SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · aaai_29784 · AAAI · 2024

Separate the Wheat from the Chaff: Model Deficiency Unlearning via Parameter-Efficient Module Operation

Min Zhang, Dongfang Li, Baotian Hu, Zihao Zheng, Zhenyu Liu, Xinshuo Hu

Open in the Atlas

Code that ran

3 of 13 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; 2 failures of our instrument; 10 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/aaai_29784.json · badge: /api/ran/aaai_29784.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
HITsz-TMG/Ext-Sub canonical (classifier label)
classifier basis: census
3 ran · 10 no recorded run
FunctionStatusWhere it lives
calculate_cross_entropy Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
HITsz-TMG/Ext-Sub/eval/halueval_eval.py
code served (permissive licence) · get_code("8e0bf627e1b05197")
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
from torch.nn import CrossEntropyLoss

torch.manual_seed(0)

# Create lm_logits: shape (batch_size=2, seq_len=5, vocab_size=4)
# After shift operation it becomes (2, 4, 4)
lm_logits = torch.randn(2, 5, 4)

# Create labels: shape (batch_size=2, seq_len=5)
# After shift operation it becomes (2, 4)
labels = torch.randint(0, 4, (2, 5), dtype=torch.long)

ARGS = {
    "lm_logits": lm_logits,
    "labels": labels,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
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
jload Ran
ran; our contract draft was wrong, not the code
HITsz-TMG/Ext-Sub/training/utils.py
code served (permissive licence) · get_code("d07d04439cd1d44f")
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) and returned dict. The image it ran in was not recorded. If your copy returns the same type 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 io
import json
import tempfile
import os

# Create a temporary JSON file with test data
temp_dir = tempfile.mkdtemp()
temp_file = os.path.join(temp_dir, "test.json")

test_data = {
    "key1": "value1",
    "key2": [1, 2, 3],
    "key3": {"nested": "dict"}
}

with open(temp_file, 'w') as f:
    json.dump(test_data, f)

# Define ARGS with the file path
ARGS = {
    "f": temp_file,
    "mode": "r"
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
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
ngram_metrics Ran
ran on a synthesized input
HITsz-TMG/Ext-Sub/eval/ngram_rep_eval.py
code served (permissive licence) · get_code("31b68e2e8ba3e815")
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 defaultdict. The image it ran in was not recorded. If your copy returns the same type 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
from collections import defaultdict, Counter
from nltk import ngrams

torch.manual_seed(0)

# The function takes:
# - token_list: List[int] - a list of token IDs
# - pad: int - a padding token ID (default -100)

# The function processes token_list by:
# 1. Checking if pad value is in token_list
# 2. If so, truncating at the first occurrence of pad
# 3. Computing n-grams for n in range(1, 5)
# 4. Computing repetition statistics for 1-grams through 4-grams

# Create a valid token_list with some repeated tokens
# Mix of unique and repeated tokens to make metrics non-trivial
token_list = [10, 20, 10, 30, 20, 40, 10, 50, -100, 99, 99]

ARGS = {
    'token_list': token_list,
    'pad': -100
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
nltk, numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type defaultdict — a defaultdict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a defaultdict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
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
batch_llm_generate Not yet run
unverified (harvested, no recorded run)
HITsz-TMG/Ext-Sub/eval/toxic_eval_generation.py
code served (permissive licence) · get_code("e58ceb7971f091eb")
How it ran

no invocation recorded.

calculate_loss_list Not yet run
unverified (harvested, no recorded run)
HITsz-TMG/Ext-Sub/eval/halueval_eval.py
code served (permissive licence) · get_code("3ba643874feca8b4")
How it ran

no invocation recorded.

get_peft_config Not yet run
unverified (harvested, no recorded run)
HITsz-TMG/Ext-Sub/training/train_peft.py
code served (permissive licence) · get_code("cc48407bea0a7d20")
How it ran

no invocation recorded.

load_model Not yet run
unverified (harvested, no recorded run)
HITsz-TMG/Ext-Sub/eval/toxic_eval_generation.py
code served (permissive licence) · get_code("8a1e619bcc580a63")
How it ran

no invocation recorded.

