Papers › Well-Read Students Learn Better: On the Importance of Pre-training Compact Models

Well-Read Students Learn Better: On the Importance of Pre-training Compact Models

23 Aug 2019ICLR 2020 1arXiv:1908.08962archive 2025-07-28

Iulia Turc, Ming-Wei Chang, Kenton Lee, Kristina Toutanova

Recent developments in natural language representations have been accompanied by large and expensive models that leverage vast amounts of general-domain text through self-supervised pre-training. Due to the cost of applying such models to down-stream tasks, several model compression techniques on pre-trained language representations have been proposed (Sun et al., 2019; Sanh, 2019). However, surprisingly, the simple baseline of just pre-training and fine-tuning compact models has been overlooked. In this paper, we first show that pre-training remains important in the context of smaller architectures, and fine-tuning pre-trained compact models can be competitive to more elaborate methods proposed in concurrent work. Starting with pre-trained compact models, we then explore transferring task knowledge from large fine-tuned models through standard knowledge distillation. The resulting simple, yet effective and general algorithm, Pre-trained Distillation, brings further improvements. Through extensive experiments, we more generally explore the interaction between pre-training and distillation under two variables that have been under-studied: model size and properties of unlabeled task data. One surprising observation is that they have a compound effect even when sequentially applied on the same data. To accelerate future research, we will make our 24 pre-trained miniature BERT models publicly available.

PaperPDFConference PDFCodeCode Syntology ran

In Syntology View this paper on Syntology, its page in Syntology's graph. That page adds which of its functions are also present in other papers' code Syntology holds, with dates, and its dated run record as JSON and a README badge. It also lists the repositories linked to the paper, the abstract and the calls for agents.

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

For agents, on Syntology's MCP service (how to connect):

Code

No sample from the archive's official repository (google-research/bert) is linked to this paper; the 32 samples below come from 12 community repositories. Syntology harvested code from google-research/bert for another paper.

google-research/bert: Syntology's graph links these samples to NapSS: Paragraph-level Medical Text Simplification via Narrative Prompting and... (arXiv:2302.05574; that paper's own run record: 17 ran (of which 0 constructed an object rather than computing a result; 13 with no instrument failure: 1 honoured, 0 violated, 12 with no contract checked; 4 where Syntology's instrument failed) · 7 unverified); nothing checks that they implement this paper's method.

Syntology 32 code samples harvested from 12 repositories linked to this paper: 21 ran (of which 0 constructed an object rather than computing a result; 15 with no instrument failure: 1 honoured, 0 violated, 14 with no contract checked; 6 where Syntology's instrument failed) · 11 unverified. Where Syntology's instrument failed: 6 where our contract draft was wrong; failures of Syntology's instrument, not of the code.

By repository: community (archive-listed): 32 samples from 12 repositories, 21 ran (15 with no instrument failure; 6 where Syntology's instrument failed). The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

No sample is linked to this paper from 9 of its repositories that Syntology harvested for other papers:

Levels written between 2026-09-22 and 2026-09-26 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

40 repositories listed; official and paper-mentioned ones first.

