beat issues
https://gitlab.idiap.ch/groups/beat/-/issues
2018-12-17T21:50:13Z
https://gitlab.idiap.ch/beat/docs/-/issues/5
[beat.web] documentation on algorithm
2018-12-17T21:50:13Z
Flavio TARSETTI
[beat.web] documentation on algorithm
@andre.anjos:
I think this answers the question:
[8._Algorithms___beat_2018.06.pdf](/uploads/c82a54740e60b9089bb1d5f1f74f9c3e/8._Algorithms___beat_2018.06.pdf)
Please see attached file.
So do we have documentation on how to write algo...
@andre.anjos:
I think this answers the question:
[8._Algorithms___beat_2018.06.pdf](/uploads/c82a54740e60b9089bb1d5f1f74f9c3e/8._Algorithms___beat_2018.06.pdf)
Please see attached file.
So do we have documentation on how to write algorithms on beat.web (1.4 branch) ?
From my point of view the answer is yes. Do we agree ?
If yes we can close this issue.
https://gitlab.idiap.ch/beat/beat.editor/-/issues/138
Can we use Conda's Node.js version?
2019-01-28T04:37:00Z
Jaden DIEFENBAUGH
Can we use Conda's Node.js version?
Pretty sure we can't, but someone should look into it anyways.
Pretty sure we can't, but someone should look into it anyways.
https://gitlab.idiap.ch/beat/beat.editor/-/issues/137
Move the `beateditor` command to beat.cmdline's Click framework
2018-09-05T06:41:52Z
Jaden DIEFENBAUGH
Move the `beateditor` command to beat.cmdline's Click framework
Add another command to `beat.cmdline` via the click framework, moving:
```
$ beateditor
```
to:
```
$ beat editor
```
Add another command to `beat.cmdline` via the click framework, moving:
```
$ beateditor
```
to:
```
$ beat editor
```
Flavio TARSETTI
Flavio TARSETTI
https://gitlab.idiap.ch/beat/beat.cmdline/-/issues/40
`list` commands have a default verbosity level above 1
2018-12-12T09:14:12Z
Jaden DIEFENBAUGH
`list` commands have a default verbosity level above 1
I recently updated `beat.cmdline` for writing the documentation examples and it seems that the verbosity level isn't `0` by default. Running `beat exp list` for the `beat.tutorial.prefix` prefix gives me:
```
Loaded configuration file `/...
I recently updated `beat.cmdline` for writing the documentation examples and it seems that the verbosity level isn't `0` by default. Running `beat exp list` for the `beat.tutorial.prefix` prefix gives me:
```
Loaded configuration file `/home/jae/.beatrc'
anjos/tutorial/eigenface/1/atnt-eigenfaces-172
[05/Jul/2018 15:25:21 - beat.cmdline.common] INFO: anjos/tutorial/eigenface/1/atnt-eigenfaces-172
test/tutorial/eigenface/1/atnt
[05/Jul/2018 15:25:21 - beat.cmdline.common] INFO: test/tutorial/eigenface/1/atnt
test/test/iris_advanced/1/iris
[05/Jul/2018 15:25:21 - beat.cmdline.common] INFO: test/test/iris_advanced/1/iris
test/test/iris/1/iris
[05/Jul/2018 15:25:21 - beat.cmdline.common] INFO: test/test/iris/1/iris
test/test/iris/1/error
[05/Jul/2018 15:25:21 - beat.cmdline.common] INFO: test/test/iris/1/error
test/test/iris/1/means
[05/Jul/2018 15:25:21 - beat.cmdline.common] INFO: test/test/iris/1/means
```
This isn't the correct functionality, right?
Samuel GAIST
Samuel GAIST
https://gitlab.idiap.ch/beat/docs/-/issues/4
[beat.web] issue when adding the package to the docs
2018-07-06T12:40:18Z
Flavio TARSETTI
[beat.web] issue when adding the package to the docs
I am adding beat.web to the docs, I cloned it but after running:
```
../bob.admin/conda/conda-build.sh --python=3 conda
```
I get this error:
`
(14:38:03.372) Error: Command Failed "conda build --no-anaconda-upload --variant-config-fil...
I am adding beat.web to the docs, I cloned it but after running:
```
../bob.admin/conda/conda-build.sh --python=3 conda
```
I get this error:
`
(14:38:03.372) Error: Command Failed "conda build --no-anaconda-upload --variant-config-files /home/neo/bob.admin/conda/../gitlab/conda_build_config.yaml --python=3 conda"
`
any idea on this one @samuel.gaist and @andre.anjos ?
https://gitlab.idiap.ch/beat/beat.editor/-/issues/135
Algorithms made with a group & no group names!
2018-12-19T19:02:44Z
Jaden DIEFENBAUGH
Algorithms made with a group & no group names!
The group names shouldnt be shown/editable by users
The group names shouldnt be shown/editable by users
Miscellaneous Editor-specific Improvements
https://gitlab.idiap.ch/beat/beat.editor/-/issues/134
Implement cache (save/cancel options) for block editor
2018-12-19T19:02:44Z
Jaden DIEFENBAUGH
Implement cache (save/cancel options) for block editor
Toolchain Editor: UX Improvements
https://gitlab.idiap.ch/beat/beat.core/-/issues/59
Tight "soft-loops" controlling algorithmic training
2020-11-04T17:09:18Z
André Anjos
Tight "soft-loops" controlling algorithmic training
In the context of the ALLIES project, a so-far-avoided requirement has popped up again the meeting: block loops.
I created this bug to differentiate from #29, because the requirements here are a bit different: in the use-case to be cons...
In the context of the ALLIES project, a so-far-avoided requirement has popped up again the meeting: block loops.
I created this bug to differentiate from #29, because the requirements here are a bit different: in the use-case to be considered, platform users need to be able to train a model depending on the output of another block. The main research question of the project needs to answer is: is it possible to implement life-long learning systems (LLLS) that can be taught by a human and improve over time? The outcome of the project must be reproducible (therefore our involvement).
In order to simulate this, partners provided the following diagram:
![modes](/uploads/a4e1aa953cd2bced7fa834da5d300e63/modes.png)
On the left, you'll see conventional ways to "adapt" statistical models from new input data. Given the model and more data, the model can be adapted to new incoming information (a.k.a. "source" in these diagrams).
In the context of ALLIES, partners want to have reproducible toolchains that allow third-parties to create the training system on the right of this image. In this context, the yellow boxes under the "training" (cyan) boxes simulate a human-user interaction with the system. E.g.: it could be a human telling a machine translation system the translation of a sentence (or paragraph) needs to be corrected, alongside with the correction, or it could be a human telling a speaker diarization system that a certain segment belongs to a different speaker. Because humans have limited ability to provide corrections, the consortium wants to control the implementation and limits of the "yellow" boxes, while third-parties implement the "trainers" that go into the cyan boxes.
It is possible to "unroll" these loops for practical scenarios, but in order to test limits of LLLS, one can envisage hundreds if not thousand iterations which is not very practical to code in BEAT as of today.
One of the key questions while considering #29 was that of the cache: a system that loops will issue multiply different cache outputs which, using the current architecture, would have to be kept. In this new use-case, such a caching requirement is not present. Partners only want to ensure that the code in "yellow" boxes can be properly tagged by the consortium and that they can conduct searches and build leaderboards on experiments using those (and only those). This prevents implementors of the "cyan" training boxes to misinterpret the role and data from the simulated human (yellow boxes). As a consequence, **no caching is required within this loop**.
Soft loops
https://gitlab.idiap.ch/beat/beat.editor/-/issues/133
tell the user why objects are invalid
2020-02-12T13:06:14Z
Jaden DIEFENBAUGH
tell the user why objects are invalid
this is super hazy and possibly very complicated
this is super hazy and possibly very complicated
https://gitlab.idiap.ch/beat/beat.editor/-/issues/132
graphviz dot: require space between blocks
2018-12-19T19:02:44Z
Jaden DIEFENBAUGH
graphviz dot: require space between blocks
They can be touching - not good
They can be touching - not good
Toolchain Editor: Fixes
https://gitlab.idiap.ch/beat/beat.editor/-/issues/131
Weird edgecases with toolchain validity
2018-12-19T19:02:44Z
Jaden DIEFENBAUGH
Weird edgecases with toolchain validity
Rarely it seems that a toolchain is valid but somehow it's not...
will hopefully probably be fixed by the refactoring & test fixing
Rarely it seems that a toolchain is valid but somehow it's not...
will hopefully probably be fixed by the refactoring & test fixing
Toolchain Editor: Fixes
https://gitlab.idiap.ch/beat/beat.editor/-/issues/130
Don't allow dashes in names
2018-12-19T19:02:44Z
Jaden DIEFENBAUGH
Don't allow dashes in names
It breaks alot of stuff
It breaks alot of stuff
Toolchain Editor: Fixes
https://gitlab.idiap.ch/beat/beat.editor/-/issues/129
Block editor should allow switching inputs
2018-12-19T19:02:44Z
Jaden DIEFENBAUGH
Block editor should allow switching inputs
This is a little tricky - say you have two inputs, `input1` and `input2`. Trying to change the first name to `input2` will fail obviously since there's already an `input2`. But you want to switch them. The only solution now is using a 3r...
This is a little tricky - say you have two inputs, `input1` and `input2`. Trying to change the first name to `input2` will fail obviously since there's already an `input2`. But you want to switch them. The only solution now is using a 3rd name to switch, but this isn't super logical...
not sure how to fix this.
Toolchain Editor: UX Improvements
https://gitlab.idiap.ch/beat/beat.editor/-/issues/128
Dataset blocks should hide/show the same as other blocks
2018-12-19T19:02:44Z
Jaden DIEFENBAUGH
Dataset blocks should hide/show the same as other blocks
Dataset blocks always exist at the bottom of the exp editor - they should only pop up if you click a dataset block. But if you click one dataset block, all of them should pop up
Dataset blocks always exist at the bottom of the exp editor - they should only pop up if you click a dataset block. But if you click one dataset block, all of them should pop up
Experiment Editor: Fixes
https://gitlab.idiap.ch/beat/beat.editor/-/issues/127
Don't remove type inference after an algorithm has been selected for a block
2018-12-19T19:02:44Z
Jaden DIEFENBAUGH
Don't remove type inference after an algorithm has been selected for a block
Before an algorithm is selected, the dropdown menu for algs is filtered via type-inference as it should be. But, after you select an alg and go to select a different one, the type inference filtering is dropped.
