SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2602.21788 · 2026

Efficient Scaling of LLM Training with Flexible Context Parallelism

Wei Zhou, Jia Li, Yifan Niu, Han Xiao, Dongyi Liu

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 (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; 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/2602.21788.json · badge: /api/ran/2602.21788.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
ivanniu/FCP canonical (classifier label)
classifier basis: census
6 ran · 2 no recorded run
FunctionStatusWhere it lives
get_batch Ran
ran; our contract draft was wrong, not the code
ivanniu/FCP/pretrain_transformers.py
pointer only (licence: NONE) · get_code("6d2504a551125090")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, ivanniu/FCP 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_img_processor_from_dataset Ran
ran; our contract draft was wrong, not the code
ivanniu/FCP/hybrid_dataloader.py
pointer only (licence: NONE) · get_code("11e5b987e3b2cb44")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, ivanniu/FCP 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.

make_pixel_loader Ran
ran; our contract draft was wrong, not the code
ivanniu/FCP/hybrid_dataloader.py
pointer only (licence: NONE) · get_code("25e79d5a03eda3d4")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, ivanniu/FCP 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.

monitor_memory Ran
ran; our contract draft was wrong, not the code
ivanniu/FCP/scheduler.py
pointer only (licence: NONE) · get_code("f062990405d7c60f")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, ivanniu/FCP 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.

pad_len_to Ran
ran, honoured the contract we drafted
ivanniu/FCP/scheduler.py
pointer only (licence: NONE) · get_code("6605653d855804dc")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, ivanniu/FCP 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.

timing Ran
ran; our contract draft was wrong, not the code
ivanniu/FCP/scheduler.py
pointer only (licence: NONE) · get_code("3b2b0b026933db79")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, ivanniu/FCP 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.

build_synthetic_dataloader Not yet run
unverified (harvested, no recorded run)
ivanniu/FCP/synthetic_dataloader.py
pointer only (licence: NONE) · get_code("1a295e561341f218")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, ivanniu/FCP 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. No invocation recorded.

make_length_sampler Not yet run
unverified (harvested, no recorded run)
ivanniu/FCP/synthetic_dataloader.py
pointer only (licence: NONE) · get_code("ef3269b7f382d564")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, ivanniu/FCP 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. 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, 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 8 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

Scaling long-context capabilities is crucial for Large Language Models (LLMs). However, real-world data contain a large number of sequences with heterogeneous lengths. Existing training libraries for LLMs rely on static parallelism strategies, which suffer from severe load imbalance, redundant communication, and suboptimal hardware utilization under data heterogeneity. In this work, we propose Flexible Context Parallelism (FCP), an efficient parallelism strategy that adaptively reconfigures communication groups and context parallelism degrees during LLM training. We generalize more flexible non-power-of-two parallelism degrees and develop a polynomial-time algorithm to generate near-optimal parallelism strategies with only millisecond-level overhead per training batch. FCP is able to maintain high hardware efficiency even under extreme data heterogeneity. Experimental results demonstrate that FCP significantly outperforms Megatron-LM and Deep-Speed in both LLM and MLLM training, achieving up to 1.46 × speedup in average throughput while maintaining near-linear scaling efficiency across large-scale clusters. For extremely unbalanced batches, FCP even achieves 2.24 × speedup. The code is available at https://github.com/ivanniu/FCP.

For agents

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

get_harvested_code_for_paper("2602.21788")
get_code_for_paper("2602.21788")
have("2602.21788")

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

curl https://syntology.ai/api/ran/2602.21788.json

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

[![Syntology run record](https://syntology.ai/api/ran/2602.21788.svg)](https://syntology.ai/paper/2602.21788)

Connect an agent — have() is free.