Add "trainremote" command #906
Open
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
@BillyCheung10botics - this is a great addition. I have three comments:
|
|
@BillyCheung10botics - can you have a look at addressing the installation issues from the checks above, and a couple of minor comments? It would be great to have that change in Donkey Car. |
DocGarbanzo
requested changes
Sep 3, 2021
| 'donkeycar_version': str(__version__) | ||
| } | ||
| ) | ||
| print(f"URL submitted to: {submit_job_url}") |
There was a problem hiding this comment.
Please switch to logging statements instead of print, that's what we are using now.
| # The MultipartEncoder provides the content-type header with the boundary: | ||
| headers={'Content-Type': mp_encoder.content_type} | ||
| ) | ||
| print(f"Submission response: HTTP {r.status_code}") |
| if ("job_uuid" in r.json()): | ||
| try: | ||
| uuid = r.json()['job_uuid'] | ||
| print(f"Submitted a Training Job to the remote server {submit_job_url}\n uuid: {uuid}") |
| print(f"Submitted a Training Job to the remote server {submit_job_url}\n uuid: {uuid}") | ||
| return uuid | ||
| except Exception as e: | ||
| print(e) |
| def get_latest_job_status_from_hq(self, refresh_job_statuses, job_uuids): | ||
| import requests | ||
|
|
||
| print(f"Getting lastest job status for uuid {job_uuids}") |
| while checkagain and (run_time < self.TIMEOUT): | ||
| result = self.get_latest_job_status_from_hq(url_status, uuid)[0] | ||
| checkagain = result['status'] == "SCHEDULED" or result['status'] == "TRAINING" | ||
| print(f"Training Status:{result['status']} at {datetime.now().strftime('%Y-%m-%d, %H:%M:%S')}") |
| raise Exception(f"Failed to train the submitted job\nError : {result['status']}") | ||
|
|
||
| run_time = time.time()-start_time | ||
| print(f"Time spent: {run_time} s") |
| @@ -33,6 +33,9 @@ dependencies: | |||
| - psutil | |||
| - kivy=2.0.0 | |||
| - plotly | |||
| - requests | |||
| - requests_toolbelt | |||
There was a problem hiding this comment.
It seems we can't find this package in conda main and conda forge channels, maybe it's only available in pip?
| @@ -35,6 +35,9 @@ dependencies: | |||
| - kivy=2.0.0 | |||
| - plotly | |||
| - tensorflow==2.2.0 | |||
| - requests | |||
| - requests_toolbelt | |||
There was a problem hiding this comment.
This seems to produce the same problem as in the Mac conda yaml file.
| @@ -34,6 +34,9 @@ dependencies: | |||
| - kivy=2.0.0 | |||
| - plotly | |||
| - psutil | |||
| - requests | |||
| - requests_toolbelt | |||
There was a problem hiding this comment.
No idea if this works in windows... do you have any means to check that?
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This commit adds a new command "trainremote" that allows users to train their ai model on our AWS server. Currently, the training is hardcoded to train a KerasLinear model with max epoch of 30. You can use this command as the following example shown,
with mandatory arguments
--tub,--pathand--model,, and optional arguments--urland--get--tub: specify the path(s) of your data tub(s)
--path: specify the path of your mycar
--model: specify the file name (without extension) of the resulting model, loss function plot and movie
--url: the url to post a training request
--get: the url to get the status of the submitted job
After submitting a job via the "trainremote", the training takes place on our AWS server with a max. duration of 15 min. You can view your submitted job (and others' jobs) on https://hq.robocarstore.com/train/. Usually, the training takes around 10 min to complete, a model file, a plot and a movie are then downloaded from our server, and saved to the folder 'models' and 'movies' respectively.