Papers › Lifting the Curse of Capacity Gap in Distilling Language Models

Lifting the Curse of Capacity Gap in Distilling Language Models

20 May 2023arXiv:2305.12129archive 2025-07-28

Chen Zhang, Yang Yang, Jiahao Liu, Jingang Wang, Yunsen Xian, Benyou Wang, Dawei Song

Pretrained language models (LMs) have shown compelling performance on various downstream tasks, but unfortunately they require a tremendous amount of inference compute. Knowledge distillation finds a path to compress LMs to small ones with a teacher-student paradigm. However, when the capacity gap between the teacher and the student is large, a curse of capacity gap appears, invoking a deficiency in distilling LMs. While a few studies have been carried out to fill the gap, the curse is not yet well tackled. In this paper, we aim at lifting the curse of capacity gap via enlarging the capacity of the student without notably increasing the inference compute. Largely motivated by sparse activation regime of mixture of experts (MoE), we propose a mixture of minimal experts (MiniMoE), which imposes extra parameters to the student but introduces almost no additional inference compute. Experimental results on GLUE and CoNLL demonstrate the curse of capacity gap is lifted by the magic of MiniMoE to a large extent. MiniMoE also achieves the state-of-the-art performance at small FLOPs compared with a range of competitive baselines. With a compression rate as much as ∼50×, MiniMoE preserves ∼95% GLUE score of the teacher.

PaperPDFCodeCode Syntology ranMethod as code

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

Syntology 13 code samples harvested from 1 repository linked to this paper: 7 ran (of which 0 constructed an object rather than computing a result; 6 with no instrument failure: 0 honoured, 0 violated, 6 with no contract checked; 1 where Syntology's instrument failed) · 6 unverified. Where Syntology's instrument failed: 1 where the contract check could not run; failures of Syntology's instrument, not of the code.

By repository: official repository: 13 samples from 1 repository, 7 ran (6 with no instrument failure; 1 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.

Levels written between 2026-09-22 and 2026-09-24 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

genezc/minimoe officialmentioned in papermentioned on GitHubpytorchSyntology: 13 samples, 7 ran (1 instrument-failed) · 6 unverified 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

13 samples harvested: 7 ran (of which 0 constructed an object rather than computing a result; 6 with no instrument failure: 0 honoured, 0 violated, 6 with no contract checked; 1 where Syntology's instrument failed) · 6 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page has not changed since , the first build that kept a record date for it; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

1ran; the contract check could not run
6ran
6unverified

Licence: 0 of the 13 samples are pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from genezc/minimoe. “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.

