SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2605.28079 · 2026

ATLAS: All-round Testing of Long-context Abilities across Scales

Chen Zhang, Xiaoyu Li, Yixin Cao, Meituan, Jiaqi Zhang, Jingang Wang, Xunliang Cai, Hongyin Tang, Cunguang Wang, Xuezhi Cao, Anchun Gui, Zhe Tang, and 7 more

arXiv · PDF · Open in the Atlas

Code that ran

1 of 1 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (1 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 1 has no contract verdict; 0 failures of our instrument; 0 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/2605.28079.json · badge: /api/ran/2605.28079.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
princeton-nlp/HELMET community
Papers with Code archive
1 ran · 0 no recorded run
FunctionStatusWhere it lives
LLM Ran
ran on a synthesized input · constructed an object; no result was computed
princeton-nlp/HELMET/model_utils.py
code served (permissive licence) · get_code("49c91caa4809ddb1")
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 an instance of its own class, LLM: the run constructed the object and did not call it, so no result was computed. 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 numpy as np
from typing import Optional, List, Dict, Any

torch.manual_seed(0)

# Define minimal stand-in classes if needed
class MinimalLLM:
    """Minimal concrete implementation of LLM for testing"""
    def __init__(self, model_name: str, temperature: float=0.9, top_p: float=0.9, 
                 max_length: int=32768, generation_max_length: int=2048, 
                 generation_min_length: int=0, do_sample: bool=True, 
                 stop_new_line: bool=False, use_chat_template: bool=False, 
                 system_message: Optional[str]='You are a helpful assistant.'):
        self.model_name = model_name
        self.temperature = temperature
        self.top_p = top_p
        self.max_length = max_length
        self.generation_max_length = generation_max_length
        self.generation_min_length = generation_min_length
        self.do_sample = do_sample
        self.use_chat_template = use_chat_template
        self.system_message = system_message
        self.stops = None
        if stop_new_line:
            self.stops = ['\n', '\n\n']
    
    def prepare_inputs(self, test_item: Dict[str, Any], data: Dict[str, Any]) -> Any:
        return test_item.get('input', 'test input')
    
    def generate(self, inputs: Optional[Any]=None, prompt: Optional[str]=None, **kwargs) -> Optional[Dict[str, Any]]:
        return {
            'output': 'generated text',
            'input_len': 10,
            'output_len': 5,
            'input_text': prompt or str(inputs)
        }
    
    def generate_batch(self, inputs: Optional[List[Any]]=None, prompt: Optional[List[str]]=None, **kwargs) -> List[Optional[Dict[str, Any]]]:
        outputs = []
        if inputs is None:
            for p in prompt:
                outputs.append(self.generate(prompt=p, **kwargs))
        else:
            for i in inputs:
                outputs.append(self.generate(inputs=i, **kwargs))
        return outputs

# Construct ARGS dictionary with all parameters
ARGS = {
    'model_name': 'test-model-v1',
    'temperature': 0.7,
    'top_p': 0.95,
    'max_length': 4096,
    'generation_max_length': 1024,
    'generation_min_length': 10,
    'do_sample': True,
    'stop_new_line': False,
    'use_chat_template': True,
    'system_message': 'You are a helpful assistant that answers questions.'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch, tqdm — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type LLM — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a LLM output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not 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

Also present in other papers' code

Of 1 samples lifted from this paper's linked repositories, 1 carry a function identity and 0 do not. Of the 1 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 1 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

Long-context language models now advertise context windows up to millions of tokens, yet evaluations typically report a single length or a narrow task family, masking two failure modes: performance can collapse as length grows, and strong retrieval need not transfer to downstream use. We present ATLAS, a benchmarking framework that redefines longcontext evaluation as length-dependent capability profiling. ATLAS contributes three methodological principles: (i) a layered taxonomy separating foundational operations from application workloads so failures can be attributed, (ii) length-aware AUC scoring that integrates score-length curves over a fixed 8K-1M grid, replacing single-point metrics with full degradation profiles, and (iii) ATLAScore, a harmonic-mean aggregate over taxonomy categories that penalizes imbalanced profiles, with end-to-end uncertainty propagation from subset scores through the nonlinear final aggregate. We instantiate the framework across eight capability dimensions with nine auditable components and 6,438 instances, and evaluate 26 models. Gemini-3.1-Pro-Preview leads at 128K, Claude-Opus-4.6 leads at 1M. Rankings reshuffle substantially between ATLAScore@8K-128K and ATLAScore@8K-1M: 7 models move by at least two ranks, and the two taxonomy layers share only 61% of cross-model variance, with individual rank gaps up to 12 positions. These results support reporting longcontext quality by capability and length, not by a single headline score.

For agents

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

get_harvested_code_for_paper("2605.28079")
get_code_for_paper("2605.28079")
have("2605.28079")

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

curl https://syntology.ai/api/ran/2605.28079.json

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

[![Syntology run record](https://syntology.ai/api/ran/2605.28079.svg)](https://syntology.ai/paper/2605.28079)

Connect an agent — have() is free.