Unbreak torch/distributed/rpc/batch

[msaroufim]

These examples were just crashing even on older versions of torch

(create) ➜  batch git:(main) ✗ python reinforce.py
(create) ➜  batch git:(main) ✗ python reinforce.py
[Gloo] Rank 0 is connected to 1 peer ranks. Expected number of connected peer ranks is : 1
[Gloo] Rank 1 is connected to 1 peer ranks. Expected number of connected peer ranks is : 1
Episode 0       Last reward: 22.00      Average reward: 11.00
Episode 1       Last reward: 45.00      Average reward: 28.00
Episode 2       Last reward: 11.00      Average reward: 19.50
Episode 3       Last reward: 33.00      Average reward: 26.25
Episode 4       Last reward: 10.00      Average reward: 18.12
Episode 5       Last reward: 41.00      Average reward: 29.56
Episode 6       Last reward: 18.00      Average reward: 23.78
Episode 7       Last reward: 18.00      Average reward: 20.89
Episode 8       Last reward: 60.00      Average reward: 40.45
Episode 9       Last reward: 55.00      Average reward: 47.72
[Gloo] Rank 1 is connected to 1 peer ranks. Expected number of connected peer ranks is : 1
[Gloo] Rank 0 is connected to 1 peer ranks. Expected number of connected peer ranks is : 1
Episode 0       Last reward: 22.00      Average reward: 11.00
Episode 1       Last reward: 45.00      Average reward: 28.00
Episode 2       Last reward: 11.00      Average reward: 19.50
Episode 3       Last reward: 33.00      Average reward: 26.25
Episode 4       Last reward: 10.00      Average reward: 18.12
Episode 5       Last reward: 41.00      Average reward: 29.56
Episode 6       Last reward: 18.00      Average reward: 23.78
Episode 7       Last reward: 18.00      Average reward: 20.89
Episode 8       Last reward: 60.00      Average reward: 40.45
Episode 9       Last reward: 55.00      Average reward: 47.72
2, 13.118742942810059, 13.514524221420288


