Project description

ML inference tools

Requirements

For model export onnx package is required.

Convert to ONNX

Below are some examples:

Convert t5-small:

PYTHONPATH=. python mlit to-onnx --model-type t5 --model-name t5-small --export-dir tmp

Check that it is working:

PYTHONPATH=. python mlit inference --model-name t5-small --base-dir tmp --model-input "translate English to French: How does this model work?" --model-type t5

Convert custom checkpoint:

PYTHONPATH=. python mlit to-onnx --model-type t5 --model-name "../my_custom_model" --export-dir tmp

Check that it is working:

PYTHONPATH=. python mlit inference --model-name my_custom_model --base-dir tmp --model-input "translate English to French: How does this model work?" --model-type t5 --tokenizer-name "t5-small"

Project details

These details have not been verified by PyPI

Release history Release notifications | RSS feed

0.1.6

May 23, 2021

0.1.5

Mar 17, 2021

0.1.4

Mar 14, 2021

0.1.3

Mar 12, 2021

0.1.2

Mar 11, 2021

This version

0.1.1

Mar 7, 2021

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

mlit-0.1.1.tar.gz (9.2 kB view details)

Uploaded Mar 7, 2021 Source

Built Distribution

mlit-0.1.1-py3-none-any.whl (11.8 kB view details)

Uploaded Mar 7, 2021 Python 3

File details

Details for the file mlit-0.1.1.tar.gz.

File metadata

Download URL: mlit-0.1.1.tar.gz
Upload date: Mar 7, 2021
Size: 9.2 kB
Tags: Source
Uploaded using Trusted Publishing? No
Uploaded via: twine/3.3.0 pkginfo/1.7.0 requests/2.25.0 setuptools/49.2.1 requests-toolbelt/0.9.1 tqdm/4.52.0 CPython/3.8.7

File hashes

Hashes for mlit-0.1.1.tar.gz
Algorithm	Hash digest
SHA256	`d14a32fef31ec1289565c6de1fe98ba7ff2b28fdad7a09752b659716075f4ed2`
MD5	`3dbe9802e33cd208b4574b2bec9057ec`
BLAKE2b-256	`f882acdc9f20bdbd52741e5d1d67672b247da503b11dfa75537befb192fef953`

See more details on using hashes here.

File details

Details for the file mlit-0.1.1-py3-none-any.whl.

File metadata

Download URL: mlit-0.1.1-py3-none-any.whl
Upload date: Mar 7, 2021
Size: 11.8 kB
Tags: Python 3
Uploaded using Trusted Publishing? No
Uploaded via: twine/3.3.0 pkginfo/1.7.0 requests/2.25.0 setuptools/49.2.1 requests-toolbelt/0.9.1 tqdm/4.52.0 CPython/3.8.7

File hashes

Hashes for mlit-0.1.1-py3-none-any.whl
Algorithm	Hash digest
SHA256	`df0283223d41708dc06a8a4b4b032a480afa373e941b27d5622046420dd7ef6f`
MD5	`41056c470030b9da0f5162dbad8b6674`
BLAKE2b-256	`adf00cdfd31f2291a6a72ea2e9ff31e201d6986eba9de4982cd065e223e321d4`