RNAdiffusion

Closed weights Princeton University,Tsinghua University,Stanford University September 2024

No estimate

No hardware requirements for this model

The weights for this model have not been published, so it cannot be downloaded or run on your own hardware at any size. It is reachable only through its provider, and no graphics card changes that.

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
Princeton University,Tsinghua University,Stanford University
Organisation type
Academia,Academia,Academia
Country
United States of America, China
Published
15 September 2024
Authors
Kaixuan Huang, Yukang Yang, Kaidi Fu, Yanyi Chu, Le Cong, Mengdi Wang

What it does

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

Domain
Biology
Task
RNA sequence generation
Base model
RNA-FM

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.

Training data
tokens

Diffusion part: 1.1M sequences. 205k UTR sequences Reward model: 101k sequences Total: 1100000+205000+101000=1406000 sequences Assuming average token length of 300: 1406000*300=421800000

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
2.5 × 10¹⁹ FLOP

Autoencoder: 8.5h per epoch on single H100, 2 epochs total Diffusion model: 3.5h per epoch on single H100, 3 epochs total Reward model: 7h on a single A100 Total training time: 2*8.5+3*3.5+7=34.5 Compute: ((2*8.5+3*3.5)*756000000000000+7*312000000000000)*60*60*0.3=2.481192e+19

How it was established
Hardware

The training run

What it physically took to train: which chips, how many, for how long, and what that drew from the wall.

Training hardware
NVIDIA H100 PCIe,NVIDIA A100
Chips used
1
Wall-clock time
35 hours

How it is classified

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

Record confidence
Likely
Citations
9

Sources

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

Reference
Latent Diffusion Models for Controllable RNA Sequence Generation
Last updated
25 May 2026

What the numbers mean

What this model is

RNAdiffusion was published by Princeton University,Tsinghua University,Stanford University, in United States of America, in September 2024. It comes out of academia,Academia,Academia.

It works in Biology, and is recorded as doing rNA sequence generation.

It is derived from RNA-FM rather than trained from scratch, which is the usual way a specialised model is produced.

Because the weights are not available, none of the hardware figures elsewhere on this site apply to it.

Training and provenance

Training it took roughly 2.5 × 10¹⁹ FLOP of computation, on NVIDIA H100 PCIe,NVIDIA A100 — a measure of what producing the model cost, not of how fast it answers.

Answers

RNAdiffusion — common questions

01

What is RNAdiffusion used for?

RNAdiffusion works in Biology, and is recorded as handling rNA sequence generation. These are the areas it was designed around; they describe intent rather than a hard boundary.

02

How much compute was used to train RNAdiffusion?

Around 2.5 × 10¹⁹ FLOP, on NVIDIA H100 PCIe,NVIDIA A100. 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.

03

What GPU do I need to run RNAdiffusion?

None. RNAdiffusion is a closed model — its weights were never published, so it cannot be downloaded or run on your own hardware at any price. It is reachable only through its provider.

04

Is RNAdiffusion open source?

The licensing for RNAdiffusion was never recorded in our source data. We treat unstated licensing as closed, because an unrecorded licence is not one to rely on.

05

How many parameters does RNAdiffusion have?

No parameter count has been published for RNAdiffusion, which is why no memory or speed figure appears on this page.

06

Who created RNAdiffusion?

RNAdiffusion was published by Princeton University,Tsinghua University,Stanford University, based in United States of America, categorised as academia,Academia,Academia.

07

When was RNAdiffusion released?

RNAdiffusion was published in September 2024. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.

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.