google-research/bert officialmentioned on GitHubtfApache-2.0Syntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to NapSS: Paragraph-level Medical Text Simplification via Narrative Prompting and... (arXiv:2302.05574; that paper's own run record: 17 ran (of which 0 constructed an object rather than computing a result; 13 with no instrument failure: 1 honoured, 0 violated, 12 with no contract checked; 4 where Syntology's instrument failed) · 7 unverified); nothing checks that they implement this paper's method. report
Arthurizijar/Bert_Airport mentioned on GitHubtfApache-2.0Syntology: 10 samples, 7 ran (3 instrument-failed) · 3 unverified report
Caesarzhang/bert-zh mentioned on GitHubtfApache-2.0Syntology: not harvested report
Chonwai/Learning_BERT mentioned on GitHubtfApache-2.0Syntology: 3 samples, 0 ran · 3 unverified report
DeligientSloth/QQsim mentioned on GitHubtfSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding (arXiv:1810.04805; that paper's own run record: 300 ran (of which 75 constructed an object rather than computing a result; 234 with no instrument failure: 18 honoured, 5 violated, 211 with no contract checked; 66 where Syntology's instrument failed) · 359 unverified); nothing checks that they implement this paper's method. report
DeligientSloth/bert-tensorflow mentioned on GitHubtfSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding (arXiv:1810.04805; that paper's own run record: 300 ran (of which 75 constructed an object rather than computing a result; 234 with no instrument failure: 18 honoured, 5 violated, 211 with no contract checked; 66 where Syntology's instrument failed) · 359 unverified); nothing checks that they implement this paper's method. report
MOHAMEDELDAKDOUKY/bert_adjusted mentioned on GitHubtfApache-2.0Syntology: 1 sample, 0 ran · 1 unverified report
Maz101/Bert mentioned on GitHubtfApache-2.0Syntology: 2 samples, 1 ran (0 instrument-failed) · 1 unverified report
Maz101/pretrainingdataset mentioned on GitHubtfApache-2.0Syntology: not harvested report
PAIR-code/lit mentioned on GitHubpytorchApache-2.0Syntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to The Language Interpretability Tool: Extensible, Interactive Visualizations and... (arXiv:2008.05122; that paper's own run record: 8 ran (of which 0 constructed an object rather than computing a result; 8 with no instrument failure: 0 honoured, 0 violated, 8 with no contract checked; 0 where Syntology's instrument failed) · 1 unverified); nothing checks that they implement this paper's method. report
RenXiangyuan/tf_bert mentioned on GitHubtfApache-2.0Syntology: not harvested report
SpikeKing/My-Bert mentioned on GitHubtfApache-2.0Syntology: 1 sample, 1 ran (0 instrument-failed) · 0 unverified report
StoneGH/bert mentioned on GitHubtfApache-2.0Syntology: 2 samples, 2 ran (0 instrument-failed) · 0 unverified report
TonyX19/bert_hyperpartisan mentioned on GitHubtfApache-2.0Syntology: 2 samples, 2 ran (0 instrument-failed) · 0 unverified report
Walter-B/bert-20-classes mentioned on GitHubtfApache-2.0Syntology: not harvested report
XINZXZHANG/Bert_repo mentioned on GitHubtfApache-2.0Syntology: not harvested report
abhiagg92/bert_sentiment mentioned on GitHubtfSyntology: not harvested report
coronazap/bert_client mentioned on GitHubtfApache-2.0Syntology: 4 samples, 2 ran (0 instrument-failed) · 2 unverified report
cospplay/bert-master mentioned on GitHubtfApache-2.0Syntology: not harvested report
cuber2460/bert mentioned on GitHubtfApache-2.0Syntology: not harvested report
dzqjorking/transpose mentioned on GitHubtfSyntology: not harvested report
geondopark/ckd mentioned on GitHubpytorchMITSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to Distilling Linguistic Context for Language Model Compression (arXiv:2109.08359; that paper's own run record: 8 ran (of which 0 constructed an object rather than computing a result; 7 with no instrument failure: 0 honoured, 0 violated, 7 with no contract checked; 1 where Syntology's instrument failed) · 3 unverified); nothing checks that they implement this paper's method. report
