ESM2quinox
An implementation of ESM2 in Equinox+JAX
Installation
pip install esm2quinox
Public API
See their docstrings for details:
esm2quinox
.ESM2
.__init__(self, num_layers: int, embed_size: int, num_heads: int, token_dropout: bool, key: PRNGKeyArray)
.__call__(self, tokens: Int[np.ndarray | jax.Array, " length"]) -> esm2quinox.ESM2Result
.ESM2Result
.hidden: Float[Array, "length embed_size"]
.logits: Float[Array, "length alphabet_size"]
.tokenise(proteins: list[str], length: None | int = None, key: None | PRNGKeyArray = None)
.from_torch(torch_esm2: esm.ESM2) -> esm2quinox.ESM2
Quick examples
Load an equivalent pretrained model from PyTorch:
import esm # pip install fair-esm==2.0.0
import esm2quinox
torch_model, _ = esm.pretrained.esm2_t6_8M_UR50D()
model = esm2quinox.from_torch(torch_model)
Create a randomly-initialised model:
import esm2quinox
import jax.random as jr
key = jr.key(1337)
model = esm2quinox.ESM2(num_layers=3, embed_size=32, num_heads=2, token_dropout=False, key=key)
Forward pass (note the model operates on unbatched data):
proteins = esm2quinox.tokenise(["SPIDERMAN", "FOO"])
out = jax.vmap(model)(proteins)
out.hidden # hidden representation from last layer
out.logits # logits for masked positions
Metadata
Release files for esm2quinox 0.2.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| esm2quinox-0.2.0.tar.gz | 12.4 kB | Without provenance | Details for esm2quinox-0.2.0.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| esm2quinox-0.2.0-py3-none-any.whl | Python 3 | none | any | Details for esm2quinox-0.2.0-py3-none-any.whl |
Total release size: 29.5 kB
Release files / esm2quinox-0.2.0.tar.gz
| Download URL | esm2quinox-0.2.0.tar.gz |
|---|---|
| Size | 12.4 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
12c11632f0b3efd3e3a4f94b031efb3f50f010f8fcbddd9a95b38da716bbb983
|
|
BLAKE2b-256 checksum How to use checksums |
e487d48d18a1faa94ee4560f14cfc559413fcfc610854c29af897bcf9fb91cf9
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.1.0 CPython/3.11.12
|
Release files / esm2quinox-0.2.0-py3-none-any.whl
| Download URL | esm2quinox-0.2.0-py3-none-any.whl |
|---|---|
| Size | 17.1 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
ac89d44ba1e141a60de1a76e10b345fab70d629d7c438031a76f53bd405fc244
|
|
BLAKE2b-256 checksum How to use checksums |
bc34b72aeafa3521e42e300117c1d1cb749a6db534ae88d9c5b28c1dd054001a
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.1.0 CPython/3.11.12
|
Security for esm2quinox 0.2.0
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.