You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: README.md
+37-29Lines changed: 37 additions & 29 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -280,46 +280,54 @@ To configure automatic updates consistent with the [update protocol](https://git
280
280
-**Double dipping**: Double dipping statistics in different behavior epochs and conditions.
281
281
282
282
### Automatic training
283
-
1. In the main dialog, press `Auto Train` button <imgsrc="https://github.com/AllenNeuralDynamics/dynamic-foraging-task/assets/24734299/9ef26192-044b-4c22-928f-b328b7ab36ab"width="90"> or `Ctrl + Alt + A` to open the Automatic Training dialog
283
+
1. In the main dialog, press `Auto Train` button <imgsrc="https://github.com/AllenNeuralDynamics/dynamic-foraging-task/assets/24734299/836a4432-b1b2-4f92-9c66-2441a9d77a82"width="150"> or `Ctrl + Alt + A` to open the Automatic Training dialog
284
284
> [!IMPORTANT]
285
285
> If the dialog fails to open, check AWS credentials at `~/.aws/credentials`. See [instructions](#for-initial-installation)
- Click buttons to see interative diagrams in browser <br><imgsrc="https://github.com/AllenNeuralDynamics/dynamic-foraging-task/assets/24734299/bb845129-a12b-445a-8639-0e981a60deb9"height="30">
292
-
- Click `Set curriculum` button <imgsrc="https://github.com/AllenNeuralDynamics/dynamic-foraging-task/assets/24734299/f1fac3e0-1c84-42e9-9f12-ba11896845b8"width="60"> to confirm
293
-
- Now a new entry with `session = 0` is added in `Training history`, and `STAGE_1` of the selected curriculum is suggested by default. <br>
1. ‼️ So far, the automatic training system cannot handle our previous Stage 1.1 --> Stage 1.2.<br>
289
+
For a new mouse, please still use `Coupled Baiting - Stage 1.1` in the "Parameters" page to warm up the mouse as before.<br>
290
+
Note we always start with `Coupled Baiting - Stage 1.1` even if `curriculum_name == "Uncoupled ..."` (see [this issue](https://github.com/AllenNeuralDynamics/aind-behavior-blog/issues/241))<br>
Once you think the mouse is ready for Stage 1.2, please follow the steps below to start using automatic training. <br>(`Stage 1` in `Auto Train`= `Stage 1.2` in the `Parameters` tab.)<br><br>
293
+
294
+
2. Confirm that this is a new mouse in the automatic training system<br>
4. (Optional) Click buttons to see interative diagrams in browser <br><imgsrc="https://github.com/AllenNeuralDynamics/dynamic-foraging-task/assets/24734299/99038b93-aef7-48bb-b756-a387bf435a6e"height="30">
299
+
5. Click `Set curriculum` button <imgsrc="https://github.com/AllenNeuralDynamics/dynamic-foraging-task/assets/24734299/2e3f9abd-afda-4a8c-8475-dfb4f2d8065d"width="300"> to confirm
300
+
6. Now a new entry with `session = 0` is added in `Training history`, and `STAGE_1` of the selected curriculum is suggested by default. <br>
-Press `Show all training history` <imgsrc="https://github.com/AllenNeuralDynamics/dynamic-foraging-task/assets/24734299/ff1354b6-4b78-4740-a317-b5ec1b83686e"height="30"> to open an interactive plotly chart that visualize all training history in browser
3. (Optional) Press `Show auto-training history in Streamlit` <imgsrc="https://github.com/AllenNeuralDynamics/dynamic-foraging-task/assets/24734299/c5dc9f2f-8485-47c3-a7e7-ed23947cc4a1"height="30"> to open the Streamlit app showing all training history in browser
3. Note that you can still modify some items in `Training parameters`, such as `Valve open time`, `Give left/right`, and `Next block`.
315
+
4. You could now close the Auto Training dialog.
316
+
5. Start the training as usual.
317
+
311
318
5. Override parameters (not recommended)
312
-
- Once `Apply and lock` is pressed, you can press it again to unlock the parameters and override any of them. But in this case, the automatic training mode is disengaged, and this session is considered "off-curriculum".
313
-
319
+
1. Once `Apply and lock` is pressed, you can press it again to unlock the parameters and override any of them. But in this case, the automatic training mode is disengaged, and this session is considered "off-curriculum". <br>
- Check `Override stage` to override the suggested stage. In the example below, `STAGE_FINAL` is suggested, but `STAGE_3` will be actually used (see the green button).
1. Check `Override stage` to override the suggested stage. In the example below, `STAGE_FINAL` is suggested, but `STAGE_3` will be actually used (see the green button).
- If you somehow decide to change the curriculum during training, press `Override curriculum` and set a new curriculum.
320
-
- In this case, since all stages from the old curriculum now become "irrelevant", you should always manually select a stage in the ***new*** curriculum to override.
327
+
1. If you somehow decide to change the curriculum during training, press `Override curriculum` and set a new curriculum.
328
+
2. In this case, since all stages from the old curriculum now become "irrelevant", you should always manually select a stage in the ***new*** curriculum to override.
# The selected curriculum is the same as the one in use
2176
2177
logger.info(f"Selected curriculum is the same as the one in use. No change is made.")
2177
-
QMessageBox.information(self, "Info", "Selected curriculum is the same as the one in use. No change is made.")
2178
+
QMessageBox.information(self, "Box {}, Info".format(self.MainWindow.box_letter), "Selected curriculum is the same as the one in use. No change is made.")
2178
2179
return
2179
2180
else:
2180
2181
# Confirm with the user about overriding the curriculum
0 commit comments