SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2402.14526 · 2024

Balanced Data Sampling for Language Model Training with Clustering

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, 0 violated it and 0 have no contract verdict; 3 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/2402.14526.json · badge: /api/ran/2402.14526.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
choosewhatulike/cluster-clip canonical (classifier label)
classifier basis: classifier_2025
4 ran · 1 no recorded run
FunctionStatusWhere it lives
distance_dot_product Ran
ran, honoured the contract we drafted
choosewhatulike/cluster-clip/kmeans.py
code served (permissive licence) · get_code("091a632e49097468")
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, 3) 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 small tensors for the function
# x: shape (batch_size, feature_dim) = (2, 4)
# centroids: shape (num_centroids, feature_dim) = (3, 4)

x = torch.randn(2, 4)
centroids = torch.randn(3, 4)

ARGS = {
    "x": x,
    "centroids": centroids,
}
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, 3) — shape (2, 3) 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
first_pooling Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
choosewhatulike/cluster-clip/run_gen_emb.py
code served (permissive licence) · get_code("9293dcb452ed9a4a")
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

torch.manual_seed(0)

# The function expects:
# - model_output: indexable by [0] to get token_embeddings
# - attention_mask: passed but not used in the function body
# - token_embeddings should be indexable by [:, 0] (2D tensor)

# Create model_output as a tuple where first element is token_embeddings
# Shape: (batch_size=2, sequence_length=5, hidden_dim=4)
token_embeddings = torch.randn(2, 5, 4)
model_output = (token_embeddings,)

# Create attention_mask (required parameter but not used in function)
attention_mask = torch.ones(2, 5)

ARGS = {
    "model_output": model_output,
    "attention_mask": attention_mask,
}
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
load_embeddings Ran
ran; our contract draft was wrong, not the code
choosewhatulike/cluster-clip/kmeans.py
code served (permissive licence) · get_code("0fb325eb5b6a375b")
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 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 os
import tempfile
import shutil

torch.manual_seed(0)

# Create a temporary directory structure with mock embedding files
temp_dir = tempfile.mkdtemp()
doc_embedding_dir = os.path.join(temp_dir, 'doc_embedding')
os.makedirs(doc_embedding_dir, exist_ok=True)

# Create mock embedding files
for i in range(4):
    embedding_tensor = torch.randn(3 + i, 4)  # Different sizes for each file
    filepath = os.path.join(doc_embedding_dir, f'embedding_{i}.pt')
    torch.save(embedding_tensor, filepath)

ARGS = {
    'input_dir': temp_dir,
    'world_size': 2,
    'rank': 0
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
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
mean_pooling Ran
ran; our contract draft was wrong, not the code
choosewhatulike/cluster-clip/run_gen_emb.py
code served (permissive licence) · get_code("ee3e60cc57e00353")
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

torch.manual_seed(0)

# Construct model_output as a tuple containing token_embeddings
# The function accesses model_output[0], so we need a tuple/list
batch_size = 2
seq_length = 3
embedding_dim = 4

token_embeddings = torch.randn(batch_size, seq_length, embedding_dim)
model_output = (token_embeddings,)

# Construct attention_mask
# Shape should be (batch_size, seq_length) to match token_embeddings
attention_mask = torch.ones(batch_size, seq_length)

ARGS = {
    "model_output": model_output,
    "attention_mask": attention_mask
}
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
init_memory Not yet run
unverified (harvested, no recorded run)
choosewhatulike/cluster-clip/kmeans.py
code served (permissive licence) · get_code("c8d10e3fa8c47d55")
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, 1 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 0 later) and 4 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
mean_pooling 2 other papers
2 dated earlier
2101.06804 (2021-01)
What Makes Good In-Context Examples for GPT-$3$?
2108.01850 (2021-08)
Controlled Text Generation as Continuous Optimization with Multiple Constraints

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

Data plays a fundamental role in the training of Large Language Models (LLMs). While attention has been paid to the collection and composition of datasets, determining the data sampling strategy in training remains an open question. Most LLMs are trained with a simple strategy, random sampling. However, this sampling strategy ignores the unbalanced nature of training data distribution, which can be sub-optimal. In this paper, we propose ClusterClip Sampling to balance the text distribution of training data for better model training. Specifically, ClusterClip Sampling utilizes data clustering to reflect the data distribution of the training set and balances the common samples and rare samples during training based on the cluster results. A repetition clip operation is introduced to mitigate the overfitting issue led by samples from certain clusters. Extensive experiments validate the effectiveness of ClusterClip Sampling, which outperforms random sampling and other cluster-based sampling variants under various training datasets and large language models.

For agents

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

get_harvested_code_for_paper("2402.14526")
get_code_for_paper("2402.14526")
have("2402.14526")

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

curl https://syntology.ai/api/ran/2402.14526.json

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

[![Syntology run record](https://syntology.ai/api/ran/2402.14526.svg)](https://syntology.ai/paper/2402.14526)

Connect an agent — have() is free.