SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2108.13161 · ICLR · 2022

Differentiable Prompt Makes Pre-trained Language Models Better Few-shot Learners

Chuanqi Tan, Xiang Chen, Fei Huang, Ningyu Zhang, Luoqiu Li, Shumin Deng, Zhen Bi, Huajun Chen

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 (3 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 3 have no contract verdict; 0 failures 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/2108.13161.json · badge: /api/ran/2108.13161.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
zjunlp/DART marked official in Papers with Code
Papers with Code archive
2 ran · 0 no recorded run
zhaohan-xi/plm-prompt-defense community
Papers with Code archive
1 ran · 1 no recorded run
FunctionStatusWhere it lives
DiffPET Ran
ran on a synthesized input · constructed an object; no result was computed
zhaohan-xi/plm-prompt-defense/src/model.py
code served (permissive licence) · get_code("a878fc6c8d69a825")
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 an instance of its own class, DiffPET: the run constructed the object and did not call it, so no result was computed. 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
from torch import nn
from transformers import GPT2Tokenizer

torch.manual_seed(0)

# Create a minimal tokenizer stand-in
class MinimalTokenizer:
    def __init__(self):
        self.vocab_size = 50257
    
    def encode(self, text, add_special_tokens=False, add_prefix_space=False):
        # Return a list of token IDs based on text length
        return [100 + len(text)] if text else [100]

# Create a reader with PATTERN and VERBALIZERS
class MinimalReader:
    PATTERN = ["[X]", "is", "[Y]"]
    VERBALIZERS = ["positive", "negative"]

# Create a minimal model
class MinimalModel(nn.Module):
    def __init__(self, vocab_size=50257, hidden_size=4):
        super().__init__()
        self.embedding = nn.Embedding(vocab_size, hidden_size)
        self.output_proj = nn.Linear(hidden_size, vocab_size)
    
    def forward(self, input_ids, attention_mask=None, token_type_ids=None):
        # Return logits of shape [batch_size, seq_len, vocab_size]
        embedded = self.embedding(input_ids)
        logits = self.output_proj(embedded)
        return (logits,)
    
    def get_input_embeddings(self):
        return self.embedding

# Create batch data
def create_batch():
    batch_size = 2
    seq_len = 5
    batch = {
        'input_ids': torch.randint(0, 1000, (batch_size, seq_len), dtype=torch.long),
        'attention_mask': torch.ones(batch_size, seq_len, dtype=torch.long),
        'token_type_ids': torch.zeros(batch_size, seq_len, dtype=torch.long),
        'pet_flags': torch.full((batch_size, seq_len), -1, dtype=torch.long),
        'label_ids': torch.tensor([0, 1], dtype=torch.long),
        'pet_logits': torch.randn(batch_size, seq_len, 50257),
    }
    return batch

# Setup arguments
tokenizer = MinimalTokenizer()
reader = MinimalReader()
model = MinimalModel(vocab_size=50257, hidden_size=4)
device = torch.device('cpu')

ARGS = {
    'tokenizer': tokenizer,
    'reader': reader,
    'model': model,
    'device': device,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, transformers — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DiffPET — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a DiffPET 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
DiffPET Ran
ran on a synthesized input · constructed an object; no result was computed
zjunlp/DART/src/model.py
code served (permissive licence) · get_code("fae89cb97e18344e")
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 an instance of its own class, DiffPET: the run constructed the object and did not call it, so no result was computed. 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
from torch import nn
from transformers import GPT2Tokenizer

torch.manual_seed(0)

# Create a minimal tokenizer stand-in
class MinimalTokenizer:
    def __init__(self):
        self.vocab_size = 50257
    
    def encode(self, text, add_special_tokens=False, add_prefix_space=False):
        # Return deterministic token ids based on text
        return [hash(text) % self.vocab_size]

# Create a minimal reader with PATTERN and VERBALIZERS
class MinimalReader:
    PATTERN = ["[PROMPT]", "text", "[MASK]"]
    VERBALIZERS = ["positive", "negative"]

# Create a minimal model
class MinimalModel(nn.Module):
    def __init__(self, vocab_size=50257, hidden_dim=4):
        super().__init__()
        self.embeddings = nn.Embedding(vocab_size, hidden_dim)
        self.output_layer = nn.Linear(hidden_dim, vocab_size)
    
    def get_input_embeddings(self):
        return self.embeddings
    
    def forward(self, input_ids, attention_mask, token_type_ids):
        # Return logits of shape (batch_size, seq_len, vocab_size)
        embedded = self.embeddings(input_ids)
        logits = self.output_layer(embedded)
        return (logits,)

# Create instances
tokenizer = MinimalTokenizer()
reader = MinimalReader()
model = MinimalModel(vocab_size=50257, hidden_dim=4)
device = 'cpu'

ARGS = {
    'tokenizer': tokenizer,
    'reader': reader,
    'model': model,
    'device': device
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, transformers — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DiffPET — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a DiffPET 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
PET Ran
ran on a synthesized input · constructed an object; no result was computed
zjunlp/DART/src/model.py
code served (permissive licence) · get_code("f50a2e625707084d")
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 an instance of its own class, PET: the run constructed the object and did not call it, so no result was computed. 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
from torch import nn
from transformers import GPT2Tokenizer

torch.manual_seed(0)

# Stand-in tokenizer
class StandInTokenizer:
    def __init__(self):
        pass
    
    def encode(self, text, add_special_tokens=False, add_prefix_space=False):
        # Return a list with a single token ID
        return [42]

# Stand-in reader with VERBALIZERS
class StandInReader:
    VERBALIZERS = ["yes", "no"]

# Minimal model that returns logits
class MinimalModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 50257)  # GPT2 vocab size
    
    def forward(self, input_ids, attention_mask, token_type_ids):
        # Return tuple with logits as first element
        batch_size = input_ids.shape[0]
        seq_len = input_ids.shape[1]
        logits = self.linear(torch.randn(batch_size, seq_len, 4))
        return (logits,)

# Create instances
tokenizer = StandInTokenizer()
reader = StandInReader()
model = MinimalModel()
device = 'cpu'

ARGS = {
    'tokenizer': tokenizer,
    'reader': reader,
    'model': model,
    'device': device
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, transformers — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type PET — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a PET 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
PET Not yet run
unverified (harvested, no recorded run)
zhaohan-xi/plm-prompt-defense/src/model.py
code served (permissive licence) · get_code("9f8852fc911e2924")
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 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-scale pre-trained language models have contributed significantly to natural language processing by demonstrating remarkable abilities as few-shot learners. However, their effectiveness depends mainly on scaling the model parameters and prompt design, hindering their implementation in most real-world applications. This study proposes a novel pluggable, extensible, and efficient approach named DifferentiAble pRompT (DART), which can convert small language models into better few-shot learners. The main principle behind this approach involves reformulating potential natural language processing tasks into the task of a pre-trained language model and differentially optimizing the prompt template as well as the target label with backpropagation. Furthermore, the proposed approach can be: (i) Plugged to any pre-trained language models; (ii) Extended to widespread classification tasks. A comprehensive evaluation of standard NLP tasks demonstrates that the proposed approach achieves a better few-shot performance 1 .

For agents

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

get_harvested_code_for_paper("2108.13161")
get_code_for_paper("2108.13161")
have("2108.13161")

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

curl https://syntology.ai/api/ran/2108.13161.json

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

[![Syntology run record](https://syntology.ai/api/ran/2108.13161.svg)](https://syntology.ai/paper/2108.13161)

Connect an agent — have() is free.