SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2403.01570 · ICLR · 2025

Small Models are LLM Knowledge Triggers for Medical Tabular Prediction

Jimeng Sun, Bo Zheng, Jintai Chen, Chaowen Hu, Jiahuan Yan, Jian Wu, Yaojun Hu

arXiv · PDF · Open in the Atlas

Code that ran

6 of 8 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (1 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; 5 failures of our instrument; 2 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/2403.01570.json · badge: /api/ran/2403.01570.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
jyansir/sersal community
Papers with Code archive
6 ran · 2 no recorded run
FunctionStatusWhere it lives
MultiheadGEAttention Ran
ran on a synthesized input · constructed an object; no result was computed
jyansir/sersal/models/t2g.py
code served (permissive licence) · get_code("1050129a2986d571")
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, MultiheadGEAttention: 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
import torch.nn as nn
import math

torch.manual_seed(0)

# Define small stand-in compression modules
class CompressionModule(nn.Module):
    def __init__(self, in_dim, out_dim):
        super().__init__()
        self.linear = nn.Linear(in_dim, out_dim)
    
    def forward(self, x):
        return self.linear(x)

# Set up dimensions
d = 8  # embedding dimension
n_heads = 2  # number of heads
dropout = 0.1
initialization = 'xavier'
n = 3  # number of columns (n + 1 = 4 total)

# Create compression modules
key_compression = CompressionModule(d, d)
value_compression = CompressionModule(d, d)

# Create input tensors
batch_size = 2
n_tokens = 5
x_head = torch.randn(batch_size, n_tokens, d)
x_tail = torch.randn(batch_size, n_tokens, d)

ARGS = {
    'd': d,
    'n_heads': n_heads,
    'dropout': dropout,
    'initialization': initialization,
    'n': n,
    'sym_weight': True,
    'sym_topology': False,
    'nsi': True,
}
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 MultiheadGEAttention — 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 MultiheadGEAttention 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
geglu Ran
ran; our contract draft was wrong, not the code
jyansir/sersal/models/t2g.py
code served (permissive licence) · get_code("d7bace13fb369137")
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 float, shape (2, 2) 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
import torch.nn.functional as F

torch.manual_seed(0)

# The geglu function takes a single parameter x which is a Tensor
# It chunks x along the last dimension into 2 equal parts
# So x needs to have an even size in the last dimension
# Let's create a small tensor with batch size 2 and last dimension 4 (even)

x = torch.randn(2, 4)

ARGS = {
    "x": x
}
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 float, shape (2, 2) — shape (2, 2) 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
get_activation_fn Ran
ran; our contract draft was wrong, not the code
jyansir/sersal/models/t2g.py
code served (permissive licence) · get_code("7627d2c02619ab87")
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 function. 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.functional as F

torch.manual_seed(0)

# The function get_activation_fn takes a single parameter: name (str)
# It returns a callable that takes a Tensor and returns a Tensor
# We need to test it with different activation function names

ARGS = {
    "name": "geglu"
}
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 function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function 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
get_nonglu_activation_fn Ran
ran; our contract draft was wrong, not the code
jyansir/sersal/models/t2g.py
code served (permissive licence) · get_code("44d9a7feaba1dc68")
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 builtin_function_or_method. 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.functional as F
from torch import Tensor

torch.manual_seed(0)

ARGS = {
    'name': 'geglu'
}
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 builtin_function_or_method — a builtin_function_or_method output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a builtin_function_or_method 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
reglu Ran
ran; our contract draft was wrong, not the code
jyansir/sersal/models/t2g.py
code served (permissive licence) · get_code("2fb665ac3ecd7c20")
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 float, 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
import torch.nn.functional as F

torch.manual_seed(0)

# The function reglu takes a Tensor x, chunks it in half along the last dimension,
# and returns a * relu(b). So we need a tensor with an even last dimension.
x = torch.randn(2, 8)  # batch size 2, last dim 8 (will be chunked into two size-4 tensors)

