SAVRN
Search Contact SAVRN

Open-weight model · Fill mask

esm2_t33_650M_UR50D

by AI at Meta facebook/esm2_t33_650M_UR50D

ESM-2 is a state-of-the-art protein model trained on a masked language modelling objective. It is suitable for fine-tuning on a wide range of tasks that take protein sequences as input.

Parameters652M
Context1,026
Weights7.8 GB
Licensemit
AccessOpen weights
Monthly Downloads1.5M

Runs On

What it takes to serve esm2_t33_650M_UR50D (652M parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.

PrecisionWeightsMemory neededCheapest setupPer hourAlso fits
16-bit 1.3 GB 1.6 GB 1x MI300X (192 GB)
Vultr		 $1.85 1x H100 $1.99 · 1x MI325X $2.00
8-bit 0.7 GB 0.8 GB 1x MI300X (192 GB)
Vultr		 $1.85 1x H100 $1.99 · 1x MI325X $2.00
4-bit 0.3 GB 0.4 GB 1x MI300X (192 GB)
Vultr		 $1.85 1x H100 $1.99 · 1x MI325X $2.00

Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Sep 18, 2026.

SAVRN's Notes on esm2_t33_650M_UR50D

AI at Meta built this checkpoint for teams that fine-tune on protein sequences, and the numbers say it will never fill an accelerator. Serving one copy takes 1.6 GB at 16-bit, 0.8 GB at 8-bit, or 0.4 GB at 4-bit. The cheapest Index listing at every precision is one MI300X with 192 GB at $1.85 per hour, so a single instance would use less than one percent of the card. Stack fine-tuning jobs or parallel copies on that one device rather than dedicating it.

The MIT license permits commercial use, modification and redistribution provided the copyright and permission notices ship with the files, so a fine-tuned variant is clear for a commercial product. Two checks before committing: the window is 1,026 tokens per sequence, and the download is 7.8 GB across nine files, stored in float32 in safetensors, pytorch and tf formats, so pull only the format your stack loads.

SAVRN Research, 2026-09-18

Model Card

By AI at Meta, published under mit, revision 08e4846e5371.

ESM-2 is a state-of-the-art protein model trained on a masked language modelling objective. It is suitable for fine-tuning on a wide range of tasks that take protein sequences as input. For detailed information on the model architecture and training data, please refer to the accompanying paper. You may also be interested in some demo notebooks (PyTorch, TensorFlow) which demonstrate how to fine-tune ESM-2 models on your tasks of interest. Several ESM-2 checkpoints are available in the Hub with varying sizes. Larger sizes generally have somewhat better accuracy, but require much more memory and time to train

Read AI at Meta's full model card

ESM-2

ESM-2 is a state-of-the-art protein model trained on a masked language modelling objective. It is suitable for fine-tuning on a wide range of tasks that take protein sequences as input. For detailed information on the model architecture and training data, please refer to the accompanying paper. You may also be interested in some demo notebooks (PyTorch, TensorFlow) which demonstrate how to fine-tune ESM-2 models on your tasks of interest.

Several ESM-2 checkpoints are available in the Hub with varying sizes. Larger sizes generally have somewhat better accuracy, but require much more memory and time to train:

Checkpoint name Num layers Num parameters
esm2_t48_15B_UR50D 48 15B
esm2_t36_3B_UR50D 36 3B
esm2_t33_650M_UR50D 33 650M
esm2_t30_150M_UR50D 30 150M
esm2_t12_35M_UR50D 12 35M
esm2_t6_8M_UR50D 6 8M

Configuration

Architecture
EsmForMaskedLM
Context length (tokens)
1,026
Layers
33
Hidden size
1,280
Feed-forward size
5,120
Attention heads
20
Vocabulary size
33
Stored precision
float32
Model type
esm

Identity and Version

Repository
facebook/esm2_t33_650M_UR50D
Publisher
AI at Meta
Task
Fill mask
Modality
Text
Library
transformers
Parameters
652M parameters
Languages
tf, esm
Revision
08e4846e537177426273712802403f7ba8261b6c
First published
2022-09-27
Last updated
2023-03-21

Files and Weights

9 files, 7.8 GB in total. The weights are 3 files totalling 7.8 GB in bin, h5, safetensors.

Weights3 files · 7.8 GB
Configuration2 files · 849 B
Tokenizer2 files · 188 B
Documentation1 file · 1.7 KB
Repository1 file · 1.4 KB
Every file
FileTypeSizeSHA-256
model.safetensorsWeights2.6 GB a08adabb949f
pytorch_model.binWeights2.6 GB c874668852c7
tf_model.h5Weights2.6 GB 3110b0ee07a4
config.jsonConfiguration724 B —
special_tokens_map.jsonConfiguration125 B —
README.mdDocumentation1.7 KB —
.gitattributesRepository1.4 KB —
tokenizer_config.jsonTokenizer95 B —
vocab.txtTokenizer93 B —

License and Download

License
mit
Access
Open weights, no gate
Download size
7.8 GB
Download from AI at Meta

Released by AI at Meta through its official repository on Hugging Face. Read the license.

