Decoupled weight decay regularization TPS calculator

Open weights University of Freiburg 36.5M parameters January 2019

Each card below is assessed against this model at the context length and minimum quality you choose. Speed is an estimate for a single request, calculated from the card's memory bandwidth and the size of the model once compressed.

Calculated for this model

818 cards that can run it

818 cards we hold specifications for

Smallest card that fits

Tesla C1080

4 GB · Q8_0 · 1,010 tok/s

Fastest card

B200

92,828 tok/s · 180 GB

Which GPUs can run Decoupled weight decay regularization?

Set the inputs, read the answer

Context length tokens

A longer conversation needs more memory, which can push this model off smaller cards.

Minimum quality

Hides cards that would only fit the model by compressing it below this point.

Search cards

Card memory

Also show cards that cannot run it

818 cards match

Calculating

					Needs	Quantisation	Fit
92,828 tok/s 55,697–148,525 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	0.7 GB	Q8_0	Comfortable
92,828 tok/s 55,697–148,525 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	0.7 GB	Q8_0	Comfortable
74,126 tok/s 44,475–118,601 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	0.7 GB	Q8_0	Comfortable
74,126 tok/s 44,475–118,601 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	0.7 GB	Q8_0	Comfortable
59,283 tok/s 35,570–94,852 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	0.7 GB	Q8_0	Comfortable
56,741 tok/s 34,045–90,786 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	0.7 GB	Q8_0	Comfortable
56,741 tok/s 34,045–90,786 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	0.7 GB	Q8_0	Comfortable
54,305 tok/s 32,583–86,887 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	0.7 GB	Q8_0	Comfortable
48,195 tok/s 28,917–77,113 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	0.7 GB	Q8_0	Comfortable
48,195 tok/s 28,917–77,113 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	0.7 GB	Q8_0	Comfortable
48,195 tok/s 28,917–77,113 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	0.7 GB	Q8_0	Comfortable
45,718 tok/s 27,431–73,149 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
38,988 tok/s 23,393–62,381 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
38,988 tok/s 23,393–62,381 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	0.7 GB	Q8_0	Comfortable
38,988 tok/s 23,393–62,381 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
38,988 tok/s 23,393–62,381 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
38,988 tok/s 23,393–62,381 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
29,687 tok/s 17,812–47,498 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	0.7 GB	Q8_0	Comfortable
29,687 tok/s 17,812–47,498 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	0.7 GB	Q8_0	Comfortable
24,739 tok/s 14,843–39,582 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	0.7 GB	Q8_0	Comfortable
24,211 tok/s 14,526–38,737 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	0.7 GB	Q8_0	Comfortable
23,671 tok/s 14,203–37,874 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	0.7 GB	Q8_0	Comfortable
23,671 tok/s 14,203–37,874 · low confidence	A100X NVIDIA	80 GB	2,040 GB/s	Jun 2021	0.7 GB	Q8_0	Comfortable
23,671 tok/s 14,203–37,874 · low confidence	A800 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Aug 2022	0.7 GB	Q8_0	Comfortable
23,671 tok/s 14,203–37,874 · low confidence	H100 CNX NVIDIA	80 GB	2,040 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable

Speeds are estimates for a single request — one conversation at a time — calculated from memory bandwidth, model size and quantisation. Real throughput varies with the inference runtime and its version. Figures published by hardware vendors measure many simultaneous requests and are much higher.

On record

Full specification

Everything on record for this model. Most of it describes how it was trained rather than how it runs — useful context for judging how much work went into it, and how it compares with models built at a different scale.

Origin

Who built this model, where, and when it was published.

Organisation: University of Freiburg
Organisation type: Academia
Country: Germany
Published: 4 January 2019
Authors: Ilya Loshchilov and Frank Hutter

What it does

The problem areas the model was built for. A model can carry several of each.

Domain: Vision
Task: Image classification

Size

How large the model is and how much data it was trained on. Parameters are the figure that decides whether it fits on a given graphics card.

Parameters: 36.5M
Training data: 50,000 tokens

Training compute

The arithmetic performed to train the model, measured in floating-point operations. It is a measure of what the training run cost, not of how fast the finished model answers you.

