SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2406.06110 · 2024

Recurrent Context Compression: Efficiently Expanding the Context Window of LLM

arXiv · PDF · Open in the Atlas

Code that ran

3 of 3 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; 3 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/2406.06110.json · badge: /api/ran/2406.06110.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
WUHU-G/RCC_Transformer canonical (classifier label)
classifier basis: census
3 ran · 0 no recorded run
FunctionStatusWhere it lives
generate_sublists Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
WUHU-G/RCC_Transformer/scrip/pretrain_long_mamba.py
code served (permissive licence) · get_code("96bd663092bc659c")
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 list. 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 random

random.seed(0)

# Define arguments for generate_sublists
lst = [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20]
N = 15  # total length of all sublists combined
M = 2   # minimum length of each sublist
X = 6   # maximum length of each sublist

ARGS = {
    "lst": lst,
    "N": N,
    "M": M,
    "X": X
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list — a list 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
load_and_merge_safetensors Ran
ran; our contract draft was wrong, not the code
WUHU-G/RCC_Transformer/chat_example.py
code served (permissive licence) · get_code("293e0b40e5f61121")
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 dict. 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 tempfile
import os
from safetensors.torch import save_file

torch.manual_seed(0)

# Create a temporary directory with sample .safetensors files
temp_dir = tempfile.mkdtemp()

# Create first safetensors file with some tensors
state_dict_1 = {
    'layer1.weight': torch.randn(2, 4),
    'layer1.bias': torch.randn(4),
}
save_file(state_dict_1, os.path.join(temp_dir, 'model_part_1.safetensors'))

# Create second safetensors file with same keys but different values
# Use same shape for keys that will be concatenated
state_dict_2 = {
    'layer1.weight': torch.randn(2, 4),
    'layer1.bias': torch.randn(4),
}
save_file(state_dict_2, os.path.join(temp_dir, 'model_part_2.safetensors'))

# Create third safetensors file with a new key
state_dict_3 = {
    'layer2.weight': torch.randn(3, 4),
}
save_file(state_dict_3, os.path.join(temp_dir, 'model_part_3.safetensors'))

ARGS = {
    'directory_path': temp_dir,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
safetensors, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
make_pretrain_data_module Ran
ran; our contract draft was wrong, not the code
WUHU-G/RCC_Transformer/scrip/pretrain_long_mamba.py
code served (permissive licence) · get_code("d8892cfebae56c76")
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 dict. 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
import transformers
from tempfile import TemporaryDirectory
import os

torch.manual_seed(0)
np.random.seed(0)

# Create a minimal tokenizer stand-in
class StandInTokenizer:
    def __init__(self):
        self.pad_token_id = 0
        self.eos_token_id = 2

# Create temporary memmap files for train and val data
def create_memmap_file():
    """Create a temporary memmap file with dummy token IDs."""
    # Create a temporary file
    tmp_dir = TemporaryDirectory()
    filepath = os.path.join(tmp_dir.name, "data.bin")
    
    # Create memmap with enough data for at least 2 samples
    # sequence_length will be 1024, so we need at least 2048 tokens
    sequence_length = 1024
    num_samples = 3
    total_length = sequence_length * num_samples
    
    # Create the memmap file
    fp = np.memmap(filepath, dtype=np.uint16, mode='w+', shape=(total_length,))
    fp[:] = np.random.randint(0, 1000, total_length, dtype=np.uint16)
    fp.flush()
    
    return filepath, tmp_dir

# Create train and val data files
train_filepath, train_tmp = create_memmap_file()
val_filepath, val_tmp = create_memmap_file()

tokenizer = StandInTokenizer()

ARGS = {
    "tokenizer": tokenizer,
    "train_data_path": train_filepath,
    "val_data_path": val_filepath,
    "cutoff_len": 1024,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch, transformers — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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 3 samples lifted from this paper's linked repositories, 3 carry a function identity and 0 do not. Of the 3 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 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.

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

To extend the context length of Transformer-based large language models (LLMs) and improve comprehension capabilities, we often face limitations due to computational resources and bounded memory storage capacity. This work introduces a method called Recurrent Context Compression (RCC), designed to efficiently expand the context window length of LLMs within constrained storage space. We also investigate the issue of poor model responses when both instructions and context are compressed in downstream tasks, and propose an instruction reconstruction method to mitigate this problem. We validated the effectiveness of our approach on multiple tasks, achieving a compression rate of up to 32x on text reconstruction tasks with a BLEU4 score close to 0.95, and nearly 100\% accuracy on a passkey retrieval task with a sequence length of 1M. Finally, our method demonstrated competitive performance in long-text question-answering tasks compared to non-compressed methods, while significantly saving storage resources in long-text inference tasks. Our code, models, and demo are available at https://github.com/WUHU-G/RCC_Transformer

For agents

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

get_harvested_code_for_paper("2406.06110")
get_code_for_paper("2406.06110")
have("2406.06110")

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

curl https://syntology.ai/api/ran/2406.06110.json

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

[![Syntology run record](https://syntology.ai/api/ran/2406.06110.svg)](https://syntology.ai/paper/2406.06110)

Connect an agent — have() is free.