Release v2.5.0 (#315)

* Add documentation for pipeline steps

* Replace 0.0.0.0 by localhost for Windows

* Bump version

* Update release notes

* Apply suggestions from code review

* Apply suggestions from code review

Co-authored-by: Lindsay Brin <[email protected]>

Co-authored-by: Lindsay Brin <[email protected]>

Author: gabegma

CHANGELOG.md

@@ -11,25 +11,13 @@ Released changes are shown in the
 ## [Not released]
 
 ### Added
-- Azimuth now works on French datasets (and pipelines)! Language can be selected in the config,
-  and language-specific defaults for syntax-tagging and behavioral tests (neutral tokens) will be
-  set dynamically (or can be altered manually).
-- New class overlap section, with class overlap detection. More details are available in the User Guide and in the Key Concepts.
-- Pipeline pre/post-processing steps breakdown in the utterance detail page.
 
 ### Changed
-- The default port for the front end will now be 3000, the React default, instead of 3005.
-- Improved class overlap threshold number picker.
-  - Control input value with a `string` instead of with a `number` so for example when hitting backspace after `0.01`, you get `0.0` instead of `0` (as maybe you wanted to type `2` and get `0.02`).
-  - Debounce number picker to reduce calls to back end.
 
 ### Deprecated/Breaking Changes
 
 ### Removed
 
 ### Fixed
-- ONNX models on GPU.
-- Preserve white spaces in utterances. That includes `\n`s, `\t`s, and consecutive spaces.
-- Fix predictions with 100% confidence not showing up in the confidence histogram and in the ECE.
 
 ### Security

CITATION.cff

@@ -18,7 +18,7 @@ authors:
 - family-names: "Babu"
   given-names: "Nandhini"
 title: "Azimuth, an open-source dataset and error analysis tool for text classification"
-version: 2.4
+version: 2.5
 doi: 10.5281/zenodo.6511558
 date-released: 2022-08-17
 url: "https://github.com/ServiceNow/azimuth"

README.md

@@ -68,7 +68,7 @@ make download_demo
 make CFG_PATH=/config/development/clinc/conf.json launch
 ```
 
-Once the startup tasks are completed, you will be able to access Azimuth at `http://0.0.0.0:8080`.
+Once the startup tasks are completed, you will be able to access Azimuth at `http://localhost:8080`.
 To learn more about Azimuth's features, refer to
 our [Key Concepts](https://servicenow.github.io/azimuth/key-concepts) and
 our [User Guide](https://servicenow.github.io/azimuth/user-guide).

azimuth/routers/README.md

@@ -16,4 +16,4 @@ api_router.include_router(
 ```
 
 You can now test your application by running `poetry run python runner.py $CFG_PATH --debug`.
-The swagger will be up at `0.0.0.0:8091/docs` and you can test your route.
+The swagger will be up at `localhost:8091/docs` and you can test your route.

docs/docs/_static/images/exploration-space/utterance-details-behavioral-tests.png

@@ -92,7 +92,7 @@ running a demo.
        make CFG_PATH=/config/development/clinc/conf.json launch
        ```
 
-3. The **app will be accessible** at [http://0.0.0.0:8080](http://0.0.0.0:8080) after a few minutes
+3. The **app will be accessible** at [http://localhost:8080](http://localhost:8080) after a few minutes
    of waiting. The screen will indicate that the start-up tasks have started. When it is completed,
    the application will be loaded.
 4. Skim the [:material-link: Key Concepts](../key-concepts/index.md) section to get a high-level

docs/docs/_static/images/exploration-space/utterance-details-similarity.png

@@ -105,7 +105,7 @@ dataset and model are available in `config/examples` (`CLINC` is also shown belo
     ```
     make launch
     ```
-4. The **app will be accessible** at `http://0.0.0.0:8080` after a few minutes of waiting. The
+4. The **app will be accessible** at `http://localhost:8080` after a few minutes of waiting. The
    start-up tasks will start.
 