Training compute: 4.7 × 10¹⁷ FLOP
How it was established: Operation counting

Availability

Whether you can obtain the model and run it on your own hardware, which is what decides if any of the graphics-card figures on this page apply.

Weights: Open — downloadable
Model access: Open weights (unrestricted)
Training code: Open source

How it is classified

Labels the source dataset applies when tracking notable models, and how confident it is in the entry.

Record confidence: Confident
Citations: 2,658

Sources

Where this record came from and when it was last checked.

Reference: Decoupled weight decay regularization.
Last updated: 25 May 2026

The extremes

The ten fastest GPUs that run Decoupled weight decay regularization

Ranked by estimated tokens per second, newest card first where speeds tie. Because generation is bound by memory bandwidth, this ordering follows bandwidth rather than any gaming benchmark.

The smallest GPUs that still run Decoupled weight decay regularization

The cheapest route in, by memory capacity. A tight fit runs the model but leaves nothing spare for a longer conversation.

What the numbers mean

The hardware side

Minimum card

Tesla C1080

Memory needed

0.7 GB

Fastest

92,828 tok/s

Decoupled weight decay regularization reaches a parameter count of 36.5M. That is small enough that hardware is rarely the obstacle, including on cards several years old. The number of cards we track that can run it: 818.

At the low end it is handled by Tesla C1080, with a memory capacity of 4 GB, running it at a compression of Q8_0 and producing around 1,010 tokens per second.

At the other end sits B200, generating roughly 92,828 tokens per second on the strength of a memory bandwidth of 8,000 GB/s.

What this model is

Decoupled weight decay regularization was published by University of Freiburg, in the country recorded as Germany, during January 2019. The category the publisher falls under is academia.

It works in the domain of Vision, and is recorded as performing the task of image classification.

The weights being open is what puts this page in the calculator rather than only in the catalogue: it is a model you can actually hold.

What decides the speed

Across every card that can run it, the middle of the range sits at 2,606.6 tokens per second. Exceeding reading speed outright: 818 of them.

Every weight participates in every token here, so bandwidth is the whole story: the ranking below is effectively a ranking of memory throughput.

Memory here is estimated from size rather than computed from the architecture, which is not recorded for this model — the numbers are indicative rather than exact.

How it was trained

The training run consumed about 4.7 × 10¹⁷ FLOP. That figure measures what producing the model cost, and has no bearing on how fast it answers.

It was trained on a corpus of about 50,000 tokens of text.

Step by step

How to choose a GPU for Decoupled weight decay regularization

The table above has already assessed every card we hold specifications for against this model. Getting to your answer takes six steps.

01

Start from the memory column

Every card here has been checked against Decoupled weight decay regularization, needing around 0.7 GB at a compression of Q8_0. No amount of processing power compensates for a card that cannot hold it.
02

Set the context length you will work at

The conversation occupies memory too, and grows as it goes. Set the slider to the length you expect, because at long context a card that handles short questions easily can be dropped by Decoupled weight decay regularization.
03

Choose how far you will compress it

Compression is what makes a model fit smaller cards, at some cost in accuracy, reaching a compression of Q8_0 on the smallest card that fits. Setting a minimum quality drops the cards that only manage it by squeezing further than you would want, and holds the comparison at one level.
04

Sort by speed

Ranking by tokens per second follows memory bandwidth rather than core counts, for Decoupled weight decay regularization. It will not match a gaming ordering, because generation is bound by memory bandwidth. The card topping the list is B200, at 92,828 tok/s.
05

Read the fit column last

The fit column separates cards that just manage it from those with room to spare, in the case of Decoupled weight decay regularization. Comfortable means you can grow the context later. That difference matters more than a few tokens per second, so buy for comfortable if you expect to.
06

Check the card from the other side

Following a card through to its own page shows every other model it can hold, which is the question that follows once you have settled on Decoupled weight decay regularization.

Answers

Decoupled weight decay regularization — common questions

Decoupled weight decay regularization— can I run it on a GPU holding 24 GB?

