SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2409.01143 · 2024

HexiScale: Facilitating Large Language Model Training over Heterogeneous Hardware

arXiv · PDF · Open in the Atlas

Code that ran

1 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; 1 failure of our instrument; 4 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/2409.01143.json · badge: /api/ran/2409.01143.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
relaxed-system-lab/flashflex canonical (classifier label)
classifier basis: census
1 ran · 4 no recorded run
FunctionStatusWhere it lives
forward_step_func Ran
ran; our contract draft was wrong, not the code
relaxed-system-lab/flashflex/llama_train.py
code served (permissive licence) · get_code("401ab1df19f0bb57")
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
import torch.nn as nn
from typing import Tuple, List

torch.manual_seed(0)

# Create a simple model that can accept either tuple/list unpacking or single input
class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
    
    def forward(self, *args, **kwargs):
        # Handle both unpacked args (from tuple/list) and single tensor
        if len(args) == 1 and isinstance(args[0], torch.Tensor):
            x = args[0]
        elif len(args) > 1:
            # Multiple args passed - concatenate them
            x = torch.cat([arg for arg in args if isinstance(arg, torch.Tensor)], dim=-1)
        else:
            x = args[0]
        
        return self.linear(x)

model = SimpleModel()

# Test case 1: inputs as a single tensor
inputs_single = torch.randn(2, 4)

# Test case 2: inputs as a tuple of tensors
inputs_tuple = (torch.randn(2, 4),)

# Test case 3: inputs as a list of tensors
inputs_list = [torch.randn(2, 4)]

# Use the single tensor case
ARGS = {
    "inputs": inputs_single,
    "model": model
}
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
construct_graph Not yet run
unverified (harvested, no recorded run)
relaxed-system-lab/flashflex/experimental/graph.py
code served (permissive licence) · get_code("a25d59327ea592b2")
How it ran

no invocation recorded.

gen_layer_dp_rank_group Not yet run
unverified (harvested, no recorded run)
relaxed-system-lab/flashflex/experimental/cost_modeling.py
code served (permissive licence) · get_code("f0f91a238ab5b0be")
How it ran

no invocation recorded.

get_distributed_dataloader Not yet run
unverified (harvested, no recorded run)
relaxed-system-lab/flashflex/llama_train.py
code served (permissive licence) · get_code("d0d437ff5cce7f03")
How it ran

no invocation recorded.

throughput Not yet run
unverified (harvested, no recorded run)
relaxed-system-lab/flashflex/experimental/evaluate.py
code served (permissive licence) · get_code("7916d03a2e93897a")
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, 2 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 2 later) and 3 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
gen_layer_dp_rank_group 1 other paper
1 later
2510.16415 (2025-10)
MeCeFO: Enhancing LLM Training Robustness via Fault-Tolerant Optimization
throughput 1 other paper
1 later
2510.16415 (2025-10)
MeCeFO: Enhancing LLM Training Robustness via Fault-Tolerant Optimization

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

Training large language models (LLMs) is a computationally intensive task, which is typically conducted in data centers with homogeneous high-performance GPUs. In this paper, we explore an alternative approach by deploying training computations across heterogeneous GPUs to enable better flexibility and efficiency for heterogeneous resource utilization. Toward this end, we propose a novel system, HexiScale, that can flexibly support asymmetric partition of training computations in the scope of data-, pipeline-, and tensor model parallelism. We further formalize the allocation of asymmetric partitioned training computations over a set of heterogeneous GPUs as a constrained optimization problem and propose an efficient hierarchical graph partitioning algorithm. Our approach effectively allocates training computations across heterogeneous GPUs, fully leveraging the available computational power. We compare the performance of HexiScale with state-of-the-art homogeneous and heterogeneous training systems. When training LLMs at different scales (from 7B to 30B), empirical results demonstrate that: (i) compared to state-of-the-art homogeneous baselines running over homogeneous GPUs, HexiScale achieves similar performance when running over heterogeneous GPUs with the same theoretical FLOPS; (ii) compared to state-of-the-art heterogeneous baselines running on the same heterogeneous clusters, HexiScale delivers $1.5\times$ to $2.4\times$ higher throughput.

For agents

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

get_harvested_code_for_paper("2409.01143")
get_code_for_paper("2409.01143")
have("2409.01143")

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

curl https://syntology.ai/api/ran/2409.01143.json

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

[![Syntology run record](https://syntology.ai/api/ran/2409.01143.svg)](https://syntology.ai/paper/2409.01143)

Connect an agent — have() is free.