open-thought/reasoning-gym
1
0
Fork 0
mirror of https://github.com/open-thought/reasoning-gym.git synced 2026-04-30 17:40:45 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

added updates

Browse source
This commit is contained in:
joesharratt1229 2025-03-29 08:07:57 +00:00
parent 774d23664d
commit 9f9f816902
10 changed files with 461 additions and 503 deletions
Download patch file Download diff file Expand all files Collapse all files

25
eval/configs/eval_config.yaml
View file

@ -1,25 +0,0 @@
# Model configuration
model: Qwen/Qwen2.5-0.5B-Instruct # Change to the smaller model
hf_path: /workspace/joe/verl_grpo_qwen_3b_curr
max_tokens: 1024 # From max_response_length in training config
temperature: 0.7 # Lower temperature for more focused responses
top_p: 0.9 # From rollout top_p
developer_prompt: DeepSeekZero
developer_role: system # Standard role for system prompts
# Output configuration
output_dir: eval_results
save_metadata: true
save_full_results: true
# Categories and datasets to evaluate
categories:
- category: reasoning
datasets:
- dataset: spell_backward
size: 1000 # From training dataset_size
seed: 42
params:
min_word_len: 3 # From training config
max_word_len: 10
data_file: holdout_words.txt