SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2201.03514 · ICML · 2022

Black-Box Tuning for Language-Model-as-a-Service

Tianxiang Sun, Yunfan Shao, Hong Qian, Xuanjing Huang, Xipeng Qiu

arXiv · PDF · Open in the Atlas

Code that ran

3 of 4 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 2 have no contract verdict; 1 failure of our instrument; 1 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/2201.03514.json · badge: /api/ran/2201.03514.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
jubgjf/plmtuningcompetition community (classifier label)
classifier basis not recorded on the link
3 ran · 1 no recorded run
FunctionStatusWhere it lives
convert_to_features Ran
ran on a synthesized input
jubgjf/plmtuningcompetition/dataloader.py
code served (permissive licence) · get_code("0dfc4ac31a46001f")
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 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 torch
import torch.nn as nn

torch.manual_seed(0)

class MockTokenizer:
    def __init__(self):
        self.mask_token_id = 103
    
    def batch_encode_plus(self, texts, add_special_tokens=True):
        # Return mock encodings with input_ids and attention_mask
        batch_size = len(texts)
        # Create input_ids with mask token present for the first element
        input_ids = [
            [101, self.mask_token_id, 102, 0],  # First has mask token at index 1
            [101, 2054, self.mask_token_id, 102]  # Second has mask token at index 2
        ][:batch_size]
        
        attention_mask = [
            [1, 1, 1, 0],
            [1, 1, 1, 1]
        ][:batch_size]
        
        return {
            'input_ids': input_ids,
            'attention_mask': attention_mask
        }

example_batch = {
    'input_text': ['what is this', 'who are you'],
    'target_text': ['answer one', 'answer two']
}

tokenizer = MockTokenizer()

ARGS = {
    'example_batch': example_batch,
    'tokenizer': tokenizer
}
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 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
create_position_ids_from_input_ids Ran
ran; our contract draft was wrong, not the code
jubgjf/plmtuningcompetition/models/deep_modeling_roberta.py
code served (permissive licence) · get_code("bbfc47d976ee7225")
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 Tensor, ndim 2, dtype family int, shape (2, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape 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

torch.manual_seed(0)

# Create a small batch of input_ids tensor
# Shape: (batch_size=2, seq_length=4)
# Values: token indices, with some padding tokens mixed in
input_ids = torch.tensor([
    [1, 2, 0, 3],
    [5, 0, 0, 4]
], dtype=torch.long)

# padding_idx is typically 0 in most transformer models
padding_idx = 0

ARGS = {
    "input_ids": input_ids,
    "padding_idx": padding_idx,
}
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 Tensor, ndim 2, dtype family int, shape (2, 4) — shape (2, 4) as a later run recorded it
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
hinge_loss Ran
ran on a synthesized input
jubgjf/plmtuningcompetition/utils.py
code served (permissive licence) · get_code("7bc8bc9cbcbec1d0")
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 Tensor, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape 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

torch.manual_seed(0)

# Parameters for hinge_loss function
# logit: (N, C, d_1, d_2, ..., d_K) - let's use (N=2, C=4, d_1=3)
logit = torch.randn(2, 4, 3)

# target: (N, d_1, d_2, ..., d_K) - let's use (N=2, d_1=3)
# target values should be valid class indices in range [0, C-1], so [0, 3]
target = torch.randint(0, 4, (2, 3))

# margin: float - a small positive value
margin = 1.0

# reduction: str - one of 'sum', 'mean', 'none'
reduction = 'sum'

ARGS = {
    'logit': logit,
    'target': target,
    'margin': margin,
    'reduction': reduction
}
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 Tensor, ndim 0, dtype family float, shape () — shape () as a later run recorded it
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
load_hf_dataset Not yet run
unverified (harvested, no recorded run)
jubgjf/plmtuningcompetition/dataloader.py
code served (permissive licence) · get_code("94689a629c7580ff")
How it ran

no invocation recorded.

Also present in other papers' code

Of 4 samples lifted from this paper's linked repositories, 4 carry a function identity and 0 do not. Of the 4 distinct functions with an identity, 3 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 3 later) and 1 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
create_position_ids_from_input_ids 8 other papers
2 dated earlier, 6 later
2021.emnlp-main.154 (2021 (year only))
Reconstruction Attack on Instance Encoding for Language Understanding
2104.08247 (2021-04)
What to Pre-Train on? Efficient Intermediate Task Selection
2023.acl-long.264 (2023 (year only))
Towards Robust Low-Resource Fine-Tuning with Multi-View Compressed Representations
and 5 more not named here
convert_to_features 1 other paper
1 later
2305.00593 (2023-05)
Reliable Gradient-free and Likelihood-free Prompt Tuning
hinge_loss 1 other paper
1 later
2305.00593 (2023-05)
Reliable Gradient-free and Likelihood-free Prompt Tuning

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

Extremely large pre-trained language models (PTMs) such as GPT-3 are usually released as a service. It allows users to design task-specific prompts to query the PTMs through some blackbox APIs. In such a scenario, which we call Language-Model-as-a-Service (LMaaS), the gradients of PTMs are usually unavailable. Can we optimize the task prompts by only accessing the model inference APIs? This paper proposes the black-box tuning framework to optimize the continuous prompt prepended to the input text via derivative-free optimization. Instead of optimizing in the original high-dimensional prompt space, which is intractable for traditional derivative-free optimization, we perform optimization in a randomly generated subspace due to the low intrinsic dimensionality of large PTMs. The experimental results show that the black-box tuning with RoBERTa on a few labeled samples not only significantly outperforms manual prompt and GPT-3's in-context learning, but also surpasses the gradient-based counterparts, i.e., prompt tuning and full model tuning.

For agents

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

get_harvested_code_for_paper("2201.03514")
get_code_for_paper("2201.03514")
have("2201.03514")

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

curl https://syntology.ai/api/ran/2201.03514.json

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

[![Syntology run record](https://syntology.ai/api/ran/2201.03514.svg)](https://syntology.ai/paper/2201.03514)

Connect an agent — have() is free.