load_tokenizer Not yet run
unverified (harvested, no recorded run)
HITsz-TMG/Ext-Sub/eval/ngram_rep_eval.py
code served (permissive licence) · get_code("e65a8e07bf87b809")
How it ran

no invocation recorded.

merge_lora_weight Not yet run
unverified (harvested, no recorded run)
HITsz-TMG/Ext-Sub/ext_sub.py
code served (permissive licence) · get_code("725b6eefc92ed3b6")
How it ran

no invocation recorded.

predict Not yet run
unverified (harvested, no recorded run)
HITsz-TMG/Ext-Sub/eval/toxic_eval_generation.py
code served (permissive licence) · get_code("3a911824709b51f8")
How it ran

no invocation recorded.

preprocess Not yet run
unverified (harvested, no recorded run)
HITsz-TMG/Ext-Sub/eval/halueval_eval.py
code served (permissive licence) · get_code("fcb5a8afcdcfee99")
How it ran

no invocation recorded.

preprocess Not yet run
unverified (harvested, no recorded run)
HITsz-TMG/Ext-Sub/training/train_peft.py
code served (permissive licence) · get_code("39422400c498cfe4")
How it ran

no invocation recorded.

rep_ngram_eval Not yet run
unverified (harvested, no recorded run)
HITsz-TMG/Ext-Sub/eval/ngram_rep_eval.py
code served (permissive licence) · get_code("5eec89a5b721401b")
How it ran

no invocation recorded.

Also present in other papers' code

Of 13 samples lifted from this paper's linked repositories, 13 carry a function identity and 0 do not. Of the 13 distinct functions with an identity, 3 are also present in other papers whose code we hold (3 with at least one paper dated earlier, 0 the same month, 1 later) and 10 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.

FunctionAlso present inEarliest-datedOthers named
jload 46 other papers
18 dated earlier, 5 later, 23 same year, order not determinable
2302.13971 (2023-02)
LLaMA: Open and Efficient Foundation Language Models
2305.12870 (2023-05)
Lion: Adversarial Distillation of Proprietary Large Language Models
2305.13068 (2023-05)
Making Language Models Better Tool Learners with Execution Feedback
and 43 more not named here
preprocess 23 other papers
8 dated earlier, 15 same year, order not determinable
2305.12870 (2023-05)
Lion: Adversarial Distillation of Proprietary Large Language Models
2306.17194 (2023-06)
On the Exploitability of Instruction Tuning
2308.07286 (2023-08)
The Devil is in the Errors: Leveraging Large Language Models for Fine-grained Machine Translation Evaluation
and 20 more not named here
calculate_cross_entropy 1 other paper
1 dated earlier
2308.08090 (2023-08)
Separate the Wheat from the Chaff: Model Deficiency Unlearning via Parameter-Efficient Module Operation

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 been widely used in various applications but are known to suffer from issues related to untruthfulness and toxicity. While parameter-efficient modules (PEMs) have demonstrated their effectiveness in equipping models with new skills, leveraging PEMs for deficiency unlearning remains underexplored. In this work, we propose a PEMs operation approach, namely Extractionbefore-Subtraction (Ext-Sub), to enhance the truthfulness and detoxification of LLMs through the integration of "expert" PEM and "anti-expert" PEM. Remarkably, even antiexpert PEM possess valuable capabilities due to their proficiency in generating fabricated content, which necessitates language modeling and logical narrative competence. Rather than merely negating the parameters, our approach involves extracting and eliminating solely the deficiency capability within anti-expert PEM while preserving the general capabilities. To evaluate the effectiveness of our approach in terms of truthfulness and detoxification, we conduct extensive experiments on LLMs, encompassing additional abilities such as language modelling and mathematical reasoning. Our empirical results demonstrate that our approach effectively improves truthfulness and detoxification, while largely preserving the fundamental abilities of LLMs.

For agents

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

get_harvested_code_for_paper("aaai_29784")
get_code_for_paper("aaai_29784")
have("aaai_29784")

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

curl https://syntology.ai/api/ran/aaai_29784.json

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

[![Syntology run record](https://syntology.ai/api/ran/aaai_29784.svg)](https://syntology.ai/paper/aaai_29784)

Connect an agent — have() is free.