SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2402.03898 · ICML · 2024

DISTILLM: Towards Streamlined Distillation for Large Language Models

Tianyi Chen, Se-Young Yun, Sungnyun Kim, Jongwoo Ko

arXiv · PDF · Open in the Atlas

Code that ran

5 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; 0 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 5 failures of our instrument; 0 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/2402.03898.json · badge: /api/ran/2402.03898.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
jongwooko/distillm canonical (classifier label)
classifier basis: classifier_2025
3 ran · 0 no recorded run
ghwang-s/abkd community
Papers with Code archive
1 ran · 0 no recorded run
microsoft/LMOps community
Papers with Code archive
1 ran · 0 no recorded run
FunctionStatusWhere it lives
ab_div Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
ghwang-s/abkd/distillation_llm/distillm/losses.py
pointer only (licence: NONE) · get_code("3595a0bc0ec56f02")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, ghwang-s/abkd when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

forward_kl Ran
ran; our contract draft was wrong, not the code
jongwooko/distillm/distillm/losses.py
pointer only (licence: NONE) · get_code("575e937196d69d18")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, jongwooko/distillm when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

get_distil_loss Ran
ran; our contract draft was wrong, not the code
microsoft/LMOps/minillm/finetune.py
code served (permissive licence) · get_code("cd958b79a0882e30")
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 a minimal tokenizer stand-in
class MinimalTokenizer:
    def __init__(self):
        self.vocab_size = 100
    
    def encode(self, text):
        return [1, 2, 3]
    
    def decode(self, ids):
        return "text"

# Create a minimal model that returns logits
class MinimalModel(nn.Module):
    def __init__(self, vocab_size=100, hidden_dim=4):
        super().__init__()
        self.embedding = nn.Embedding(vocab_size, hidden_dim)
        self.lm_head = nn.Linear(hidden_dim, vocab_size)
    
    def forward(self, input_ids, attention_mask=None, use_cache=False, **kwargs):
        x = self.embedding(input_ids)
        logits = self.lm_head(x)
        
        class Output:
            pass
        
        output = Output()
        output.logits = logits
        return output

# Create args object
class Args:
    def __init__(self):
        self.model_parallel = False

# Build components
tokenizer = MinimalTokenizer()
model = MinimalModel(vocab_size=100, hidden_dim=4)
teacher_model = MinimalModel(vocab_size=100, hidden_dim=4)
args = Args()

# Create batch tensors
batch_size = 2
seq_length = 3
vocab_size = 100

input_ids = torch.randint(0, vocab_size, (batch_size, seq_length))
attention_mask = torch.ones(batch_size, seq_length)

model_batch = {
    'input_ids': input_ids,
    'attention_mask': attention_mask,
}

no_model_batch = {
    'loss_mask': torch.ones(batch_size, seq_length),
    'label': torch.randint(-100, vocab_size, (batch_size, seq_length)),
}

# Create logits from model
with torch.no_grad():
    model_output = model(**model_batch, use_cache=False)
    logits = model_output.logits

ARGS = {
    'args': args,
    'tokenizer': tokenizer,
    'model': model,
    'teacher_model': teacher_model,
    'model_batch': model_batch,
    'no_model_batch': no_model_batch,
    'logits': logits,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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
reverse_kl Ran
ran; our contract draft was wrong, not the code
jongwooko/distillm/distillm/losses.py
pointer only (licence: NONE) · get_code("283147d481c8695c")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, jongwooko/distillm when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

symmetric_kl Ran
ran; our contract draft was wrong, not the code
jongwooko/distillm/distillm/losses.py
pointer only (licence: NONE) · get_code("49aa907f43ee85d3")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, jongwooko/distillm when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is 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 (0 with at least one paper dated earlier, 0 the same month, 1 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
symmetric_kl
brick of this paper's assembly: entry
1 other paper
1 later
2410.16236 (2024-10)
LLaVA-KD: A Framework of Distilling Multimodal Large Language Models

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

Knowledge distillation (KD) is widely used for compressing a teacher model to a smaller student model, reducing its inference cost and memory footprint while preserving model capabilities. However, current KD methods for autoregressive sequence models (e.g., large language models) suffer from missing a standardized objective function. Moreover, the recent use of studentgenerated outputs to address training-inference mismatches has significantly escalated computational costs. To tackle these issues, we introduce DISTILLM, a more effective and efficient KD framework for auto-regressive language models. DISTILLM comprises two components: (1) a novel skew Kullback-Leibler divergence loss, where we unveil and leverage its theoretical properties, and (2) an adaptive off-policy approach designed to enhance the efficiency in utilizing student-generated outputs. Extensive experiments, including instruction-following tasks, demonstrate the effectiveness of DISTILLM in building high-performing student models while achieving up to 4.3× speedup compared to recent KD methods.

For agents

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

get_harvested_code_for_paper("2402.03898")
get_code_for_paper("2402.03898")
have("2402.03898")

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

curl https://syntology.ai/api/ran/2402.03898.json

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

[![Syntology run record](https://syntology.ai/api/ran/2402.03898.svg)](https://syntology.ai/paper/2402.03898)

Connect an agent — have() is free.