goodluck4s/bert- mentioned on GitHubtfSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to Attention Is All You Need (arXiv:1706.03762; that paper's own run record: 610 ran (of which 293 constructed an object rather than computing a result; 529 with no instrument failure: 45 honoured, 15 violated, 469 with no contract checked; 81 where Syntology's instrument failed) · 336 unverified) and BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding (arXiv:1810.04805; that paper's own run record: 300 ran (of which 75 constructed an object rather than computing a result; 234 with no instrument failure: 18 honoured, 5 violated, 211 with no contract checked; 66 where Syntology's instrument failed) · 359 unverified); nothing checks that they implement this paper's method. report
google-research/bleurt mentioned on GitHubtfApache-2.0Syntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to FEA-SLT: A Gloss-Free End-to-End Framework for Facial-Expression-Aware Sign Language... (arXiv:2601.03549; that paper's own run record: 12 ran (of which 0 constructed an object rather than computing a result; 4 with no instrument failure: 0 honoured, 1 violated, 3 with no contract checked; 8 where Syntology's instrument failed) · 0 unverified); nothing checks that they implement this paper's method. report
google-research/tapas mentioned on GitHubtfSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to DoT: An efficient Double Transformer for NLP tasks with tables (arXiv:2106.00479; that paper's own run record: 5 ran (of which 1 constructed an object rather than computing a result; 1 with no instrument failure: 0 honoured, 0 violated, 1 with no contract checked; 4 where Syntology's instrument failed) · 4 unverified) and TableFormer: Robust Transformer Modeling for Table-Text Encoding (arXiv:2203.00274; that paper's own run record: 1 ran (of which 0 constructed an object rather than computing a result; 0 with no instrument failure: 0 honoured, 0 violated, 0 with no contract checked; 1 where Syntology's instrument failed) · 0 unverified); nothing checks that they implement this paper's method. report
halo090770/bert mentioned on GitHubtfApache-2.0Syntology: 1 sample, 1 ran (0 instrument-failed) · 0 unverified report
krantirk/BERT-Pretrained-model mentioned on GitHubtfApache-2.0Syntology: not harvested report
llx666/git_demo mentioned on GitHubtfApache-2.0Syntology: not harvested report
paolanu/BERT_epitope mentioned on GitHubtfApache-2.0Syntology: 1 sample, 1 ran (0 instrument-failed) · 0 unverified report
pingheng001/Cnn-Bert mentioned on GitHubtfApache-2.0Syntology: not harvested report
ricardordb/bert mentioned on GitHubtfSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to Attention Is All You Need (arXiv:1706.03762; that paper's own run record: 610 ran (of which 293 constructed an object rather than computing a result; 529 with no instrument failure: 45 honoured, 15 violated, 469 with no contract checked; 81 where Syntology's instrument failed) · 336 unverified) and BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding (arXiv:1810.04805; that paper's own run record: 300 ran (of which 75 constructed an object rather than computing a result; 234 with no instrument failure: 18 honoured, 5 violated, 211 with no contract checked; 66 where Syntology's instrument failed) · 359 unverified); nothing checks that they implement this paper's method. report
rohan400/PepperQA mentioned on GitHubtfApache-2.0Syntology: not harvested report
somiltg/bert mentioned on GitHubtfApache-2.0Syntology: 3 samples, 3 ran (3 instrument-failed) · 0 unverified report
the-Quert/iNLPfun mentioned on GitHubtfApache-2.0Syntology: 2 samples, 1 ran (0 instrument-failed) · 1 unverified report
theQuert/inlpfun mentioned on GitHubtfApache-2.0Syntology: not harvested report
thecodemasterk/BERT mentioned on GitHubtfSyntology: not harvested report
thousandvoices/ok_ml_cup mentioned on GitHubpytorchSyntology: not harvested report
xiaopp123/bert_explain mentioned on GitHubtfApache-2.0Syntology: not harvested report
xitianxiaofeixue/BERT mentioned on GitHubtfApache-2.0Syntology: not harvested report
yuhangT/tf_bert mentioned on GitHubtfApache-2.0Syntology: not harvested 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.

