Skip to content

fix: change grpo default to use 64 prompts per step and 32 generation… #111

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

Merged
parthchadha merged 2 commits into from
Apr 1, 2025
Merged

fix: change grpo default to use 64 prompts per step and 32 generation… #111

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
a962e89
fix: change grpo default to use 64 prompts per step and 32 generation…
parthchadha Apr 1, 2025
0733368
Merge branch 'main' into pchadha/grpo-default
parthchadha Apr 1, 2025
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
2 changes: 1 addition & 1 deletion examples/configs/grpo_math_1B.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -27,7 +27,7 @@ policy:
model_name: "meta-llama/Llama-3.2-1B-Instruct"
train_global_batch_size: 512
train_micro_batch_size: 4
generation_batch_size: 32
generation_batch_size: 32 # Only used when generating using HF backend
logprob_batch_size: 4
max_total_sequence_length: 512
precision: "bfloat16"
Expand Down
6 changes: 5 additions & 1 deletion examples/configs/grpo_math_8B.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,11 +1,15 @@
# GRPO Algorithm Configuration
defaults: "grpo_math_1B.yaml"

grpo:
num_prompts_per_step: 64
num_generations_per_prompt: 32

policy:
model_name: "meta-llama/Llama-3.1-8B-Instruct"
train_global_batch_size: 512
train_micro_batch_size: 1
generation_batch_size: 32
generation_batch_size: 32 # Only used when generating using HF backend
logprob_batch_size: 2
max_total_sequence_length: 4096
precision: "bfloat16"
Expand Down