Memory Requirements

PrecisionWeights in memory
As published7.8 GB
16-bit1.3 GB
8-bit0.7 GB
4-bit0.3 GB

Weights only, from the published parameter count; the key-value cache and runtime add to this.

Compare esm2_t33_650M_UR50D

Questions About esm2_t33_650M_UR50D

How much GPU memory does esm2_t33_650M_UR50D need?

About 1.6 GB at 16-bit and 0.4 GB at 4-bit: the weights (652M parameters) plus a working margin. A long context needs more.

What is the cheapest GPU to run esm2_t33_650M_UR50D on?

At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.

Can I use esm2_t33_650M_UR50D commercially?

Yes. esm2_t33_650M_UR50D is released under MIT License. The MIT License is a short permissive license. It permits commercial use, modification and redistribution, provided the copyright notice and permission notice are included.

What is esm2_t33_650M_UR50D's context length?

1,026 tokens, from the maximum position embeddings in its published configuration.

Similar Models

XLM-RoBERTa model pre-trained on 2.5TB of filtered CommonCrawl data containing 100 languages. It was introduced in the paper Unsupervised Cross-lingual Representation Learning at Scale by Conneau et al. and first released in this repository. Disclaimer: The team releasing XLM-RoBERTa did not write a model card for this model so this model card has been written by the Hugging Face team. XLM-RoBERTa is a multilingual version of RoBERTa. It is pre-trained on 2.5TB of filtered CommonCrawl data containing 100 languages. RoBERTa is a transformers model pretrained on a large corpus in a self-supervised fashion. This means it was pretrained on the raw texts only, with no humans labelling them in…

Open weights mit 561M parameters 514 tokens transformers
View model

Model · Fill mask

ModernBERT-large

Answer.AI

ModernBERT is a modernized bidirectional encoder-only Transformer model (BERT-style) pre-trained on 2 trillion tokens of English and code data with a native context length of up to 8,192 tokens. ModernBERT leverages recent architectural improvements such as: - Rotary Positional Embeddings (RoPE) for long-context support. - Local-Global Alternating Attention for efficiency on long inputs. - Unpadding and Flash Attention for efficient inference. ModernBERT’s native long context length makes it ideal for tasks that require processing long documents, such as retrieval, classification, and semantic search within large corpora. The model was trained on a large corpus of text and code, making it…

Open weights apache-2.0 396M parameters 8,192 tokens transformers
View model

Pretrained model on English language using a masked language modeling (MLM) objective. It was introduced in this paper and first released in this repository. This model is case-sensitive: it makes a difference between english and English. Disclaimer: The team releasing RoBERTa did not write a model card for this model so this model card has been written by the Hugging Face team. RoBERTa is a transformers model pretrained on a large corpus of English data in a self-supervised fashion. This means it was pretrained on the raw texts only, with no humans labelling them in any way (which is why it can use lots of publicly available data) with an automatic process to generate inputs and labels…

Open weights mit 355M parameters 514 tokens transformers
View model

Model · Fill mask

bert-large-uncased

BERT community

Pretrained model on English language using a masked language modeling (MLM) objective. It was introduced in this paper and first released in this repository. This model is uncased: it does not make a difference between english and English. Disclaimer: The team releasing BERT did not write a model card for this model so this model card has been written by the Hugging Face team. BERT is a transformers model pretrained on a large corpus of English data in a self-supervised fashion. This means it was pretrained on the raw texts only, with no humans labelling them in any way (which is why it can use lots of publicly available data) with an automatic process to generate inputs and labels from…

Open weights apache-2.0 336M parameters 512 tokens transformers
View model

This is a Japanese DeBERTa V2 large model pre-trained on Japanese Wikipedia, the Japanese portion of CC-100, and the Japanese portion of OSCAR. This model is trained with character-level tokenization and whole word masking. You can use this model for masked language modeling as follows: You can also fine-tune this model on downstream tasks. There is no need to tokenize texts in advance, and you can give raw texts to the tokenizer. The texts are tokenized into character-level tokens by sentencepiece. We used the following corpora for pre-training: - Japanese Wikipedia (as of 20221020, 3.2GB, 27M sentences, 1.3M documents) - Japanese portion of CC-100 (85GB, 619M sentences, 66M documents)…

Open weights cc-by-sa-4.0 330M parameters 512 tokens transformers
View model

XLM-RoBERTa model pre-trained on 2.5TB of filtered CommonCrawl data containing 100 languages. It was introduced in the paper Unsupervised Cross-lingual Representation Learning at Scale by Conneau et al. and first released in this repository. Disclaimer: The team releasing XLM-RoBERTa did not write a model card for this model so this model card has been written by the Hugging Face team. XLM-RoBERTa is a multilingual version of RoBERTa. It is pre-trained on 2.5TB of filtered CommonCrawl data containing 100 languages. RoBERTa is a transformers model pretrained on a large corpus in a self-supervised fashion. This means it was pretrained on the raw texts only, with no humans labelling them in…

Open weights mit 279M parameters 514 tokens transformers
View model