|
1 | 1 | <!-- markdown-link-check-disable --> |
2 | 2 |
|
| 3 | +# 0.58.0 |
| 4 | + |
| 5 | +## New Annotators |
| 6 | +This release brings in three new integrations for our annotator stack component: [Prodigy](https://prodi.gy/), [Argilla](https://github.com/argilla-io/argilla) and [Pigeon](https://github.com/agermanidis/pigeon). |
| 7 | +* Pigeon works within Jupyter notebooks and supports a limited feature set but is great for |
| 8 | +experimentation and demos. |
| 9 | +* Argilla works both locally-deployed and when the annotation instance lives in the cloud |
| 10 | +(i.e. in the Hugging Face Spaces deployment which they recommend). |
| 11 | +* Prodigy is a powerful closed-source annotation tool that allows for efficient data labeling. With this |
| 12 | +integration, users can now connect ZenML with Prodigy and leverage its annotation capabilities |
| 13 | +in their ML pipelines. |
| 14 | + |
| 15 | +## Retry configuration for steps |
| 16 | +This release also includes new `retry` configuration for the steps. The following parameters |
| 17 | +can be set: |
| 18 | + |
| 19 | +- _**max_retries**_: The maximum number of times the step should be retried in case of failure. |
| 20 | +- _**delay**_: The initial delay in seconds before the first retry attempt. |
| 21 | +- _**backoff**_: The factor by which the delay should be multiplied after each retry attempt. |
| 22 | + |
| 23 | +To use this in your code: |
| 24 | + |
| 25 | +```python |
| 26 | +from zenml.config.retry_config import StepRetryConfig |
| 27 | + |
| 28 | +@step(retry=StepRetryConfig(max_retries=3, delay=10, backoff=2)) |
| 29 | +def step_3() -> None: |
| 30 | + # Step implementation |
| 31 | + raise Exception("This is a test exception") |
| 32 | +``` |
| 33 | + |
| 34 | +or using a `config.yaml`: |
| 35 | + |
| 36 | +```yaml |
| 37 | +steps: |
| 38 | + my_step: |
| 39 | + retry: |
| 40 | + max_retries: 3 |
| 41 | + delay: 10 |
| 42 | + backoff: 2 |
| 43 | + ``` |
| 44 | +
|
| 45 | +In addition, this release includes a number of bug fixes and documentation updates, such |
| 46 | +as a new LLM finetuning template powered by PEFT and BitsAndBytes and instructions for the |
| 47 | +new annotators. |
| 48 | +
|
| 49 | +
|
| 50 | +## Breaking changes |
| 51 | +* The interface for the base class of the annotator stack component has been updated to |
| 52 | +account for the fact that not all annotators will launch with a specific URL. So there is |
| 53 | +no longer an url argument passed in. |
| 54 | +
|
| 55 | +## 🥳 Community Contributions 🥳 |
| 56 | +
|
| 57 | +We'd like to give a special thanks to @christianversloot who contributed to this release |
| 58 | +by bumping the `mlflow` version to 2.12.2 |
| 59 | + |
| 60 | +## What's Changed |
| 61 | +* Add more failure logs for code repositories and build reuse by @schustmi in https://github.com/zenml-io/zenml/pull/2697 |
| 62 | +* Prodigy annotator by @strickvl in https://github.com/zenml-io/zenml/pull/2655 |
| 63 | +* Bump mlflow support to version 2.12.2 by @christianversloot in https://github.com/zenml-io/zenml/pull/2693 |
| 64 | +* add 0.57.1 to migration test scripts by @safoinme in https://github.com/zenml-io/zenml/pull/2702 |
| 65 | +* Pigeon annotator by @strickvl in https://github.com/zenml-io/zenml/pull/2641 |
| 66 | +* Allow credentials expiry to be configured for service connectors by @stefannica in https://github.com/zenml-io/zenml/pull/2704 |
| 67 | +* Argilla annotator by @strickvl in https://github.com/zenml-io/zenml/pull/2687 |
| 68 | +* Add `MySQL` and `mariadb` migration tests to Slow CI by @safoinme in https://github.com/zenml-io/zenml/pull/2686 |
| 69 | +* Misc small fixes by @schustmi in https://github.com/zenml-io/zenml/pull/2712 |
| 70 | +* Allow resetting server and user metadata by @schustmi in https://github.com/zenml-io/zenml/pull/2666 |
| 71 | +* Fix Docker failures in the CI by @avishniakov in https://github.com/zenml-io/zenml/pull/2716 |
| 72 | +* Add note about helm dependencies by @strickvl in https://github.com/zenml-io/zenml/pull/2709 |
| 73 | +* Add retry config for failing steps by @safoinme in https://github.com/zenml-io/zenml/pull/2627 |
| 74 | +* Update pyparsing version by @strickvl in https://github.com/zenml-io/zenml/pull/2710 |
| 75 | +* New ruff issue by @avishniakov in https://github.com/zenml-io/zenml/pull/2718 |
| 76 | +* PEFT LLM Template by @avishniakov in https://github.com/zenml-io/zenml/pull/2719 |
| 77 | +* Add `model_version_id` as part of the Model config by @avishniakov in https://github.com/zenml-io/zenml/pull/2703 |
| 78 | +* Add more runners to fast CI by @safoinme in https://github.com/zenml-io/zenml/pull/2706 |
| 79 | +* Fail faster on notebook installation and only clone / download the branch we need for `zenml go` by @strickvl in https://github.com/zenml-io/zenml/pull/2721 |
| 80 | +* Make a clear separation between server and dashboard API in the server configuration by @stefannica in https://github.com/zenml-io/zenml/pull/2722 |
| 81 | +* Update pymysql to fix CVE-2024-36039 by @stefannica in https://github.com/zenml-io/zenml/pull/2714 |
| 82 | +* Allow specifying privileged mode for Kubernetes orchestrator containers by @schustmi in https://github.com/zenml-io/zenml/pull/2717 |
| 83 | +* Don't use pod resources/affinity for kubernetes orchestrator pod by @schustmi in https://github.com/zenml-io/zenml/pull/2707 |
| 84 | +* Extra test for artifact listing by @avishniakov in https://github.com/zenml-io/zenml/pull/2715 |
| 85 | +* Pipeline run not tracked in cached artifact version by @avishniakov in https://github.com/zenml-io/zenml/pull/2713 |
| 86 | + |
| 87 | + |
| 88 | +**Full Changelog**: https://github.com/zenml-io/zenml/compare/0.57.1...0.58.0 |
| 89 | + |
| 90 | + |
3 | 91 | # 0.57.1 |
4 | 92 |
|
5 | 93 | This a minor release that brings a variety of enhancements for |
|
0 commit comments