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.

Sign up for GitHub

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

Jump to bottom

updating gptj-config #109

Merged
merged 9 commits into from
Dec 4, 2022
Merged

updating gptj-config #109

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
630c95c
updating gptj-config
Dahoas Nov 22, 2022
39990b6
added distributed config logging to wandb
Dahoas Nov 23, 2022
2e4aeb2
Merge branch 'main' into update-gptj
Dahoas Nov 23, 2022
4b2c532
update
Dahoas Nov 23, 2022
e78e47a
merge
Dahoas Nov 23, 2022
25399c2
black fix
Dahoas Nov 23, 2022
fb701e7
adding check for ds_plugin
Dahoas Nov 23, 2022
65ee9c0
removing wandb entity name from default config
Dahoas Nov 23, 2022
f3f43b7
Merge branch 'main' into update-gptj
LouisCastricato Dec 4, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 4 additions & 0 deletions configs/ppo_gptj.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -35,6 +35,10 @@ method:
cliprange: 0.2 # clip range
cliprange_value: 0.2 # clip range
vf_coef: 0.2 # value term weight
scale_reward: False # False | "ref" | "running" estimate against which to scale rewards
ref_mean: null
ref_std: null # rescale rewards with this deviation
cliprange_reward: 10
gen_kwargs:
max_length: 48 # LM max sample gen length
min_length: 48 # LM min sample gen length
Expand Down
3 changes: 1 addition & 2 deletions examples/ppo_sentiments.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,7 +17,7 @@ def get_positive_score(scores):
return dict(map(lambda x: tuple(x.values()), scores))["POSITIVE"]


default_config = yaml.safe_load(open("configs/ppo_config.yml"))
default_config = yaml.safe_load(open("configs/ppo_gptj.yml"))
Copy link
Collaborator

Choose a reason for hiding this comment

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

are you sure you want to change the default to gpt-j? I wouldn't mind the second script which imports main from here and changes config as the most simplest option right now, until we switch to hydra or something else

Copy link
Contributor

Choose a reason for hiding this comment

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

Agreed, this is a bad idea.

Copy link
Collaborator Author

Choose a reason for hiding this comment

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

Yeah I just forgot to change this back.



def main(hparams={}):
Expand Down Expand Up @@ -46,7 +46,6 @@ def reward_fn(samples: List[str]) -> List[float]:
prompts = [" ".join(review.split()[:4]) for review in imdb["text"]]

model = trlx.train(
"lvwerra/gpt2-imdb",
reward_fn=reward_fn,
prompts=prompts,
eval_prompts=["I don't know much about Hungarian underground"] * 64,
Expand Down