M-0827_rl_reflect_countdown__0epoch_3and4args__grpo_minibs32_lr1e-6_rolloutn16-rl

Model Details

• Training Method: VeRL Reinforcement Learning (RL)
• Stage Name: rl
• Experiment: 0827_rl_reflect_countdown__0epoch_3and4args__grpo_minibs32_lr1e-6_rolloutn16
• RL Framework: VeRL (Versatile Reinforcement Learning)

Training Configuration

Experiment Tracking

🔗 View complete experiment details: https://huggingface.co/datasets/TAUR-dev/D-ExpTracker__0827_rl_reflect_countdown__0epoch_3and4args__grpo_minibs32_lr1e-6_rolloutn16__v1

Usage

from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("TAUR-dev/M-0827_rl_reflect_countdown__0epoch_3and4args__grpo_minibs32_lr1e-6_rolloutn16-rl")
model = AutoModelForCausalLM.from_pretrained("TAUR-dev/M-0827_rl_reflect_countdown__0epoch_3and4args__grpo_minibs32_lr1e-6_rolloutn16-rl")