genezc/minimoe official repository 13 samples, 7 ran (1 instrument-failed) · 6 unverified
GateLinear modules/modeling_moebert_gate.py ran Apache-2.0 (permissive) · 6abd1347d0c44be3 · report fixture: signature input:Tensor/2/float · returned Tensor · level recorded 2026-09-24
MoEBertAttention modules/modeling_moebert_gate.py ran Apache-2.0 (permissive) · 0906fd483dd042ca · report fixture: signature hidden_states:Tensor/3/float, attention_mask:Tensor/4/float, head_mask:NoneType/0/?, output_attentions:bool/0/bool · returned NoneType · level recorded 2026-09-24
MoEBertEmbeddings modules/modeling_moebert_gate.py ran · metamorphic tier: invariant Apache-2.0 (permissive) · d059c2309fca697b · report fixture: signature input_ids:Tensor/2/int, token_type_ids:NoneType/0/?, position_ids:NoneType/0/?, inputs_embeds:NoneType/0/? · returned Tensor · level recorded 2026-09-22
MoEBertIntermediate modules/modeling_moebert_gate.py ran fingerprinted Apache-2.0 (permissive) · 16f3cea91fa35e41 · report fixture: signature hidden_states:Tensor/3/float, expert_idx:NoneType/0/? · returned Tensor · level recorded 2026-09-24
MoEBertOutput modules/modeling_moebert_gate.py ran Apache-2.0 (permissive) · fbfad05bb8c37ef0 · report fixture: signature hidden_states:Tensor/3/float, input_tensor:Tensor/3/float, expert_idx:NoneType/0/?, expert_gate:NoneType/0/? · returned Tensor · level recorded 2026-09-24
MoEBertPooler modules/modeling_moebert_gate.py ran · metamorphic tier: invariant fingerprinted Apache-2.0 (permissive) · 2af6f69a908eee12 · report fixture: signature hidden_states:Tensor/3/float · returned Tensor · level recorded 2026-09-22
one_hot in 150 papers
census: 18 implementations from 25 papers, 10 distinct outputs across 6 buckets, one shared input per bucket this sample: bucket 2 of 6, arg 1: rank 2, kind int, dtype int64
modules/modeling_moebert_gate.py ran; the contract check could not run (RAISES) Apache-2.0 (permissive) · 882d0552fedc5aa1 · report fixture: signature indices:Tensor/2/int, num_classes:int/0/int · returned Tensor · level recorded 2026-09-22
MoEBertEncoder modules/modeling_moebert_gate.py unverified Apache-2.0 (permissive) · 215b29e96dbea836 · report
MoEBertLayer modules/modeling_moebert_gate.py unverified Apache-2.0 (permissive) · ea3533cf6785ab2e · report
MoEBertModel modules/modeling_moebert_gate.py unverified Apache-2.0 (permissive) · 29c6962b86d5b64c · report
MoEBertSelfAttention modules/modeling_moebert_gate.py unverified Apache-2.0 (permissive) · a401370bc11c23ef · report
MoEBertSelfOutput modules/modeling_moebert_gate.py unverified Apache-2.0 (permissive) · 09da8fe5de680832 · report
MoELinear modules/modeling_moebert_gate.py unverified Apache-2.0 (permissive) · abc80e95eaef3d78 · report

Method as code Syntology

What Syntology's graph records as this paper's claimed method, lifted into code: each assembly names its bricks (functions and classes), each with its own role and its own verification level. An assembly is not a check that the code implements the method correctly or reproduces the paper's results. Levels are each brick's own verification level in Syntology's graph (level 0: no recorded run), listed brick by brick and never combined into one level; “ran” means the brick executed on a synthesized input, not that its output is correct, and the instrument figure counts failures of Syntology's instrument, not of the code. Where a model read the code against the paper's claim, each assembly's label verdict is one model's judgement, kept in this page's JSON twin; it is not calibrated, so no card prints it.

1 assembly on this paper: 1 harvested, 0 generated.

harvested Assembly for MINIMOE (graph method: MINIMOE). 13 bricks; individually on a fixture: 7 ran (of which 0 constructed an object rather than computing a result; 6 with no instrument failure: 0 honoured, 0 violated, 6 with no contract checked; 1 where Syntology's instrument failed) · 6 unverified. The assembly has not been run end-to-end.

  1. MoEBertModel entry · level 0 · unverified · 29c6962b86d5b64c
  2. MoEBertEmbeddings model · level 1 · ran · d059c2309fca697b
  3. MoEBertEncoder model · level 0 · unverified · 215b29e96dbea836
  4. MoEBertPooler model · level 1 · ran · 2af6f69a908eee12
  5. MoEBertLayer model · level 0 · unverified · ea3533cf6785ab2e
  6. GateLinear model · level 1 · ran · 6abd1347d0c44be3
  7. MoEBertAttention model · level 1 · ran · 0906fd483dd042ca
  8. MoEBertIntermediate model · level 1 · ran · 16f3cea91fa35e41
  9. MoEBertOutput model · level 1 · ran · fbfad05bb8c37ef0
  10. one_hot helper · level 1 · ran; the contract check could not run (RAISES) · 882d0552fedc5aa1
  11. MoEBertSelfAttention helper · level 0 · unverified · a401370bc11c23ef
  12. MoEBertSelfOutput helper · level 0 · unverified · 09da8fe5de680832
  13. MoELinear model · level 0 · unverified · abc80e95eaef3d78

Each brick ends with its code_sha256, Syntology's identity for that exact code; click one to copy the get_code call that returns it. For agents, on Syntology's MCP service: assembly(arxiv_id="2305.12129") returns these assemblies with each brick's role and wiring (how to connect).

Tasks

Knowledge DistillationMixture-of-Experts

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Methods

Knowledge Distillation

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