Methods › Natural Language Processing › Transformers › Switch Transformer
Switch Transformer
archive 2025-07-28 Description, source and code snippet are the archive's method entry.
Switch Transformer is a sparsely-activated expert Transformer model that aims to simplify and improve over Mixture of Experts. Through distillation of sparse pre-trained and specialized fine-tuned models into small dense models, it reduces the model size by up to 99% while preserving 30% of the quality gains of the large sparse teacher. It also uses selective precision training that enables training with lower bfloat16 precision, as well as an initialization scheme that allows for scaling to a larger number of experts, and also increased regularization that improves sparse model fine-tuning and multi-task training.
In Syntology repository known, no verified implementation (declared_code; repositories: 5 listed of 8 in the graph, strongest link: official). The archive credits no paper with an arXiv id for this method, so whether this answer (about arXiv:2101.03961) is about this method is not checked.
Code Syntology holds under other names, as the lookup lists them (code_nearby); not a complete list, and none of them is this method.
- Agentic Transformer · shares_name_tokens (shares name tokens with this name; a different method) · level V3 · retrieve with
get_reference_implementation(name='Agentic Transformer') - Waypoint Transformer · shares_name_tokens (shares name tokens with this name; a different method) · level V2 · retrieve with
get_reference_implementation(name='Waypoint Transformer') - Block-State Transformer · shares_name_tokens (shares name tokens with this name; a different method) · level V3 · retrieve with
get_reference_implementation(name='Block-State Transformer') - Bootstrapped Transformer · shares_name_tokens (shares name tokens with this name; a different method) · level V3 · retrieve with
get_reference_implementation(name='Bootstrapped Transformer') - Focused Transformer (FOT) · shares_name_tokens (shares name tokens with this name; a different method) · level V3 · retrieve with
get_reference_implementation(name='Focused Transformer (FOT)') - Elastic Decision Transformer · shares_name_tokens (shares name tokens with this name; a different method) · level V2 · retrieve with
get_reference_implementation(name='Elastic Decision Transformer')
Origin paper of this answer, arXiv:2101.03961 (its page on this site · its page on Syntology): Syntology's run record for its harvested code, 7 ran (of which 2 constructed an object rather than computing a result; 3 with no instrument failure: 0 honoured, 0 violated, 3 with no contract checked; 4 where Syntology's instrument failed) · 6 unverified (of 13 harvested samples) · 8 pointer-only (licence). The instrument figure counts failures of Syntology's instrument, not of the code. Ran means executed on a synthesized input, not that the code is correct.
Syntology's have check for this name (free, it answers coverage only), read 2026-10-05; ask it again live. A level is the served verification level of that one answer; nothing here is summed across methods.
What each level means, quoted from Syntology's server card: V0 “imports”; V1 “executes against a synthesized fixture and returns a value of the recorded type”; V2 “two implementations GENERATED from the paper's signature and spec agree numerically (1e-6) on fixture inputs” “neither of the two ever read the authors' own code, so V2 measures how unambiguously the paper specifies the method, not fidelity to the authors' implementation (measured 2026-09-11, commit 75eaa77: cross_checked_by is a second LLM on 2,628 of 2,831 served samples)”; V3 “paper-derived property tests pass on both of those same two generations.” The card adds: “No rung on this ladder compares the served code against the authors' implementation, so treat every level as evidence about runnability and internal agreement, not about matching the paper's results.” A level quoted in two parts is one definition the card writes as two clauses joined by a double hyphen, each clause quoted on its own (how these levels are quoted).
Papers archive 2025-07-28
archive papers tagged: 20 · with a code link: 6 · where Syntology ran a sample: 4 (4 with a run with no instrument failure, 0 where every run was a failure of Syntology's instrument) Syntology
Show: all tagged papersonly where code ran (4 of 20 tagged: 4 with a run with no instrument failure, 0 where every run was a failure of Syntology's instrument)
Page 1 of 1: papers 1 to 20 of 20, newest first by the archive's date (ties by slug), in archive order.
Repository counts are the archive's code-links table. A Syntology line states what Syntology ran from that paper's harvested code, as “N ran (of which C constructed an object rather than computing a result; K with no instrument failure: H honoured, V violated, P with no contract checked; I where Syntology's instrument failed) · U unverified”; the instrument figure counts failures of Syntology's instrument, not of the code. It is per sample and not a correctness claim. When the archive marks a repository official for the paper, the line starts with that repository's state (the archive's flag, not a verdict on who wrote the code; “community repositories only” when every sample that ran came from a community repository, “official: no sample here; runs from other or unrecorded repositories” when some came from a repository the paper names or has in its text, or from none recorded); hover it for the repositories the samples that ran came from.
