SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2310.18356 · 2023

LoRAShear: Efficient Large Language Model Structured Pruning and Knowledge Recovery

arXiv · PDF · Open in the Atlas

Code that ran

1 of 3 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; 0 honoured the contract we drafted, 0 violated it and 1 has no contract verdict; 0 failures of our instrument; 2 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; 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/2310.18356.json · badge: /api/ran/2310.18356.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
microsoft/only_train_once community (classifier label)
classifier basis not recorded on the link
1 ran · 2 no recorded run
FunctionStatusWhere it lives
get_non_stem_nodes Ran
ran on a synthesized input
microsoft/only_train_once/only_train_once/dependency_graph/pruning_dependency.py
code served (permissive licence) · get_code("3548130dc97c33b9")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

# Create a minimal Node class that satisfies the function's requirements
class Node:
    def __init__(self, node_id, stem=False, concat_axis_1=False, dummy=False):
        self.id = node_id
        self._stem = stem
        self._concat_axis_1 = concat_axis_1
        self._dummy = dummy
    
    def is_stem(self):
        return self._stem
    
    def is_concat(self, axis=None):
        if axis == 1:
            return self._concat_axis_1
        return False
    
    def is_dummy(self):
        return self._dummy

# Create a minimal Graph class that has a nodes attribute
class Graph:
    def __init__(self, nodes_dict):
        self.nodes = nodes_dict

# Construct test data
# Create various types of nodes:
# - stem nodes (should be skipped)
# - concat nodes with axis=1 (should be skipped)
# - dummy nodes (should be skipped)
# - non-stem, non-concat, non-dummy nodes (should be included)

nodes_dict = {
    1: Node(node_id=1, stem=True, concat_axis_1=False, dummy=False),      # stem - skip
    2: Node(node_id=2, stem=False, concat_axis_1=True, dummy=False),      # concat axis=1 - skip
    3: Node(node_id=3, stem=False, concat_axis_1=False, dummy=True),      # dummy - skip
    4: Node(node_id=4, stem=False, concat_axis_1=False, dummy=False),     # include
    5: Node(node_id=5, stem=False, concat_axis_1=False, dummy=False),     # include
    6: Node(node_id=6, stem=True, concat_axis_1=False, dummy=False),      # stem - skip
}

graph = Graph(nodes_dict)

# Create skip_node_ids set with node 5 to test skipping
skip_node_ids = {5}

ARGS = {
    "graph": graph,
    "skip_node_ids": skip_node_ids,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
automated_pruning_compression Not yet run
unverified (harvested, no recorded run)
microsoft/only_train_once/only_train_once/subnet_construction/pruning_compression.py
code served (permissive licence) · get_code("9c26845ee8b5aee3")
How it ran

no invocation recorded.

grow_non_stem_node_groups Not yet run
unverified (harvested, no recorded run)
microsoft/only_train_once/only_train_once/dependency_graph/pruning_dependency.py
code served (permissive licence) · get_code("c1a932acd507ff18")
How it ran

no invocation recorded.

Also present in other papers' code

Of 3 samples lifted from this paper's linked repositories, 3 carry a function identity and 0 do not. Of the 3 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 3 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

Large Language Models (LLMs) have transformed the landscape of artificial intelligence, while their enormous size presents significant challenges in terms of computational costs. We introduce LoRAShear, a novel efficient approach to structurally prune LLMs and recover knowledge. Given general LLMs, LoRAShear at first creates the dependency graphs over LoRA modules to discover minimally removal structures and analyze the knowledge distribution. It then proceeds progressive structured pruning on LoRA adaptors and enables inherent knowledge transfer to better preserve the information in the redundant structures. To recover the lost knowledge during pruning, LoRAShear meticulously studies and proposes a dynamic fine-tuning schemes with dynamic data adaptors to effectively narrow down the performance gap to the full models. Numerical results demonstrate that by only using one GPU within a couple of GPU days, LoRAShear effectively reduced footprint of LLMs by 20% with only 1.0% performance degradation and significantly outperforms state-of-the-arts. The source code will be available at https://github.com/microsoft/lorashear.

For agents

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

get_harvested_code_for_paper("2310.18356")
get_code_for_paper("2310.18356")
have("2310.18356")

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

curl https://syntology.ai/api/ran/2310.18356.json

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

[![Syntology run record](https://syntology.ai/api/ran/2310.18356.svg)](https://syntology.ai/paper/2310.18356)

Connect an agent — have() is free.