Skip to content

Commit fd3a528

Browse files
[fix] Convert the rest of configs from ymls (#346)
* chore(examples): convert the rest of configs from `ymls` * style(examples): satisfy black * feat(examples/sentiments): add config updating through cli * refactor(examples): delete config ymls * fix(ppo_hh): config reference * chore(hh/readme): update for the new config selection * refactor(trlx): update depreciation warning of implicit configs * fix(ppo_hh): conversion errors * style(examples): satisfy isort * style(trlx): break a long line
1 parent 368a194 commit fd3a528

25 files changed

+307
-771
lines changed

examples/hh/README.md

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -8,7 +8,7 @@ accelerate launch --num_processes 7 --config_file ../../configs/accelerate/zero2
88
```
99
Or if you want to train a smaller model or start from a supervised checkpoint, you can use one of the [configs](./configs)
1010
```sh
11-
CONFIG_PATH=configs/ppo_hh_125M.yml accelerate launch --num_processes 7 --config_file ../../configs/accelerate/zero2-bf16.yaml ppo_hh.py
11+
CONFIG_NAME=125M accelerate launch --num_processes 7 --config_file ../../configs/accelerate/zero2-bf16.yaml ppo_hh.py
1212
```
1313

1414
Already trained models are hosted on https://huggingface.co/reciprocate

examples/hh/configs/ilql_hh.yml

Lines changed: 0 additions & 50 deletions
This file was deleted.

examples/hh/configs/ilql_hh_125M.yml

Lines changed: 0 additions & 50 deletions
This file was deleted.

examples/hh/configs/ilql_hh_1B.yml

Lines changed: 0 additions & 50 deletions
This file was deleted.

examples/hh/configs/ilql_hh_20B.yml

Lines changed: 0 additions & 50 deletions
This file was deleted.

examples/hh/configs/ilql_hh_6B.yml

Lines changed: 0 additions & 50 deletions
This file was deleted.

examples/hh/configs/ppo_hh.yml

Lines changed: 0 additions & 57 deletions
This file was deleted.

examples/hh/configs/ppo_hh_125M.yml

Lines changed: 0 additions & 56 deletions
This file was deleted.

0 commit comments

Comments
 (0)