v3.1.2

In this release, we made two changes:

1. Brought Flash Attention and Triton Rotary embedding to the ESM C model
2. Updated forge client to enable the return of ESM C's binary serialized hidden states. (note this the return of ESM3 hidden states are not supported).

v3.1.1

In this release, we made three changes:

1. Updated ESM C tokenizer to no longer depend on ESM3. So now you can use ESM C without having to accept the ESM3 license.
2. Updated the transformer class so it returns hidden states.
3. Pined transformers version to be <4.47.0 to prevent a tokenizer issue. If you are currently seeing tokenizer issues, try reinstalling the new release.

v3.1.0

Releasing the ESM C series of models!

v3.0.8

v3.0.8 (#146)

Co-authored-by: Zeming Lin <[email protected]>

v3.0.7

• Adds model API support for ProteinComplex and associated utilities

v3.0.4

• Fix issue where .logits wasn't working properly through Forge
• Fix issue where invalid IDs were being sampled
• Added prediction and inverse fold UI ipynbs

v3.0.3

• Decoding fixes.
• SageMaker client.

v3.0.2

• logits() endpoint.
• topk parameter validation.
• minor example fixes

v3.0.1

ESM v3.0.1 has been released! We have support for a Forge client which allows for API users to use our large models. Be on the look out for your API token.

• Support for Forge API
• Support for batched generation via the batch_generate API call
• gfp_design.ipynb notebook to demonstrate the GFP design results in our paper
• esmprotein.ipynb notebook which goes into detail how to interact with the ESMProtein class.
• ... and many other minor usability improvements and bug fixes.

v3.0.0.post2

What's Changed

• bugfix: plddt now saved to PDB in the b_factor column
• bugfix: residue annotations won't error out if model detects no annotations
• bugfix: mean_embedding checking the wrong configuration argument
• bugfix: iterative sampling of structures was not working as intended.

Full Changelog: https://github.com/evolutionaryscale/esm/commits/v3.0.0post2