-
QoS-Efficient Serving of Multiple Mixture-of-Expert LLMs Using Partial Runtime Reconfiguration 10 May 2025 · 0 repositories · arXiv:2505.06481
-
ExpertRAG: Efficient RAG with Mixture of Experts -- Optimizing Context Retrieval for Adaptive LLM Responses 23 Mar 2025 · 0 repositories · arXiv:2504.08744
-
ResMoE: Space-efficient Compression of Mixture of Experts LLMs via Residual Restoration 10 Mar 2025 · 1 repository · arXiv:2503.06881Syntology official (archive's flag): 10 ran · 10 ran (of which 0 constructed an object rather than computing a result; 7 with no instrument failure: 0 honoured, 1 violated, 6 with no contract checked; 3 where Syntology's instrument failed) · 5 unverified (of 15 harvested samples) · 1 pointer-only (licence)
-
Sparse Backpropagation for MoE Training 1 Oct 2023 · 0 repositories · arXiv:2310.00811
-
SwapMoE: Serving Off-the-shelf MoE-based Large Language Models with Tunable Memory Budget 29 Aug 2023 · 0 repositories · arXiv:2308.15030
-
Condensing Multilingual Knowledge with Lightweight Language-Specific Modules 23 May 2023 · 1 repository · arXiv:2305.13993Syntology official (archive's flag): 2 ran · 2 ran (of which 2 constructed an object rather than computing a result; 1 with no instrument failure: 0 honoured, 0 violated, 1 with no contract checked; 1 where Syntology's instrument failed) · 0 unverified; every one of the 2 samples that ran constructed an object rather than computing a result (of 2 harvested samples)
-
Towards A Unified View of Sparse Feed-Forward Network in Pretraining Large Language Model 23 May 2023 · 0 repositories · arXiv:2305.13999
-
Improving Transformer Performance for French Clinical Notes Classification Using Mixture of Experts on a Limited Dataset 22 Mar 2023 · 0 repositories · arXiv:2303.12892
-
SMILE: Scaling Mixture-of-Experts with Efficient Bi-level Routing 10 Dec 2022 · 0 repositories · arXiv:2212.05191
-
Argumentative Text Generation in Economic Domain 18 Jun 2022 · 1 repository · arXiv:2206.09251
-
Automatic Summarization of Russian Texts: Comparison of Extractive and Abstractive Methods 18 Jun 2022 · 0 repositories · arXiv:2206.09253
-
Build a Robust QA System with Transformer-based Mixture of Experts 20 Mar 2022 · 1 repository · arXiv:2204.09598
-
Efficient Language Modeling with Sparse all-MLP 14 Mar 2022 · 0 repositories · arXiv:2203.06850
-
Switch Trajectory Transformer with Distributional Value Approximation for Multi-Task Reinforcement Learning 14 Mar 2022 · 0 repositories · arXiv:2203.07413
-
Mixture-of-Experts with Expert Choice Routing 18 Feb 2022 · 0 repositories · arXiv:2202.09368
-
M6-10T: A Sharing-Delinking Paradigm for Efficient Multi-Trillion Parameter Pretraining 8 Oct 2021 · 0 repositories · arXiv:2110.03888
-
Taming Sparsely Activated Transformer with Stochastic Experts 8 Oct 2021 · 1 repository · arXiv:2110.04260Syntology official (archive's flag): 2 ran · 2 ran (of which 0 constructed an object rather than computing a result; 2 with no instrument failure: 0 honoured, 0 violated, 2 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified (of 2 harvested samples)
-
Random Offset Block Embedding Array (ROBE) for CriteoTB Benchmark MLPerf DLRM Model : 1000× Compression and 3.1× Faster Inference 4 Aug 2021 · 0 repositories · arXiv:2108.02191
-
Carbon Emissions and Large Neural Network Training 21 Apr 2021 · 0 repositories · arXiv:2104.10350
-
Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity 11 Jan 2021 · 8 repositories · arXiv:2101.03961Syntology official: harvested, nothing ran · 7 ran (of which 2 constructed an object rather than computing a result; 3 with no instrument failure: 0 honoured, 0 violated, 3 with no contract checked; 4 where Syntology's instrument failed) · 6 unverified (of 13 harvested samples) · 8 pointer-only (licence)
Tasks archive 2025-07-28
20 shown of 30 tasks the archive attaches to papers tagged with this method, by distinct papers. A task without a page in the catalog is plain text.
Usage over time archive 2025-07-28
Components: the archive holds no method-to-method composition, so PwC's Components table cannot be rebuilt; the Papers list carries no Results column for the same reason (the archive does not join its leaderboard rows to method tags).
Categories archive 2025-07-28
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