A paper named beside a repository with no sample linked to this paper is shown with that paper's own run record, not this paper's: “ran” means executed on a synthesized input, not that the code is correct, and “where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Papers are listed in arXiv-id order, at most three per repository.

Code Syntology ran Syntology

32 samples harvested: 21 ran (of which 0 constructed an object rather than computing a result; 15 with no instrument failure: 1 honoured, 0 violated, 14 with no contract checked; 6 where Syntology's instrument failed) · 11 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page last changed ; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

1ran · honoured contract
6ran · our draft was wrong
14ran
11unverified

Licence: 4 of the 32 samples are pointer only, meaning Syntology does not serve that copy's text. For 4 of them, this paper's copy is under a permissive licence and cleared, but the stored copy of the same code that Syntology serves is not; the licence label's tooltip gives Syntology's recorded reason. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from 12 repositories linked to this paper, official or community; the list below is grouped by repository, each group headed by its name and what it is to the paper. “Ran” means the sample executed on a synthesized input. It does not mean the output is correct, and nothing here reproduces the paper's results. A line beginning “fixture:” under a sample that ran gives the arguments Syntology's fixture called the sample with, each as name:type/rank/element kind (? where none was recorded), and the type of what the fixture's run returned, exactly as the harness recorded them; the arguments and the returned type were recorded by two separate runs of the same fixture. No output value is shown, and a returned type says nothing about whether the output was right. “Level recorded” gives the UTC date when Syntology's graph wrote the sample's verification level, not when the code executed; where the level was written before Syntology's graph kept that date, the row says “level recorded, date not recorded”. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself. “Our draft was wrong” and “the contract check could not run” are failures of Syntology's instrument, not of the code: the sample ran on Syntology's synthesized input, and then the drafted contract did not describe it, or the check itself raised an error (RAISES), lacked a dependency (DEP_MISSING) or ran out of time (TIMEOUT); hover the label for which. A metamorphic tier, where shown, is evidence of the output's shape, not that the sample implements the method.

Each sample ends with its code_sha256. A sample's code_sha256 is Syntology's identity for that exact code. It is printed in full (16 hexadecimal characters for every sample in this build), and get_code(code_sha256="…") takes it exactly as printed. An agent calls it on Syntology's MCP service (how to connect); click an identity to copy that call.

Repository labels, per sample. official repository: The archive marks this repository official for the paper. named in the paper: The archive records that the paper mentions this repository; it is not marked official. community (archive-listed): In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper. found in paper text by Syntology: Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted. community: Not in the archive's code links for this paper; a community repository Syntology harvested. Samples are grouped by repository: a repository marked official first, then by how directly the repository is tied to the paper, then in the archive's order; never by how many ran. Licence labels name the repository's licence as recorded at harvest. “Pointer only” means Syntology does not serve that copy's text, for one of six reasons: no licence file was found; the licence was not identified; the licence is recorded as permissive but that copy's record is not marked cleared; the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar); this paper's copy was not located, so its licence is not recorded; or this paper's copy is cleared, but the stored copy of the same code that Syntology serves is not. Some licences outside that list permit redistribution, such as WTFPL, and GPL-3.0 under its conditions; they are simply not on the list. Hover a licence label for the reason. File links open the file on GitHub at the default branch, which may have changed since the harvest.

A function or class name links to its page when the name occurs in the harvested code of at least 5 papers; the name is a string, not an identity. A census tag appears when Syntology's census ran implementations with that name on one shared input per bucket: it gives how many were compared, from how many papers, and how many distinct outputs they produced, and names this sample's bucket when this sample was one of them.

