Papers › Parameter-Efficient Sparsity Crafting from Dense to Mixture-of-Experts for Instruction...

Parameter-Efficient Sparsity Crafting from Dense to Mixture-of-Experts for Instruction Tuning on General Tasks

5 Jan 2024arXiv:2401.02731archive 2025-07-28

Haoyuan Wu, Haisheng Zheng, Zhuolun He, Bei Yu

Large language models (LLMs) have demonstrated considerable proficiency in general natural language processing (NLP) tasks. Instruction tuning, a successful paradigm, enhances the ability of LLMs to follow natural language instructions and exhibit robust generalization across general tasks. However, these models often encounter performance limitations across multiple tasks due to constrained model capacity. Expanding this capacity during the instruction tuning phase poses significant challenges. To address this issue, we introduce parameter-efficient sparsity crafting (PESC), which crafts dense models into sparse models using the mixture-of-experts (MoE) architecture. PESC integrates adapters into the MoE layers of sparse models, differentiating experts without altering the individual weights within these layers. This method significantly reduces computational costs and GPU memory requirements, facilitating model capacity expansion through a minimal parameter increase when guaranteeing the quality of approximation in function space compared to original sparse upcycling. Our empirical evaluation demonstrates the effectiveness of the PESC method. Using PESC during instruction tuning, our best sparse model outperforms other sparse and dense models and exhibits superior general capabilities compared to GPT-3.5. Our code is available at https://github.com/wuhy68/Parameter-Efficient-MoE.

PaperPDFCode

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

Code

wuhy68/parameter-efficient-moe officialmentioned in papermentioned on GitHubpytorch report
ShayekhBinIslam/openrag 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

Arithmetic ReasoningCode GenerationCommon Sense ReasoningMath Word Problem SolvingMixture-of-ExpertsMulti-task Language UnderstandingQuestion AnsweringSentence Completion

1 archive task tag without a task page not shown.

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Arithmetic Reasoning GSM8K Camelidae-8×34B (5-shot) Accuracy 78.3 #78 of 164 Archive leaderboard report
Arithmetic Reasoning GSM8K Qwen2idae-16x14B (5-shot) Accuracy 77.8 #79 of 164 Archive leaderboard report
Code Generation MBPP Qwen2idae-16x14B (4-shot) Accuracy 48.6 #65 of 99 Archive leaderboard report
Code Generation MBPP Camelidae-8×34B (4-shot) Accuracy 41.4 #82 of 99 Archive leaderboard report
Common Sense Reasoning ARC (Challenge) Camelidae-8×34B Accuracy 65.2 #21 of 54 Archive leaderboard report
Common Sense Reasoning ARC (Easy) Camelidae-8×34B Accuracy 86.2 #6 of 47 Archive leaderboard report
Common Sense Reasoning WinoGrande Camelidae-8×34B Accuracy 80.9 #16 of 77 Archive leaderboard report
Math Word Problem Solving MATH Qwen2idae-16x14B (4-shot) Accuracy 29.9 #90 of 135 Archive leaderboard report
Math Word Problem Solving MATH Camelidae-8×34B (4-shot) Accuracy 22.6 #100 of 135 Archive leaderboard report
Question Answering PIQA Camelidae-8×34B Accuracy 82.7 #19 of 67 Archive leaderboard report
Sentence Completion HellaSwag Camelidae-8×34B (10-shot) Accuracy 83.2 #34 of 89 Archive leaderboard report
Sentence Completion HellaSwag Qwen2idae-16x14B (10-shot) Accuracy 82.3 #38 of 89 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

AdamAttentionAttention DropoutBPECosine AnnealingDense ConnectionsDropoutGPT-3Layer NormalizationLinear LayerLinear Warmup With Cosine AnnealingMulti-Head AttentionResidual ConnectionSoftmaxWeight Decay

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