Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Upload the model with push_to_hub in examples #297

Merged
merged 17 commits into from
Jul 7, 2021
Merged

Conversation

leshanbog
Copy link
Collaborator

Pushing model and optimizer checkpoints to HuggingFace Hub using their API

@leshanbog leshanbog force-pushed the better_model_upload branch 3 times, most recently from 716cd33 to 5c0ca8c Compare June 30, 2021 10:34
@leshanbog leshanbog force-pushed the better_model_upload branch from 5c0ca8c to f037ea6 Compare June 30, 2021 11:36
@yhn112 yhn112 self-requested a review July 1, 2021 11:08
Copy link
Member

@justheuristic justheuristic left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Everything appears to be in order.

NB: ignore the codecov message below, it's a lie.

Let's test this in one of the internal runs (we could take an existing run and add an auxiliary uploader @ QYP, then see if it can upload 2 consecutive checkpoints)

@codecov
Copy link

codecov bot commented Jul 2, 2021

Codecov Report

Merging #297 (294f474) into master (e9956b8) will increase coverage by 0.16%.
The diff coverage is 92.19%.

@@            Coverage Diff             @@
##           master     #297      +/-   ##
==========================================
+ Coverage   81.86%   82.03%   +0.16%     
==========================================
  Files          63       65       +2     
  Lines        5813     5856      +43     
==========================================
+ Hits         4759     4804      +45     
+ Misses       1054     1052       -2     
Impacted Files Coverage Δ
hivemind/averaging/group_info.py 100.00% <ø> (ø)
hivemind/averaging/load_balancing.py 95.91% <ø> (ø)
hivemind/averaging/partition.py 98.01% <ø> (ø)
hivemind/hivemind_cli/run_server.py 0.00% <0.00%> (ø)
hivemind/moe/client/expert.py 96.49% <ø> (ø)
hivemind/moe/server/expert_uid.py 69.64% <0.00%> (ø)
hivemind/moe/server/layers/lr_schedule.py 100.00% <ø> (ø)
hivemind/optim/adaptive.py 77.77% <ø> (ø)
hivemind/optim/base.py 71.42% <ø> (-1.30%) ⬇️
hivemind/p2p/p2p_daemon.py 91.18% <ø> (-0.45%) ⬇️
... and 46 more

yhn112 and others added 3 commits July 2, 2021 20:45
- Removed hivemind.utils.threading.run_in_background and HIVEMIND_THREADS
- Refactored MPFuture to be a single object instead of a linked pair of objects
- MPFuture now uses a single process-wide pipe and thread, instead of spawning new pipe/thread for each future
- MPFuture.result/exception can now only be awaited from the process that created it
- MPFuture now returns the same exception types as regular future (and as asyncio.Future in __await__)
- Added more thorough tests for MPFuture

Co-authored-by: Alexander Borzunov <[email protected]>
Co-authored-by: Max Ryabinin <[email protected]>
Co-authored-by: Michael Diskin <[email protected]>
* Split hivemind.client into hivemind.averaging and hivemind.moe

* Reduce the number of wildcard imports, update docs
@mryab mryab changed the title Change model uploading method Upload the model with push_to_hub in examples Jul 7, 2021
@mryab mryab merged commit 2436a3b into master Jul 7, 2021
@mryab mryab deleted the better_model_upload branch July 7, 2021 14:18
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants