-budget-bert officialmentioned in papermentioned on GitHubpytorchnot reachable when probed 2026-09-17 — repositories for recent papers often appear after camera-ready report

peteriz/academic-budget-bert officialmentioned in papermentioned on GitHubpytorchnot reachable when probed 2026-09-17 — repositories for recent papers often appear after camera-ready report

octanove/shiba mentioned on GitHubpytorchNOASSERTION report

yxzwang/normalized-information-payload mentioned on GitHubpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Language Modeling Language Modelling Linguistic Acceptability Natural Language Inference Question Answering Semantic Textual Similarity Sentiment Analysis

Results from the paper archive 2025-07-28

Task	Dataset	Model	Metric	Value	Rank at snapshot	Leaderboard	Report
Linguistic Acceptability	CoLA	24hBERT	Accuracy	57.1	#33 of 43	Archive leaderboard	report
Natural Language Inference	MultiNLI	24hBERT	Matched	84.4	#33 of 67	Archive leaderboard	report
Natural Language Inference	MultiNLI	24hBERT	Mismatched	83.8	#33 of 67	Archive leaderboard	report
Natural Language Inference	QNLI	24hBERT	Accuracy	90.6	#32 of 43	Archive leaderboard	report
Natural Language Inference	RTE	24hBERT	Accuracy	57.7%	#79 of 90	Archive leaderboard	report
Question Answering	Quora Question Pairs	24hBERT	Accuracy	70.7	#19 of 19	Archive leaderboard	report
Semantic Textual Similarity	MRPC	24hBERT	Accuracy	87.5%	#27 of 45	Archive leaderboard	report
Semantic Textual Similarity	STS Benchmark	24hBERT	Pearson Correlation	0.820	#27 of 66	Archive leaderboard	report
Sentiment Analysis	SST-2 Binary classification	24hBERT	Accuracy	93.0	#44 of 87	Archive leaderboard	report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

Adam Attention Attention Dropout BERT Dense Connections Dropout Layer Normalization Linear Layer Linear Warmup With Linear Decay Multi-Head Attention Residual Connection Softmax Weight Decay WordPiece

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections