Logging update: added PID and formatting

[theobjectivedad]

Adding a logging enhancement that adds PID, package name, and line number to each logging message, ex:

[2023-07-14 12:35:12,981] [INFO] [axolotl.scripts.train:218] [PID:19] loading tokenizer... /models/llama-7b-hf
[2023-07-14 12:35:12,987] [INFO] [axolotl.scripts.train:218] [PID:18] loading tokenizer... /models/llama-7b-hf
[2023-07-14 12:35:12,988] [INFO] [axolotl.scripts.train:218] [PID:17] loading tokenizer... /models/llama-7b-hf
...
[2023-07-14 12:35:33,253] [INFO] [torch.distributed.distributed_c10d._store_based_barrier:432] [PID:18] Added key: store_based_barrier_key:2 to store for rank: 1
[2023-07-14 12:35:33,276] [INFO] [torch.distributed.distributed_c10d._store_based_barrier:432] [PID:17] Added key: store_based_barrier_key:2 to store for rank: 0
[2023-07-14 12:35:33,276] [INFO] [torch.distributed.distributed_c10d._store_based_barrier:432] [PID:19] Added key: store_based_barrier_key:2 to store for rank: 2
[2023-07-14 12:35:33,277] [INFO] [torch.distributed.distributed_c10d._store_based_barrier:466] [PID:19] Rank 2: Completed store-based barrier for key:store_based_barrier_key:2 with 3 nodes.
[2023-07-14 12:35:33,283] [INFO] [torch.distributed.distributed_c10d._store_based_barrier:466] [PID:18] Rank 1: Completed store-based barrier for key:store_based_barrier_key:2 with 3 nodes.
[2023-07-14 12:35:33,286] [INFO] [torch.distributed.distributed_c10d._store_based_barrier:466] [PID:17] Rank 0: Completed store-based barrier for key:store_based_barrier_key:2 with 3 nodes.

Background:

I'm working on potential DeepSpeed + additional logging/visibility updates and find it incredibly useful to see which GPU process and exactly which module path/line number triggered a logging event.

[winglian]

lgtm

[winglian]

@theobjectivedad would you mind fixing a few things for the linter please? thanks!

pre-commit run --all-files

[winglian]

also, lmk if you want any help with the failing unit tests https://github.com/OpenAccess-AI-Collective/axolotl/actions/runs/5554165505/jobs/10144233892?pr=276

there are checks in the unit tests that check the config validation, so the output messages likely changed (or are now generators it seems). It might be easiest to not use the logging module for the validation sub-module

[theobjectivedad]

Ty @winglian, I'll have some time to take a crack at it this Sunday. I'm planning to use Axolotl for my research projects so I need to have a proper dev environment setup anyway.

[theobjectivedad]

Good morning @winglian, please see my last commit ... everything appears to be passing now locally:

=== 30 passed in 3.62s ===
check yaml...............................................................Passed
fix end of files.........................................................Passed
trim trailing whitespace.................................................Passed
black....................................................................Passed
isort....................................................................Passed
flake8...................................................................Passed
pylint...................................................................Passed
mypy.....................................................................Passed
bandit...................................................................Passed