SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2203.02485 · ICLR · 2022

Better Supervisory Signals by Observing Learning Paths

Yi Ren, Shangmin Guo, Danica Sutherland

arXiv · PDF · Open in the Atlas

Code that ran

4 of 5 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; 1 honoured the contract we drafted, 2 violated it and 0 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/2203.02485.json · badge: /api/ran/2203.02485.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
joshua-ren/better_supervisory_signal canonical (classifier label)
classifier basis: classifier_2025
4 ran · 1 no recorded run
FunctionStatusWhere it lives
ce_loss Ran
ran, honoured the contract we drafted
joshua-ren/better_supervisory_signal/utils.py
code served (permissive licence) · get_code("21fe6fa61f6e0652")
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
import torch.nn as nn

torch.manual_seed(0)

# y should be a tensor of logits with shape (batch_size, num_classes)
# CrossEntropyLoss expects raw scores, not probabilities
y = torch.randn(2, 4)

# labels should be a tensor of class indices with shape (batch_size, 1) or (batch_size,)
# The function calls .squeeze() on it, so we make it (batch_size, 1)
labels = torch.randint(0, 4, (2, 1))

ARGS = {
    "y": y,
    "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 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
distil_loss Ran
ran, violated the contract we drafted
joshua-ren/better_supervisory_signal/utils.py
code served (permissive licence) · get_code("da57ff4737709565")
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 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
import torch.nn as nn
import torch.nn.functional as F

torch.manual_seed(0)

# Create small tensors for the function
batch_size = 2
num_classes = 4

# y: logits from student model (batch_size, num_classes)
y = torch.randn(batch_size, num_classes)

# labels: ground truth labels (batch_size, 1) - will be squeezed in ce_loss
labels = torch.randint(0, num_classes, (batch_size, 1))

# teacher_scores: logits from teacher model (batch_size, num_classes)
teacher_scores = torch.randn(batch_size, num_classes)

# Temperature parameter
T = 4

# Alpha parameter for weighting KD loss vs CE loss
alpha = 0.5

ARGS = {
    'y': y,
    'labels': labels,
    'teacher_scores': teacher_scores,
    'T': T,
    'alpha': alpha
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch, torchvision — 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
get_validation Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
joshua-ren/better_supervisory_signal/main_distill.py
code served (permissive licence) · get_code("fdceb841db6be746")
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, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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
import torch.nn as nn
import numpy as np

torch.manual_seed(0)

# Define a minimal model that satisfies the function's usage
class MinimalModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)  # 4 input features, 3 output classes
    
    def forward(self, x):
        return self.linear(x)

# Create a minimal data loader
# The function expects: x, _, ny, idx from each batch
# x: input tensor, _: unused, ny: labels, idx: indices
class MinimalDataLoader:
    def __init__(self, batch_size=2):
        self.batch_size = batch_size
        self.data = [
            (torch.randn(2, 4), None, torch.randint(0, 3, (2, 1)), torch.arange(2)),
            (torch.randn(2, 4), None, torch.randint(0, 3, (2, 1)), torch.arange(2, 4))
        ]
    
    def __iter__(self):
        return iter(self.data)

# Define a stub for cal_ECE function that the code calls
def cal_ECE(pb_table, tf_table):
    return 0.1

# Patch cal_ECE into the module namespace
import sys
current_module = sys.modules[__name__]
current_module.cal_ECE = cal_ECE

model = MinimalModel()
data_loader = MinimalDataLoader(batch_size=2)

ARGS = {
    "model": model,
    "data_loader": data_loader
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family float — a tuple 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
kd_loss Ran
ran, violated the contract we drafted
joshua-ren/better_supervisory_signal/utils.py
code served (permissive licence) · get_code("741d6b3c4d0906d4")
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
import torch.nn as nn
import torch.nn.functional as F

torch.manual_seed(0)

# y: model output logits, should be a tensor
# teacher_scores: teacher model output logits, should be a tensor
# T: temperature parameter (default 1)

# Create small tensors: batch size 2, num classes 4
batch_size = 2
num_classes = 4

y = torch.randn(batch_size, num_classes)
teacher_scores = torch.randn(batch_size, num_classes)
T = 1

ARGS = {
    'y': y,
    'teacher_scores': teacher_scores,
    'T': T
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, torchvision — 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
get_init_net Not yet run
unverified (harvested, no recorded run)
joshua-ren/better_supervisory_signal/utils.py
code served (permissive licence) · get_code("edb82c7ef88ade05")
How it ran

no invocation recorded.

Also present in other papers' code

Of 5 samples lifted from this paper's linked repositories, 5 carry a function identity and 0 do not. Of the 5 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 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.

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

Better-supervised models might have better performance. In this paper, we first clarify what makes for good supervision for a classification problem, and then explain two existing label refining methods, label smoothing and knowledge distillation, in terms of our proposed criterion. To further answer why and how better supervision emerges, we observe the learning path, i.e., the trajectory of the model's predictions during training, for each training sample. We find that the model can spontaneously refine "bad" labels through a "zig-zag" learning path, which occurs on both toy and real datasets. Observing the learning path not only provides a new perspective for understanding knowledge distillation, overfitting, and learning dynamics, but also reveals that the supervisory signal of a teacher network can be very unstable near the best points in training on real tasks. Inspired by this, we propose a new knowledge distillation scheme, Filter-KD, which improves downstream classification performance in various settings.

For agents

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

get_harvested_code_for_paper("2203.02485")
get_code_for_paper("2203.02485")
have("2203.02485")

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

curl https://syntology.ai/api/ran/2203.02485.json

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

[![Syntology run record](https://syntology.ai/api/ran/2203.02485.svg)](https://syntology.ai/paper/2203.02485)

Connect an agent — have() is free.