rloo

Here are 7 public repositories matching this topic...

sinanuozdemir / oreilly-llm-rl-alignment

This training offers an intensive exploration into the frontier of reinforcement learning techniques with large language models (LLMs). We will explore advanced topics such as Reinforcement Learning with Human Feedback (RLHF), Reinforcement Learning from AI Feedback (RLAIF), Reasoning LLMs, and demonstrate practical applications such as fine-tuning

reinforcement-learning ai llama agents ppo dpo qwen deepseek reward-modeling grpo rloo

Updated Mar 9, 2026
Jupyter Notebook

hectopascal / agent-credit-bench

Star

Exact-oracle conformance tests for turn-level credit assignment in agentic RL (GRPO, RLOO, GAE, GiGPO; verl, TRL, OpenRLHF).

benchmark reinforcement-learning gae credit-assignment trl advantage-estimation llm-agents grpo verl openrlhf rloo

Updated Sep 17, 2026
Python

Rangow4562 / glm-mini-from-scratch

Star

Readable 25.7M-parameter GLM-5.3-Flash-style Hybrid-MoE built from scratch in PyTorch. Features hybrid linear/sparse attention, executable-reward RLOO, and recursive self-improvement on verified programs.

reinforcement-learning deep-learning pytorch moe code-generation glm language-models self-improvement language-model from-scratch multimodal mixture-of-experts llm recursive-self-improvement hybrid-attention rloo glm-5-3-flash

Updated Oct 10, 2026
Python

junyi-lai / shopping-agent-rl

Star

MiniCPM5-2B shopping agent, end-to-end: collection → filtering → curriculum SFT → GRPO/RLOO (veRL) → deterministic Final-200 eval. Strict success 3.0% → 76.5%.

reinforcement-learning llm-agent minicpm grpo shopping-agent rloo

Updated Sep 23, 2026
Python

JQINF / blender-rlhf-lighting

Star

RL + human-preference reward model for automatic product lighting in Blender (NN-inherit → 2-step RLOO refine → best-of-16 RM arbitration)

reinforcement-learning blender cycles-renderer preference-learning lighting-design rlhf rloo

Updated Sep 24, 2026
Python

Roberto-Pecora / rloo-vs-ppo-rlhf

Star

Reproduces Hugging Face's RLOO vs. PPO win-rate comparison on real public checkpoints, using Claude as judge instead of GPT-4. Evaluation only — no training.

nlp benchmarking machine-learning reinforcement-learning reproducible-research alignment summarization claude ppo huggingface rlhf anthropic llm-evaluation llm-as-judge rloo

Updated Aug 22, 2026
Python

marcovizcarra / frontier-curriculum-rloo

Star

Frontier-curriculum RLOO for LLM reasoning, developed for Stanford CS224R. Includes baseline comparisons, a simplified public implementation, evaluation, and results. Improved pass@1 from 0.41 to 0.49 over vanilla RLOO.

reinforcement-learning pytorch reasoning curriculum-learning post-training llm rloo

Updated Sep 3, 2026
Jupyter Notebook

Add this topic to your repo

To associate your repository with the rloo topic, visit your repo's landing page and select "manage topics."

Learn more