Arthurizijar/Bert_Airport community (archive-listed) 10 samples, 7 ran (3 instrument-failed) · 3 unverified
convert_to_unicode in 40 papers tokenization.py ran · our draft was wrong fingerprinted Apache-2.0 (permissive) here; the stored copy is pointer only · 1923fc05163d207d · report fixture: signature text:str/0/str · returned str · level recorded 2026-09-22
create_float_feature in 11 papers create_pretraining_data.py ran Apache-2.0 (permissive) · b24b3b9c700da839 · report fixture: signature values:list/1/float · returned Feature · level recorded 2026-09-25
create_int_feature in 14 papers create_pretraining_data.py ran Apache-2.0 (permissive) · dd3ef48a15aec0cd · report fixture: signature values:list/1/int · returned Feature · level recorded 2026-09-25
gelu in 308 papers
census: 43 implementations from 74 papers, 6 distinct outputs on one shared input
modeling.py ran · honoured contract fingerprinted Apache-2.0 (permissive) · d08f324f950148de · report fixture: signature x:EagerTensor/2/float · returned EagerTensor · level recorded 2026-09-22
get_activation in 218 papers modeling.py ran · our draft was wrong Apache-2.0 (permissive) · 5842ee67106321d6 · report fixture: signature activation_string:str/0/str · returned function · level recorded 2026-09-22
input_fn_builder in 13 papers extract_features.py ran Apache-2.0 (permissive) · 47a3c792cebb6603 · report fixture: signature features:list/1/Feature, seq_length:int/0/int · returned function · level recorded 2026-09-25
printable_text in 36 papers tokenization.py ran · our draft was wrong Apache-2.0 (permissive) · 0e5615f8994003cf · report fixture: signature text:str/0/str · returned str · level recorded 2026-09-26
file_based_input_fn_builder in 16 papers run_classifier.py unverified Apache-2.0 (permissive) · 1b46c9ce24fc6db3 · report
get_assignment_map_from_checkpoint in 32 papers modeling.py unverified Apache-2.0 (permissive) here; the stored copy is pointer only · 50958618b65e514e · report
load_vocab in 107 papers tokenization.py unverified Apache-2.0 (permissive) · ff83ccc8b0b6462d · report
Chonwai/Learning_BERT community (archive-listed) 3 samples, 0 ran · 3 unverified
download_and_load_datasets predicting_movie_reviews_with_bert_on_tf_hub.py unverified Apache-2.0 (permissive) · 85813dd1e897d90f · report
load_dataset in 276 papers predicting_movie_reviews_with_bert_on_tf_hub.py unverified Apache-2.0 (permissive) · a181d94c3334306d · report
load_directory_data predicting_movie_reviews_with_bert_on_tf_hub.py unverified Apache-2.0 (permissive) · 0ccbf8e553bab00d · report
MOHAMEDELDAKDOUKY/bert_adjusted community (archive-listed) 1 sample, 0 ran · 1 unverified
create_optimizer in 37 papers optimization.py unverified Apache-2.0 (permissive) · c070006ff55cd83a · report
Maz101/Bert community (archive-listed) 2 samples, 1 ran (0 instrument-failed) · 1 unverified
load_vocab in 107 papers tokenization.py ran Apache-2.0 (permissive) · 30b05f2a415dd132 · report fixture: signature vocab_file:str/0/str · returned OrderedDict · level recorded 2026-09-25
file_based_input_fn_builder in 16 papers run_classifier.py unverified Apache-2.0 (permissive) · 28b353970e44ff54 · report
SpikeKing/My-Bert community (archive-listed) 1 sample, 1 ran (0 instrument-failed) · 0 unverified
create_optimizer in 37 papers optimization.py ran Apache-2.0 (permissive) · 021cbaa59ced6dac · report fixture: signature loss:SymbolicTensor/0/float, init_lr:float/0/float, num_train_steps:int/0/int, num_warmup_steps:int/0/int, use_tpu:bool/0/bool · returned Operation · level recorded 2026-09-24
StoneGH/bert community (archive-listed) 2 samples, 2 ran (0 instrument-failed) · 0 unverified