 ## Advanced Settings

docs/docs/getting-started/b-basics.md

@@ -1,5 +1,21 @@
 # Releases
 
+## [2.5.0] - 2022-12-05
+
+### Added
+- **Support for french**: Azimuth now works on French datasets (and pipelines)! Language can be selected in the config, and language-specific defaults for syntax-tagging and behavioral tests (neutral tokens) will be set dynamically (or can be altered manually). [See Reference](../reference/configuration/language.md).
+- **Class Overlap**: New class overlap detection section. More details are available in the [User Guide](../user-guide/class-overlap.md) and in the [Key Concepts](../key-concepts/similarity.md).
+- **Pre/Post-processing Steps**: The details of the pipeline pre/post-processing steps are now visible in the utterance details page.
+- **Accuracy metric**: Accuracy was added as a default metric.
+
+### Changed
+- The default port for the front end will now be 3000, the React default, instead of 3005.
+
+### Fixed
+- ONNX models on GPU.
+- Preserve white spaces in utterances. That includes `\n`s, `\t`s, and consecutive spaces.
+- Fix predictions with 100% confidence not showing up in the confidence histogram and in the ECE.
+
 ## [2.4.1] - 2022-10-21
 
 ### Fixed

docs/docs/getting-started/c-run.md

@@ -2,7 +2,7 @@
 
 Azimuth has limited support for "custom utterances", i.e. utterances not part of the initial dataset.
 
-Azimuth offers support through API routes accessible locally at [`http://0.0.0.0:8091/docs#/Custom%20Utterances%20v1`](http://0.0.0.0:8091/docs#/Custom%20Utterances%20v1).
+Azimuth offers support through API routes accessible locally at [`http://localhost:8091/docs#/Custom%20Utterances%20v1`](http://localhost:8091/docs#/Custom%20Utterances%20v1).
 
 ## Data augmentation
 
@@ -18,7 +18,7 @@ To augment a list of utterances, you can follow these intructions:
     utterances = ["welcome to Azimuth",
                   "Don't forget to star our repo!"]
 
-    response = requests.get("http://0.0.0.0:8091/custom_utterances/perturbed_utterances",
+    response = requests.get("http://localhost:8091/custom_utterances/perturbed_utterances",
                             params={"utterances": utterances}).json()
     pprint([r["perturbedUtterance"] for r in response])
     ```
@@ -47,7 +47,7 @@ If a pipeline allows it, you can get saliency of a custom utterance by doing the
     utterances = ["welcome to Azimuth",
                   "Don't forget to star our repo!"]
 
-    response = requests.get("http://0.0.0.0:8091/custom_utterances/saliency",
+    response = requests.get("http://localhost:8091/custom_utterances/saliency",
                             params={"utterances": utterances, "pipeline_index": 0}).json()
     pprint(response)
     ```

docs/docs/getting-started/changelog.md

@@ -11,12 +11,14 @@ The top section shows the utterance details:
 * **Utterance**: The utterance and its per-token [saliency](../../key-concepts/saliency.md) (if
   available for the model's architecture). Hover over the tokens to the see the associated saliency
   value.
+  * If pre-processing steps are returned by the pipeline, blue chevrons will be shown, allowing the utterance to be displayed after each pre-processing step, from the raw utterance to the processed version sent to the model.
 * **Label**: The utterance's labeled class (the target).
 * **Predictions**:
     * The **top 3** predictions for this utterance are shown.
     * The top prediction is highlighted using the prediction outcome color.
         * If the top prediction was converted to the rejection class based on the threshold, it will
           be added as a 4th element at the top.
+    * By clicking on the blue chevrons, the prediction results after each post-processing step can be viewed, including the model's output before any post-processing.
 * **Smart tags** (where applicable): An automatically computed tag highlighting a certain
   characteristic of the utterance (e.g., long sentences, utterance is missing a verb, utterance
   contains multiple sentences). For more information,

docs/docs/user-guide/custom-utterances.md

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "azimuth"
-version = "2.4.1"
+version = "2.5.0"
 description = "Azimuth provides a unified error analysis experience to data scientists."
 readme = "README.md"
 authors = ["Azimuth team <[email protected]>"]