SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2205.14219 · NeurIPS · 2022

Controllable Text Generation with Neurally-Decomposed Oracle

Kai-Wei Chang, Sidi Lu, Tao Meng, Nanyun Peng

arXiv · PDF · Open in the Atlas

Code that ran

2 of 2 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 0 have no contract verdict; 2 failures of our instrument; 0 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/2205.14219.json · badge: /api/ran/2205.14219.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
mtsomethree/constrdecoding canonical (classifier label)
classifier basis: classifier_2025
2 ran · 0 no recorded run
FunctionStatusWhere it lives
sample_from_marianMT Ran
ran; our contract draft was wrong, not the code
mtsomethree/constrdecoding/train_MT.py
code served (permissive licence) · get_code("f8b0ed7ea83e455f")
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 tuple. 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 torch.nn as nn
import tempfile
import os

torch.manual_seed(0)

# Create a minimal mock model
class MockMarianMTModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.config = type('Config', (), {
            'pad_token_id': 0,
        })()
        self.linear = nn.Linear(4, 4)
        self._constraint_factor = 0.0
        self._temperature = 1.0
    
    def set_constraint_factor(self, value):
        self._constraint_factor = value
    
    def set_temperature(self, value):
        self._temperature = value
    
    def to(self, device):
        return super().to(device)
    
    def generate(self, input_ids, attention_mask, do_sample=False, output_scores=False, 
                 return_dict_in_generate=False, max_length=10):
        batch_size = input_ids.shape[0]
        seq_length = 6
        
        # Create mock output sequences
        sequences = torch.randint(1, 100, (batch_size, seq_length))
        sequences[:, -1] = 2  # EOS token
        
        # Create mock scores
        sequences_scores = torch.randn(batch_size)
        
        # Return object with required attributes
        output = type('GenerateOutput', (), {
            'sequences': sequences,
            'sequences_scores': sequences_scores,
        })()
        
        return output

# Create a minimal mock tokenizer
class MockTokenizer:
    def __init__(self):
        self.vocab_size = 100
    
    def __call__(self, texts, return_tensors=None, padding=False, truncation=False, max_length=None):
        batch_size = len(texts)
        seq_length = 5
        
        input_ids = torch.randint(1, 50, (batch_size, seq_length))
        attention_mask = torch.ones(batch_size, seq_length)
        
        return {
            'input_ids': input_ids.tolist(),
            'attention_mask': attention_mask.tolist(),
        }
    
    def decode(self, token_ids, skip_special_token=False):
        return "translated text sample"

# Create a minimal constraint function
def constraint_function(text):
    return len(text) > 5

# Create args object
class Args:
    def __init__(self):
        self.samples_file = os.path.join(tempfile.gettempdir(), 'test_samples.pt')
        self.temperature = 0.9
        self.device = 'cpu'
        self.max_length = 10
        self.sample_batch_size = 8
        self.batch_size = 4

# Build the arguments dictionary
model = MockMarianMTModel()
tokenizer = MockTokenizer()
source_texts = ["Hello world", "How are you"]
args = Args()

ARGS = {
    'model': model,
    'tokenizer': tokenizer,
    'source_texts': source_texts,
    'constraint_function': constraint_function,
    'args': args,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch, tqdm — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is 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
sample_from_marianMT2 Ran
ran; our contract draft was wrong, not the code
mtsomethree/constrdecoding/train_MT.py
code served (permissive licence) · get_code("644cd55e35f5f0b5")
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 tuple. 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 torch.nn as nn
from unittest.mock import Mock

torch.manual_seed(0)

# Create a minimal mock model
class MockMarianMTModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.config = Mock()
        self.config.pad_token_id = 0
        self.linear = nn.Linear(4, 4)
    
    def set_constraint_factor(self, value):
        pass
    
    def set_temperature(self, value):
        pass
    
    def generate(self, input_ids, attention_mask, do_sample=False, output_scores=False, return_dict_in_generate=False):
        batch_size = input_ids.shape[0]
        seq_len = 5
        output = Mock()
        output.sequences = torch.randint(1, 100, (batch_size, seq_len))
        output.sequences_scores = torch.randn(batch_size)
        return output

# Create a minimal mock tokenizer
class MockTokenizer:
    def __call__(self, texts, return_tensors=None, padding=False, truncation=False, max_length=None):
        batch_size = len(texts) if isinstance(texts, list) else 1
        seq_len = 6
        result = {
            'input_ids': torch.randint(1, 100, (batch_size, seq_len)).tolist(),
            'attention_mask': torch.ones(batch_size, seq_len).tolist()
        }
        return result
    
    def decode(self, token_ids, skip_special_token=False):
        return "translated text"

# Create a simple constraint function
def constraint_function(text):
    return True

# Create args object
class Args:
    def __init__(self):
        self.samples_file = "/tmp/nonexistent_samples_file_12345.pt"
        self.temperature = 0.9
        self.device = 'cpu'
        self.max_length = 128
        self.sample_batch_size = 2
        self.batch_size = 1

# Create source texts
source_texts = ["Hello world", "How are you"]

# Build ARGS dictionary
ARGS = {
    'model': MockMarianMTModel(),
    'tokenizer': MockTokenizer(),
    'source_texts': source_texts,
    'constraint_function': constraint_function,
    'args': Args()
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch, tqdm — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is 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

Also present in other papers' code

Of 2 samples lifted from this paper's linked repositories, 2 carry a function identity and 0 do not. Of the 2 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 2 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

We propose a general and efficient framework to control auto-regressive generation models with NeurAlly-Decomposed Oracle (NADO). Given a pre-trained base language model and a sequence-level boolean oracle function, we propose to decompose the oracle function into token-level guidance to steer the base model in text generation. Specifically, the token-level guidance is approximated by a neural model trained with examples sampled from the base model, demanding no additional auxiliary labeled data. Based on posterior regularization, we present the closed-form optimal solution to incorporate the token-level guidance into the base model for controllable generation. We further provide a theoretical analysis of how the approximation quality of NADO affects the controllable generation results. Experiments conducted on two tasks: (1) text generation with lexical constraints and (2) machine translation with formality control demonstrate that our framework efficiently guides the base model towards the given control factors while maintaining high generation quality.

For agents

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

get_harvested_code_for_paper("2205.14219")
get_code_for_paper("2205.14219")
have("2205.14219")

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

curl https://syntology.ai/api/ran/2205.14219.json

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

[![Syntology run record](https://syntology.ai/api/ran/2205.14219.svg)](https://syntology.ai/paper/2205.14219)

Connect an agent — have() is free.