- New York, New York
- https://jon-tow.github.io
Block or Report
Block or report jon-tow
Report abuse
Contact GitHub support about this user’s behavior. Learn more about reporting abuse.
Report abusePinned
523 contributions in the last year
Less
More
Contribution activity
December 2022
Created 49 commits in 3 repositories
Created 3 repositories
- jon-tow/transformers Python
- jon-tow/DeepSpeed Python
- jon-tow/accelerate Python
Created a pull request in CarperAI/trlx that received 7 comments
Add BloomModel hydra support
This PR adds hydra-based PPO model branching support for BloomModels.
wandb reports:
PPO Sentiment
Note: The <Arch>ModelBranch implementations sh…
+203
−5
•
7
comments
Opened 15 other pull requests in 3 repositories
CarperAI/trlx
2
open
7
merged
-
Update
pre-commitversion and addisort - Release 0.3.1
-
Refactor RL model wrapper into a
trainermodule - Set device-specific seeding with global rank
- Freeze hydra model branches
- Remove incorrect default config settings
-
Add
OptimizerConfigandSchedulerConfig -
Add hydra head support for
GPTNeo - Add unit tests to ensure valid example configs
EleutherAI/lm-evaluation-harness
5
merged
jon-tow/text-sed
1
merged
Reviewed 12 pull requests in 1 repository
CarperAI/trlx
12 pull requests
-
Update
pre-commitversion and addisort - Upgrade code quality
- Linear LR scheduler
- Logging learning rate
-
Refactor RL model wrapper into a
trainermodule - Fix seeding per process
- Fix context overflow
-
Add
OptimizerConfigandSchedulerConfig - add: load function for AccelerateRLModel
- bnb adam, grad accumulation, dtype param
-
Add
BloomModelhydra support -
Add support for more
CausalLMs
12
contributions
in private repositories
Dec 3 – Dec 22