Pretty sure I added this...
Before an algorithm is selected, the dropdown menu for algs is filtered via type-inference as it should be. But, after you select an alg and go to select a different one, the type inference filtering is dropped.
Pretty sure I added this as a feature but now it's a bug :')
Experiment Editor: Fixes
https://gitlab.idiap.ch/beat/beat.cmdline/-/issues/39
Better copy/paste support for `exp caches`
2018-06-11T08:13:15Z
Jaden DIEFENBAUGH
Better copy/paste support for `exp caches`
Right now the caches printed are file paths:
```
block: `pre_training'
algorithm: `test/iris_preprocessor/1'
output: `/home/jae/code/beat.editor/prefix/cache/a9/e2/0b/2ba5928422351e52a508bd0cbc55e3d0e5bacb05896dc28413bcd0a1a4'
block:...
Right now the caches printed are file paths:
```
block: `pre_training'
algorithm: `test/iris_preprocessor/1'
output: `/home/jae/code/beat.editor/prefix/cache/a9/e2/0b/2ba5928422351e52a508bd0cbc55e3d0e5bacb05896dc28413bcd0a1a4'
block: `pre_testing'
algorithm: `test/iris_preprocessor/1'
output: `/home/jae/code/beat.editor/prefix/cache/1e/a5/74/80c7d48bae3d0f6baf0883c3a7f78b5324830931029fcee4fd3bffff4d'
block: `training_alg'
algorithm: `test/iris_training/1'
output: `/home/jae/code/beat.editor/prefix/cache/45/61/15/3438e272598bf15e6bd729709536f4ca38b2b9d1061dd286c6dae5f288'
block: `testing_alg'
algorithm: `test/iris_testing/1'
output: `/home/jae/code/beat.editor/prefix/cache/39/d4/20/a6709aca3cac7e9a9a7cb2c36d536184554b5e19358927f2e866ad5751'
block: `analyzer'
analyzer: `test/iris_analyzer/1'
output: `/home/jae/code/beat.editor/prefix/cache/f0/a0/86/a645856dcbee4edc2729fcc60554212ed5def33334670602be6086aafb'
```
But you only use path segments relative to the cache folder to operate on that cache:
```
$ beat ca view a9/e2/0b/2ba5928422351e52a508bd0cbc55e3d0e5bacb05896dc28413bcd0a1a4
```
It would make sense for `exp caches` to only print out this bit, not the absolute path.
https://gitlab.idiap.ch/beat/beat.cmdline/-/issues/38
Offline plotting support is shaky
2020-05-26T15:19:19Z
Jaden DIEFENBAUGH
Offline plotting support is shaky
If the plotter object and the default plotterparameters exist in the local prefix, failing to fetch info from the BEAT web platform should not stop plotting.
The issue came up when using the tutorial VM. Some people didn't have an inter...
If the plotter object and the default plotterparameters exist in the local prefix, failing to fetch info from the BEAT web platform should not stop plotting.
The issue came up when using the tutorial VM. Some people didn't have an internet connection and couldn't plot because `beat.cmdline` was trying to fetch stuff from beat.web even though the plotter objects and default parameters existed.
Samuel GAIST
Samuel GAIST
https://gitlab.idiap.ch/beat/beat.editor/-/issues/126
Fix exp editor assigning datasets wrongly when choosing protocol
2018-11-20T19:14:16Z
Jaden DIEFENBAUGH
Fix exp editor assigning datasets wrongly when choosing protocol
When creating an iris experiment, training & testing were assigned wrongly (why???)
When creating an iris experiment, training & testing were assigned wrongly (why???)
Experiment Editor: Fixes
https://gitlab.idiap.ch/beat/beat.editor/-/issues/125
Update & expand toolchain editor help menu popup thingy
2018-12-19T19:02:44Z
Jaden DIEFENBAUGH
Update & expand toolchain editor help menu popup thingy
Especially after the UX pass
Especially after the UX pass
Toolchain Editor: UX Improvements
https://gitlab.idiap.ch/beat/beat.editor/-/issues/124
State management refactoring
2018-07-26T17:52:14Z
Jaden DIEFENBAUGH
State management refactoring
Thought we already had an issue open for this but I guess not.
Right now the object editors are using a `cache` state field to hold the unsaved changed. These changes aren't propogated to the redux store until the user clicks 'save', wh...
Thought we already had an issue open for this but I guess not.
Right now the object editors are using a `cache` state field to hold the unsaved changed. These changes aren't propogated to the redux store until the user clicks 'save', which will *also* save to the prefix. There's no reason to have a state cache and the redux store anymore.
Instead of using the editors' states' cache fields to store unsaved changes, just use the redux store!
State Management Refactoring