SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2602.02151 · 2026

Revisiting Adaptive Rounding with Vectorized Reparameterization for LLM Quantization

Yawei Li, Luca Benini, Guolei Sun, Yuli Zhou, Qingxuan Chen

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 (1 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 2 have no contract verdict; 1 failure 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/2602.02151.json · badge: /api/ran/2602.02151.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
zhoustan/VQRound community
Papers with Code archive
3 ran · 0 no recorded run
FunctionStatusWhere it lives
RectifiedSigmoid Ran
ran on a synthesized input · constructed an object; no result was computed
zhoustan/VQRound/vqround.py
code served (permissive licence) · get_code("9d3ab09c4512fc2e")
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, RectifiedSigmoid: 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 torch.nn as nn

torch.manual_seed(0)

class RectifiedSigmoid(nn.Module):

    def __init__(self, gamma=-0.1, zeta=1.1):
        super().__init__()
        self.gamma = gamma
        self.zeta = zeta

    def forward(self, x):
        return torch.clamp(torch.sigmoid(x) * (self.zeta - self.gamma) + self.gamma, 0, 1)

    def inverse(self, y):
        return -torch.log((self.zeta - self.gamma) / (y - self.gamma) - 1)


ARGS = {
    "gamma": -0.1,
    "zeta": 1.1
}
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 RectifiedSigmoid — 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 RectifiedSigmoid 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
VQRoundLinear Ran
ran on a synthesized input
zhoustan/VQRound/vqround.py
code served (permissive licence) · get_code("37104c4787d25cc0")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, ndim 2, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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 numpy as np

# Create a small base linear layer
base_linear = nn.Linear(in_features=6, out_features=8, bias=True)

# Initialize with small random weights
with torch.no_grad():
    base_linear.weight.data = torch.randn(8, 6) * 0.5
    base_linear.bias.data = torch.randn(8) * 0.1

# Parameters for VQRoundLinear
n_bits = 4
D = 4  # dimension of each block for VQ
K = 8  # number of codebook entries (small for testing)
kmeans_iters = 5  # fewer iterations for speed
gamma = -0.1
zeta = 1.1

INIT_ARGS = {
    'base_linear': base_linear,
    'n_bits': n_bits,
    'D': D,
    'K': K,
    'kmeans_iters': kmeans_iters,
    'gamma': gamma,
    'zeta': zeta,
    'external_scale': None,
    'external_zero': None,
    'external_base_int': None,
    'external_rest': None,
}

# Input tensor for forward pass
# batch_size=2, in_features=6 (must match base_linear.in_features)
x_input = torch.randn(2, 6)

ARGS = {
    'x': x_input,
}
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
faiss, numpy, 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 — no output shape is stated: observe_signature.py observed this class with its constructor sizes canonicalised where it could, which the node does not record, so its shape is not stated as this setup block's
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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
per_row_affine_params Ran
ran; our contract draft was wrong, not the code
zhoustan/VQRound/vqround.py
code served (permissive licence) · get_code("ee70e3e31f57a1af")
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 tuple, ndim 3, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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

torch.manual_seed(0)

# Create a small weight tensor with shape (out_features, in_features)
# Using distinct dimensions: 2 output features, 4 input features
w = torch.randn(2, 4)

# n_bits parameter - typical quantization bit width
n_bits = 8

ARGS = {
    "w": w,
    "n_bits": n_bits,
}
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 tuple, ndim 3, dtype family float — a tuple output has no array shape of its own; none is stated
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

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

Adaptive Rounding has emerged as an alternative to round-to-nearest (RTN) for posttraining quantization by enabling cross-element error cancellation. Yet, dense and elementwise rounding matrices are prohibitively expensive for billion-parameter large language models (LLMs). We revisit adaptive rounding from an efficiency perspective and propose VQRound, a parameter-efficient optimization framework that reparameterizes the rounding matrix into a compact codebook. Unlike low-rank alternatives, VQRound minimizes the element-wise worstcase error under L ∞ norm, which is critical for handling heavy-tailed weight distributions in LLMs. Beyond reparameterization, we identify rounding initialization as a decisive factor and develop a lightweight end-to-end finetuning pipeline that optimizes codebooks across all layers using only 128 samples. Extensive experiments on OPT, LLaMA, LLaMA2, and Qwen3 models demonstrate that VQRound achieves better convergence than traditional adaptive rounding at the same number of steps while using as little as 0.2% of the trainable parameters. Our results show that adaptive rounding can be made both scalable and fast-fitting. The code is available at https://github.com/zhoustan/VQRound.

For agents

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

get_harvested_code_for_paper("2602.02151")
get_code_for_paper("2602.02151")
have("2602.02151")

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

curl https://syntology.ai/api/ran/2602.02151.json

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

[![Syntology run record](https://syntology.ai/api/ran/2602.02151.svg)](https://syntology.ai/paper/2602.02151)

Connect an agent — have() is free.