(create) ➜  batch git:(main) ✗ python parameter_server.py
[Gloo] Rank 0 is connected to 5 peer ranks. Expected number of connected peer ranks is : 5
[Gloo] Rank 3 is connected to 5 peer ranks. Expected number of connected peer ranks is : 5
[Gloo] Rank 5 is connected to 5 peer ranks. Expected number of connected peer ranks is : 5
[Gloo] Rank 2 is connected to 5 peer ranks. Expected number of connected peer ranks is : 5
[Gloo] Rank 1 is connected to 5 peer ranks. Expected number of connected peer ranks is : 5
[Gloo] Rank 4 is connected to 5 peer ranks. Expected number of connected peer ranks is : 5
[Gloo] Rank 0 is connected to 5 peer ranks. Expected number of connected peer ranks is : 5
[Gloo] Rank 1 is connected to 5 peer ranks. Expected number of connected peer ranks is : 5
[Gloo] Rank 3 is connected to 5 peer ranks. Expected number of connected peer ranks is : 5
[Gloo] Rank 2 is connected to 5 peer ranks. Expected number of connected peer ranks is : 5
[Gloo] Rank 5 is connected to 5 peer ranks. Expected number of connected peer ranks is : 5
[Gloo] Rank 4 is connected to 5 peer ranks. Expected number of connected peer ranks is : 5
12:53:07 Start training
12:53:09 trainer5 processing one batch
12:53:09 trainer3 processing one batch
12:53:09 trainer1 processing one batch
12:53:09 trainer2 processing one batch
12:53:09 trainer4 processing one batch
12:53:10 trainer5 reporting grads
12:53:10 trainer1 reporting grads
12:53:11 PS got 0/5 updates
12:53:11 trainer3 reporting grads
12:53:11 trainer2 reporting grads
12:53:11 PS got 1/5 updates
12:53:11 PS got 2/5 updates
12:53:11 trainer4 reporting grads
12:53:11 PS got 3/5 updates
12:53:11 PS got 4/5 updates
12:53:11 PS updated model
12:53:11 trainer5 got updated model
12:53:11 trainer5 processing one batch
12:53:11 trainer3 got updated model
12:53:11 trainer3 processing one batch
12:53:11 trainer2 got updated model
12:53:11 trainer2 processing one batch
12:53:11 trainer5 reporting grads
12:53:11 trainer3 reporting grads
12:53:11 trainer2 reporting grads
12:53:11 trainer1 got updated model
12:53:11 trainer4 got updated model
12:53:11 trainer1 processing one batch
12:53:11 trainer4 processing one batch
12:53:11 trainer1 reporting grads
12:53:11 trainer4 reporting grads
12:53:11 PS got 0/5 updates
12:53:11 PS got 0/5 updates
12:53:11 PS got 1/5 updates
12:53:11 PS got 3/5 updates
12:53:11 PS got 4/5 updates
12:53:12 PS updated model
12:53:12 trainer3 got updated model
12:53:12 trainer3 processing one batch
12:53:12 trainer3 reporting grads
12:53:12 trainer2 got updated model
12:53:12 trainer5 got updated model
12:53:12 trainer5 processing one batch
12:53:12 trainer2 processing one batch
12:53:12 trainer4 got updated model
12:53:12 trainer4 processing one batch
12:53:12 trainer2 reporting grads
12:53:12 trainer5 reporting grads
12:53:12 trainer1 got updated model
12:53:12 trainer4 reporting grads
12:53:12 trainer1 processing one batch
12:53:12 trainer1 reporting grads
12:53:12 PS got 0/5 updates
12:53:12 PS got 1/5 updates
12:53:12 PS got 1/5 updates
12:53:12 PS got 2/5 updates
12:53:12 PS got 3/5 updates
12:53:12 PS updated model
12:53:12 trainer3 got updated model
12:53:12 trainer3 processing one batch
12:53:13 trainer3 reporting grads
12:53:13 PS got 0/5 updates
12:53:13 trainer1 got updated model
12:53:13 trainer2 got updated model
12:53:13 trainer1 processing one batch
12:53:13 trainer2 processing one batch
12:53:13 trainer1 reporting grads
12:53:13 trainer4 got updated model
12:53:13 trainer2 reporting grads
12:53:13 trainer4 processing one batch
12:53:13 trainer5 got updated model
12:53:13 trainer5 processing one batch
12:53:13 trainer4 reporting grads
12:53:13 trainer5 reporting grads
12:53:13 PS got 1/5 updates
12:53:13 PS got 1/5 updates
12:53:13 PS got 2/5 updates
12:53:13 PS got 4/5 updates
12:53:13 PS updated model
12:53:13 trainer3 got updated model
12:53:13 trainer3 processing one batch
12:53:13 trainer3 reporting grads
12:53:13 PS got 0/5 updates
12:53:13 trainer1 got updated model
12:53:13 trainer1 processing one batch
12:53:13 trainer1 reporting grads
12:53:13 trainer5 got updated model
12:53:13 trainer2 got updated model
12:53:13 trainer5 processing one batch
12:53:13 trainer2 processing one batch
12:53:13 trainer4 got updated model
12:53:13 trainer4 processing one batch
12:53:13 trainer4 reporting grads
12:53:13 trainer2 reporting grads
12:53:13 trainer5 reporting grads
12:53:13 PS got 1/5 updates
12:53:13 PS got 2/5 updates
12:53:13 PS got 2/5 updates
12:53:13 PS got 4/5 updates
12:53:13 PS updated model
12:53:14 trainer3 got updated model
12:53:14 trainer3 processing one batch
12:53:14 trainer3 reporting grads
12:53:14 PS got 0/5 updates
12:53:14 trainer1 got updated model
12:53:14 trainer2 got updated model
12:53:14 trainer1 processing one batch
12:53:14 trainer2 processing one batch
12:53:14 trainer5 got updated model
12:53:14 trainer5 processing one batch
12:53:14 trainer1 reporting grads
12:53:14 trainer2 reporting grads
12:53:14 trainer5 reporting grads
12:53:14 trainer4 got updated model
12:53:14 trainer4 processing one batch
12:53:14 trainer4 reporting grads
12:53:14 PS got 1/5 updates
12:53:14 PS got 1/5 updates
12:53:14 PS got 1/5 updates
12:53:14 PS got 2/5 updates
12:53:14 PS updated model
12:53:14 trainer3 got updated model
12:53:14 trainer5 got updated model
12:53:14 trainer2 got updated model
12:53:14 trainer1 got updated model
12:53:14 trainer4 got updated model
12:53:14 Finish training
(create) ➜  batch git:(main) ✗

[netlify]

✅ Deploy Preview for pytorch-examples-preview canceled.

Name	Link
🔨 Latest commit	3b1a2d1
🔍 Latest deploy log	https://app.netlify.com/projects/pytorch-examples-preview/deploys/68acbfcc523802000826142c