FoldFlow2
No estimate
No hardware requirements for this model
This model's weights are open, but no parameter count has been published for it. Every memory and speed figure starts from that number, so we would rather show nothing than a fabricated estimate.
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
- Dreamfold,University of Montreal / Université de Montréal,McGill University,University of Oxford
- Organisation type
- Industry,Academia,Academia,Academia
- Country
- Canada, United Kingdom of Great Britain and Northern Ireland
- Published
- 30 May 2024
- Authors
- Guillaume Huguet, James Vuckovic, Kilian Fatras, Eric Thibodeau-Laufer, Pablo Lemos, Riashat Islam, Cheng-Hao Liu, Jarrid Rector-Brooks, Tara Akhound-Sadegh, Michael Bronstein, Alexander Tong, Avishek Joey Bose
What it does
The problem areas the model was built for. A model can carry several of each.
- Domain
- Biology
- Task
- Protein generation, Protein design
- Base model
- ESM2-650M
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
- 48,000,000 tokens
Average protein length = (60 + 384) / 2 = 222 residues Total datapoints = 160,000 structures × 222 residues = 35,520,000
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
- 7.6 × 10²¹ FLOP
- How it was established
- Hardware
1. Hardware setup: 2x NVIDIA A100 40GB GPUs (3.12 x 10^14 FLOP/s per GPU) 2. Training duration: 4 days (directly provided) = 345,600 seconds 3. Utilization rate: 40% 4. Calculation: 2 GPUs × 3.12×10^14 FLOP/s × 345,600s × 0.4 = 8.6×10^19 FLOPs Base model: 7.560000000001e+21 Total: 7646000000001000000000
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
- 2
- Wall-clock time
- 96 hours
- Power draw
- 1.6 kW
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 (non-commercial)
- Training code
- Open (non-commercial)
CC NC 4.0 https://github.com/DreamFold/FoldFlow weights https://github.com/DreamFold/FoldFlow/releases/tag/0.2.0
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
- 49
Sources
Where this record came from and when it was last checked.
- Reference
- Sequence-Augmented SE(3)-Flow Matching For Conditional Protein Backbone Generation
- Last updated
- 25 May 2026
What the numbers mean
What this model is
FoldFlow2 was published by Dreamfold,University of Montreal / Université de Montréal,McGill University,University of Oxford, in Canada, in May 2024. industry,Academia,Academia,Academia is the category the publisher falls under.
It works in Biology, and is recorded as doing protein generation, Protein design.
Its starting point was ESM2-650M — most models at this scale are adapted from an existing base rather than built from nothing.
Because its weights were released, nothing about running it depends on a provider staying available — it is yours once downloaded.
Training and provenance
Producing it required around 7.6 × 10²¹ FLOP of arithmetic, on NVIDIA A100, which is a statement about the training budget rather than about inference.
It was trained on about 48,000,000 tokens of text.
Answers
FoldFlow2 — common questions
What is FoldFlow2 used for?
FoldFlow2 works in Biology, and is recorded as handling protein generation, Protein design. These are the areas it was designed around; they describe intent rather than a hard boundary.
Where can I download FoldFlow2?
The weights for FoldFlow2 are published, though we do not hold a repository link for it. This site calculates hardware requirements rather than hosting model files.
How much compute was used to train FoldFlow2?
Around 7.6 × 10²¹ FLOP, on 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.
What GPU do I need to run FoldFlow2?
We cannot say. FoldFlow2 has open weights, but no parameter count has been published for it, and every memory and speed calculation starts from that number. We would rather show nothing than a fabricated estimate.
Is FoldFlow2 open source?
Its weights are published, so FoldFlow2 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.
How many parameters does FoldFlow2 have?
No parameter count has been published for FoldFlow2, which is why no memory or speed figure appears on this page.
Who created FoldFlow2?
FoldFlow2 was published by Dreamfold,University of Montreal / Université de Montréal,McGill University,University of Oxford, based in Canada, categorised as industry,Academia,Academia,Academia.
When was FoldFlow2 released?
FoldFlow2 was published in May 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.
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.