init_predict_var freeze_graph.py ran fingerprinted Apache-2.0 (permissive) · 5edf8fc5e1137648 · report fixture: signature path:str/0/str · returned tuple · level recorded 2026-09-24
set_logger in 26 papers freeze_graph.py ran Apache-2.0 (permissive) · 9b7c0abd85734790 · report fixture: signature context:str/0/str, verbose:bool/0/bool · returned Logger · level recorded 2026-09-24
TonyX19/bert_hyperpartisan community (archive-listed) 2 samples, 2 ran (0 instrument-failed) · 0 unverified
extractLabelXml seperate.py ran Apache-2.0 (permissive) · 06627935dcb8a0fe · report fixture: signature xml_file:str/0/str · returned list · level recorded 2026-09-24
loadText seperate.py ran Apache-2.0 (permissive) · ce566df49585f2b5 · report fixture: signature text_dir:str/0/str · returned dict · level recorded 2026-09-24
coronazap/bert_client community (archive-listed) 4 samples, 2 ran (0 instrument-failed) · 2 unverified
load_vocab in 107 papers tokenization.py ran Apache-2.0 (permissive) · 21c6f1a7ac31a2bd · report fixture: signature vocab_file:str/0/str · returned OrderedDict · level recorded 2026-09-24
process_result run_squad.py ran Apache-2.0 (permissive) · b57e2fc59104beb6 · report fixture: signature result:dict/0/? · returned RawResult · level recorded 2026-09-24
process_output in 8 papers run_squad.py unverified Apache-2.0 (permissive) · 7b79d58035cd12c2 · report
read_squad_examples run_squad.py unverified Apache-2.0 (permissive) · 3cbc0c492348e83d · report
halo090770/bert community (archive-listed) 1 sample, 1 ran (0 instrument-failed) · 0 unverified
create_optimizer in 37 papers optimization_1.py ran Apache-2.0 (permissive) · 5f1c73e253f08c93 · report fixture: signature loss:SymbolicTensor/0/float, init_lr:float/0/float, num_train_steps:int/0/int, num_warmup_steps:int/0/int, use_tpu:bool/0/bool · returned Operation · level recorded 2026-09-24
paolanu/BERT_epitope community (archive-listed) 1 sample, 1 ran (0 instrument-failed) · 0 unverified
convert_example_to_feature convert_examples_to_features.py ran Apache-2.0 (permissive) · b4a25fd2a133d044 · report fixture: signature example_row:tuple/1/SimpleExample · returned InputFeatures · level recorded 2026-09-24
somiltg/bert community (archive-listed) 3 samples, 3 ran (3 instrument-failed) · 0 unverified
get_tokens in 40 papers evaluate-v2.0.py ran · our draft was wrong fingerprinted Apache-2.0 (permissive) here; the stored copy is pointer only · aa6bc32d9c17a2f1 · report fixture: signature s:str/0/str · returned list · level recorded 2026-09-22
make_qid_to_has_ans in 14 papers evaluate-v2.0.py ran · our draft was wrong Apache-2.0 (permissive) · e91c44f5680827e2 · report fixture: signature dataset:list/1/dict · returned dict · level recorded 2026-09-22
normalize_answer in 304 papers evaluate-v2.0.py ran · our draft was wrong fingerprinted Apache-2.0 (permissive) here; the stored copy is pointer only · 6a96435eba311b08 · report fixture: signature s:str/0/str · returned str · level recorded 2026-09-22
the-Quert/iNLPfun community (archive-listed) 2 samples, 1 ran (0 instrument-failed) · 1 unverified
get_tasks in 21 papers code/download_glue_data.py ran Apache-2.0 (permissive) · c54b935b72798660 · report fixture: signature task_names:str/0/str · returned list · level recorded 2026-09-24
clones in 92 papers custom/encoder_ref.py unverified Apache-2.0 (permissive) · 7ba03346caea013d · report

Tasks

Knowledge DistillationLanguage ModellingModel CompressionSentiment Analysis

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Methods

AdamAttentionAttention DropoutBERTDense ConnectionsDropoutLayer NormalizationLinear LayerLinear Warmup With Linear DecayMulti-Head AttentionResidual ConnectionSoftmaxWeight DecayWordPiece

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