ARGS = {
    "x": x
}
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 float, 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
tanglu Ran
ran; our contract draft was wrong, not the code
jyansir/sersal/models/t2g.py
code served (permissive licence) · get_code("dc107686161af683")
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 float, 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
from torch import Tensor

torch.manual_seed(0)

# The function tanglu takes a Tensor and chunks it along the last dimension
# into two equal parts, then returns a * tanh(b)
# So we need a tensor with an even last dimension

x = torch.randn(2, 8)  # batch size 2, last dim 8 (will be chunked into 2x4)

ARGS = {
    "x": x
}
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 float, 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
T2GFormer Not yet run
unverified (harvested, no recorded run)
jyansir/sersal/models/t2g.py
code served (permissive licence) · get_code("0bbf2ee0b377a20d")
How it ran

no invocation recorded.

Tokenizer Not yet run
unverified (harvested, no recorded run)
jyansir/sersal/models/t2g.py
code served (permissive licence) · get_code("d0e255c6c68af4b2")
How it ran

no invocation recorded.

Also present in other papers' code

Of 8 samples lifted from this paper's linked repositories, 8 carry a function identity and 0 do not. Of the 8 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, 0 later) and 5 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
geglu
brick of this paper's assembly: helper
3 other papers
3 dated earlier
2106.11959 (2021-06)
Revisiting Deep Learning Models for Tabular Data
2207.03208 (2022-07)
Revisiting Pretraining Objectives for Tabular Deep Learning
2211.16887 (2022-11)
T2G-Former: Organizing Tabular Features into Relation Graphs Promotes Heterogeneous Feature Interaction
reglu
brick of this paper's assembly: helper
3 other papers
3 dated earlier
2106.11959 (2021-06)
Revisiting Deep Learning Models for Tabular Data
2207.03208 (2022-07)
Revisiting Pretraining Objectives for Tabular Deep Learning
2211.16887 (2022-11)
T2G-Former: Organizing Tabular Features into Relation Graphs Promotes Heterogeneous Feature Interaction
tanglu
brick of this paper's assembly: helper
1 other paper
1 dated earlier
2211.16887 (2022-11)
T2G-Former: Organizing Tabular Features into Relation Graphs Promotes Heterogeneous Feature Interaction

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

Recent development in large language models (LLMs) has demonstrated impressive domain proficiency on unstructured textual or multi-modal tasks. However, despite with intrinsic world knowledge, their application on structured tabular data prediction still lags behind, primarily due to the numerical insensitivity and modality discrepancy that brings a gap between LLM reasoning and statistical tabular learning. Unlike textual or vision data (e.g., electronic clinical notes or medical imaging data), tabular data is often presented in heterogeneous numerical values (e.g., CBC reports). This ubiquitous data format requires intensive expert annotation, and its numerical nature limits LLMs' capability to effectively transfer untapped domain expertise. In this paper, we propose SERSAL, a general selfprompting method by synergy learning with small models to enhance LLM tabular prediction in an unsupervised manner. Specifically, SERSAL utilizes the LLM's prior outcomes as original soft noisy annotations, which are dynamically leveraged to teach a better small student model. Reversely, the outcomes from the trained small model are used to teach the LLM to further refine its real capability. This process can be repeatedly applied to gradually distill refined knowledge for continuous progress. Comprehensive experiments on widely used medical domain tabular datasets show that, without access to gold labels, applying SERSAL to OpenAI GPT reasoning process attains substantial improvement compared to linguistic prompting methods, which serves as an orthogonal direction for tabular LLM, and increasing prompting bonus is observed as more powerful LLMs appear. Codes are available at https://github.com/jyansir/sersal.

For agents

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

get_harvested_code_for_paper("2403.01570")
get_code_for_paper("2403.01570")
have("2403.01570")

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

curl https://syntology.ai/api/ran/2403.01570.json

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

[![Syntology run record](https://syntology.ai/api/ran/2403.01570.svg)](https://syntology.ai/paper/2403.01570)

Connect an agent — have() is free.