SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2609.00575 · September 2026

Residual Sparsification via Output Importance for Compressing Mixture-of-Experts LLMs

Junseok Lee, Chuck Yoo, Gyeongsik Yang, Seungmin Cha, Seungwoo Jung, Dohyeok Kwon, Yeonho Yoo

arXiv · PDF · Open in the Atlas

Code that ran

14 of 16 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 1 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 13 failures of our instrument; 2 no recorded run), measured 2026-11-22; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2609.00575.json · badge: /api/ran/2609.00575.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
OSSS-KU/PARSER canonical (classifier label)
classifier basis: census
14 ran · 2 no recorded run
FunctionStatusWhere it lives
combine_gate_up Ran
ran; our contract draft was wrong, not the code
OSSS-KU/PARSER/moe_pipeline/methods/parser.py
pointer only (licence: NONE) · get_code("72dd374fe5554019")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, OSSS-KU/PARSER when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

describe_model Ran
ran; our contract draft was wrong, not the code
OSSS-KU/PARSER/moe_pipeline/model_adapter.py
pointer only (licence: NONE) · get_code("dea0f8d0cdf4dd11")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, OSSS-KU/PARSER when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

ensure_device Ran
ran; our contract draft was wrong, not the code
OSSS-KU/PARSER/moe_pipeline/model_adapter.py
pointer only (licence: NONE) · get_code("c3f16caadeb018ba")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, OSSS-KU/PARSER when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

expert_proj_attr_map Ran
ran; our contract draft was wrong, not the code
OSSS-KU/PARSER/moe_pipeline/moe_introspection.py
pointer only (licence: NONE) · get_code("2f95370a0d5b4831")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, OSSS-KU/PARSER when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

get_experts Ran
ran; our contract draft was wrong, not the code
OSSS-KU/PARSER/moe_pipeline/moe_introspection.py
pointer only (licence: NONE) · get_code("bc1dbe535e45c676")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, OSSS-KU/PARSER when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

make_expert_key Ran
ran; our contract draft was wrong, not the code
OSSS-KU/PARSER/moe_pipeline/methods/parser.py
pointer only (licence: NONE) · get_code("558255df6b9f0304")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, OSSS-KU/PARSER when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

normalize_exclude_domains Ran
ran; our contract draft was wrong, not the code
OSSS-KU/PARSER/moe_pipeline/calibration/parser_activation.py
pointer only (licence: NONE) · get_code("3dc2d3aa615186ec")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, OSSS-KU/PARSER when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

parse_layer_spec Ran
ran, honoured the contract we drafted
OSSS-KU/PARSER/compress_blocks.py
pointer only (licence: NONE) · get_code("95f10ed2539d15cc")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, OSSS-KU/PARSER when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

permute_columns Ran
ran; our contract draft was wrong, not the code
OSSS-KU/PARSER/moe_pipeline/backbone_transport.py
pointer only (licence: NONE) · get_code("68b6edb39c63312a")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, OSSS-KU/PARSER when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

permute_rows Ran
ran; our contract draft was wrong, not the code
OSSS-KU/PARSER/moe_pipeline/backbone_transport.py
pointer only (licence: NONE) · get_code("21057ae2149ed972")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, OSSS-KU/PARSER when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

resolve_activation_fn Ran
ran; our contract draft was wrong, not the code
OSSS-KU/PARSER/moe_pipeline/methods/parser.py
pointer only (licence: NONE) · get_code("d5e8ebb743dd5b45")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, OSSS-KU/PARSER when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

resolve_device_map Ran
ran; our contract draft was wrong, not the code
OSSS-KU/PARSER/evaluate_blocks.py
pointer only (licence: NONE) · get_code("b19c6e02f95fdde6")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, OSSS-KU/PARSER when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

resolve_dtype Ran
ran; our contract draft was wrong, not the code
OSSS-KU/PARSER/evaluate_blocks.py
pointer only (licence: NONE) · get_code("1d146dee4b822985")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, OSSS-KU/PARSER when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

resolve_hf_torch_dtype Ran
ran; our contract draft was wrong, not the code
OSSS-KU/PARSER/evaluate_blocks.py
pointer only (licence: NONE) · get_code("871b82dc51e2fbbe")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, OSSS-KU/PARSER when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

build_wikitext_prompts Not yet run
unverified (harvested, no recorded run)
OSSS-KU/PARSER/benchmark_tps.py
pointer only (licence: NONE) · get_code("06341373ffd7a746")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, OSSS-KU/PARSER when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

is_supported_moe_block Not yet run
unverified (harvested, no recorded run)
OSSS-KU/PARSER/moe_pipeline/moe_introspection.py
pointer only (licence: NONE) · get_code("1bdf79198155d2e1")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, OSSS-KU/PARSER when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

Also present in other papers' code

Of 16 samples lifted from this paper's linked repositories, 16 carry a function identity and 0 do not. Of the 16 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 16 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Mixture-of-experts (MoE) architectures scale large language models efficiently, but they demand massive GPU memory. To cope with such demand, models are commonly compressed to reduce their memory footprint. Residual sparsification is a representative compression technique that decomposes each projection matrix of an expert into a shared base matrix and per-expert residual matrix, and then compresses the residuals. Existing sparsification methods compress each residual matrix independently by minimizing its compression error, thereby minimizing the error of each projection matrix. However, our analysis shows that this objective is misaligned with preserving model accuracy after compression. In an expert, the final output is produced through computations coupled across multiple projections and hidden representations. Therefore, even small errors in individual matrices can propagate through hidden representations and projection interactions, leading to large expert output errors and accuracy degradation. To address this misalignment, we propose PARSER, a new residual sparsification method that shifts the compression objective from minimizing isolated matrix errors to preserving the expert output error. PARSER achieves this by introducing output importance, which measures the actual contribution to the expert output error. Our experiments show that, compared with existing methods, PARSER narrows the accuracy gap to the uncompressed model by 1.41$\times$ on Qwen and 1.44$\times$ on DeepSeek, while achieving the same peak memory reduction. Our code is available at https://github.com/OSSS-KU/PARSER.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2609.00575")
get_code_for_paper("2609.00575")
have("2609.00575")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2609.00575.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2609.00575.svg)](https://syntology.ai/paper/2609.00575)

Connect an agent — have() is free.