PocketGen

Closed weights University of Science and Technology of China (USTC),Hefei Comprehensive National Science Center,Harvard University,Broad Institute,Harvard Data Science Initiative 7.9M parameters 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
University of Science and Technology of China (USTC),Hefei Comprehensive National Science Center,Harvard University,Broad Institute,Harvard Data Science Initiative
Organisation type
Academia,Research collective,Academia,Research collective,Academia
Country
China, United States of America
Published
23 September 2024
Authors
Zaixi Zhang, Wan Xiang Shen, Qi Liu, Marinka Zitnik

What it does

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

Domain
Biology
Task
Protein-ligand contact prediction

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
7.9M

"As a result, PocketGen requires significantly fewer trainable parameters than RFDiffusionAA [16] (7.9M versus 82.9M trainable parameters)."

Training data
tokens

"Finally, we have 40k protein-ligand pairs for training, 100 pairs for validation, and 100 pairs for testing." 40000*300 (assumed length) = 12000000

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
2.1 × 10¹⁹ FLOP

"It takes around 48 hours to finish training on 1 Tesla A100 GPU from scratch." Assume 40% utilization, FP16 tensor precision.

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 A100
Chips used
1
Wall-clock time
48 hours
Power draw
433 W

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
15

Sources

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

Reference
Efficient Generation of Protein Pockets with PocketGen
Last updated
28 November 2025

What the numbers mean

What this model is

PocketGen was published by University of Science and Technology of China (USTC),Hefei Comprehensive National Science Center,Harvard University,Broad Institute,Harvard Data Science Initiative, in the country recorded as China, during September 2024. The publishing organisation is categorised as academia,Research collective,Academia,Research collective,Academia.

It works in the domain of Biology, and is recorded as performing the task of protein-ligand contact prediction.

Its weights were never published, so it can only be reached through its provider. No graphics card changes that.

How it was trained

Producing it required arithmetic totalling around 2.1 × 10¹⁹ FLOP, on hardware recorded as NVIDIA A100. That figure measures what producing the model cost, and has no bearing on how fast it answers.

Answers

PocketGen — common questions

01

PocketGen— is it open source?

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

02

PocketGen— how many parameters does it have?

It has a parameter count of 7.9M. "As a result, PocketGen requires significantly fewer trainable parameters than RFDiffusionAA [16] (7.9M versus 82.9M trainable parameters).". 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.

03

PocketGen— who created it?

It was published by University of Science and Technology of China (USTC),Hefei Comprehensive National Science Center,Harvard University,Broad Institute,Harvard Data Science Initiative, based in China, an organisation categorised as academia,Research collective,Academia,Research collective,Academia.

04

PocketGen— when was it released?

It 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.

05

PocketGen— what is it used for?

It works in the domain of Biology, and is recorded as handling the task of protein-ligand contact prediction. These are the areas it was designed around; they describe intent rather than a hard boundary.

06

PocketGen— how much compute was used to train it?

Training consumed around 2.1 × 10¹⁹ FLOP, on hardware recorded as 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.

07

PocketGen— what GPU do I need to run it?

None. This 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.

Source

Original publication

Record last updated 28 November 2025

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.