Yes. The card GeForce RTX 5090 D V2, holding 24 GB, runs it at a compression of Q8_0, using about 0.7 GB and generating roughly 15,549 tokens per second. The fit is comfortable.

Decoupled weight decay regularization— is it open source?

Its weights are published, so it can be downloaded and run on your own hardware. Note that open weights is not the same as open source in the full sense — it says nothing about the training data, the training code, or the commercial terms attached.

Decoupled weight decay regularization— how many parameters does it have?

It has a parameter count of 36.5M. From author communication WideResNet 28-10 models with 36.5 million parameters (3.65E+07). That figure is the total, and it is what decides how much memory the model needs — roughly half a gigabyte per billion at the compression most people use.

Decoupled weight decay regularization— who created it?

It was published by University of Freiburg, based in Germany, an organisation categorised as academia.

Decoupled weight decay regularization— when was it released?

It was published in January 2019. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.

Decoupled weight decay regularization— what is it used for?

It works in the domain of Vision, and is recorded as handling the task of image classification. Models frequently carry more than one of each, and the tags describe purpose rather than capability limits.

Decoupled weight decay regularization— where can I download it?

The weights are published, though we do not hold a repository link for it. This site calculates hardware requirements rather than hosting model files.

Decoupled weight decay regularization— how much compute was used to train it?

Training consumed around 4.7 × 10¹⁷ FLOP. That measures what producing the model cost and says nothing about how quickly it answers once trained — inference speed comes from memory bandwidth, not from the training budget.

Decoupled weight decay regularization— can I run it if it does not fit in my GPU?

It can be split between the card and system memory, but it generates painfully slowly that way. Every figure here assumes the whole model is resident on the card.

Decoupled weight decay regularization— would two GPUs run it faster?

Two cards buy memory rather than speed, which matters only if one card cannot hold it. The number that can: 818. So a second card is rarely the answer here.

Decoupled weight decay regularization— why does the quantisation differ between cards?

Each card is shown running the least-compressed copy it can hold. The number of distinct compression levels across the cards that fit it: 1. Bigger cards get the more accurate version, and the quality floor above pins the comparison to one level.

Decoupled weight decay regularization— how accurate are these speed estimates?

Every figure is derived from memory bandwidth and model size, not benchmarked, which is why each is published as a range rather than a single number. One example: 55,697–148,525 tok/s on B200. The same model and card vary by thirty to fifty per cent depending on the inference software and its version.

Decoupled weight decay regularization— what GPU do I need to run it?

The smallest card in our catalogue that holds it is Tesla C1080, with a memory capacity of 4 GB. It runs the model at a compression of Q8_0 using about 0.7 GB, and produces roughly 1,010 tokens per second. The number of cards able to run it in total: 818.

Decoupled weight decay regularization— how fast is it on a GPU?

It depends on the card. The quickest we calculate is B200, at about 92,828 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and the number of cards clearing that: 818.

Decoupled weight decay regularization— how much VRAM does it need?

It needs about 0.7 GB at a compression of Q8_0, which is what the smallest card that runs it uses. Less compression needs more: the figures in the memory column above are recalculated for each card, because each one holds the least-compressed version it can.

Decoupled weight decay regularization— can I run it on a GPU holding 8 GB?

Yes. The card CMP 170HX 8 GB, holding 8 GB, runs it at a compression of Q8_0, using about 0.7 GB and generating roughly 17,289 tokens per second. The fit is comfortable.

Decoupled weight decay regularization— can I run it on a GPU holding 12 GB?

Yes. The card GeForce RTX 3080 Ti, holding 12 GB, runs it at a compression of Q8_0, using about 0.7 GB and generating roughly 10,587 tokens per second. The fit is comfortable.

Decoupled weight decay regularization— can I run it on a GPU holding 16 GB?

Yes. The card Tesla V100 SXM2 16 GB, holding 16 GB, runs it at a compression of Q8_0, using about 0.7 GB and generating roughly 13,112 tokens per second. The fit is comfortable.

Source

Original publication

Record last updated 25 May 2026

The other direction

Looking at it from the other side?

This page starts from the model. If you already own a card and want to know everything it will run, start from the hardware instead.

All AI models