RRepoGEO

REPOGEO REPORT · LITE

PKU-Alignment/safe-rlhf

Default branch main · commit e8cca166 · scanned 6/29/2026, 7:27:52 PM

GitHub: 1,606 stars · 133 forks

Scan history for this repo

Score trend below includes all ready runs (older left, newer right; scroll horizontally if needed). The table is collapsed by default—expand for newest-first rows, 10 per page.

Score trend (left → right: older → newer)

2 ready scans. Expand the table below for newest-first rows (10 per page, paginated).

AI VISIBILITY SCORE
40 /100
Critical
Category recall
0 / 2
Not recommended in any query
Rule findings
2 pass · 0 warn · 0 fail
Objective metadata checks
AI knows your name
3 / 3
Direct prompts that named your repo
HOW TO READ THIS REPORT

Action plan is what to do next — copy-pasteable changes prioritized by impact. Category visibility is the real GEO test: when a user asks an AI a brand-free question that should surface PKU-Alignment/safe-rlhf, does the AI actually recommend you — or your competitors? Objective checks verify the metadata signals AI engines weight first. Self-mention check detects whether AI even knows you exist by name.

Action plan — copy-paste fixes

3 prioritized changes generated by gemini-2.5-flash. Mark items done after you ship the fix.

OVERALL DIRECTION
  • highreadme#1
    Reposition the README's opening paragraph to highlight unique safety focus

    Why:

    CURRENT
    Beaver is a highly modular open-source RLHF framework developed by the PKU-Alignment team at Peking University. It aims to provide training data and a reproducible code pipeline for alignment research, especially constrained alignment LLM research via Safe RLHF methods.
    COPY-PASTE FIX
    Beaver is the leading open-source framework for **Safe Reinforcement Learning from Human Feedback (Safe RLHF)**, developed by the PKU-Alignment team. It provides a robust, reproducible code pipeline and extensive datasets specifically designed for **constrained value alignment of Large Language Models**, ensuring safety and mitigating undesirable behaviors.
  • mediumreadme#2
    Add a 'Comparison' section to the README

    Why:

    COPY-PASTE FIX
    Add a new section titled 'Why Choose Safe RLHF? Key Differentiators' or 'Comparison to Other RLHF Frameworks' that highlights how safe-rlhf's focus on safety constraints, cost models, and specific datasets sets it apart from general RLHF implementations.
  • lowreadme#3
    Reorder the README to place 'What's New' further down

    Why:

    CURRENT
    The 'What's New?' section immediately follows the introductory paragraph and features list.
    COPY-PASTE FIX
    Move the 'What's New?' section to appear after the 'Key features of Beaver are:' list and potentially after a 'Getting Started' or 'Usage' section, ensuring the core value is presented first.

Category GEO backends resolved for this scan: google/gemini-2.5-flash, deepseek/deepseek-v4-flash

Category visibility — the real GEO test

Brand-free queries asked to google/gemini-2.5-flash. Did AI recommend you, or someone else?

Same questions for every model — switch tabs to compare answers and rankings.

Recall
0 / 2
0% of queries surface PKU-Alignment/safe-rlhf
Avg rank
—
Lower is better. #1 = top recommendation.
Share of voice
0%
Of all named tools, what % are you?
Top rival
huggingface/trl
Recommended in 2 of 2 queries
COMPETITOR LEADERBOARD
  1. huggingface/trl · recommended 2×
  2. huggingface/transformers · recommended 2×
  3. RLHF (Reinforcement Learning from Human Feedback) by Hugging Face · recommended 1×
  4. openai/spinningup · recommended 1×
  5. vwxyzjn/cleanrl · recommended 1×
  • CATEGORY QUERY
    Need an open-source solution for safe reinforcement learning with human feedback.
    you: not recommended
    AI recommended (in order):
    1. RLHF (Reinforcement Learning from Human Feedback) by Hugging Face
    2. TRL (Transformer Reinforcement Learning) by Hugging Face (huggingface/trl)
    3. Safe Reinforcement Learning (Safe RL) by OpenAI (Baselines/Spinning Up) (openai/spinningup)
    4. CleanRL (vwxyzjn/cleanrl)
    5. Ray RLib (ray-project/ray)

    AI recommended 5 alternatives but never named PKU-Alignment/safe-rlhf. This is the gap to close.

    Show full AI answer
  • CATEGORY QUERY
    How to align large language models with human values while enforcing safety constraints?
    you: not recommended
    AI recommended (in order):
    1. Hugging Face Transformers (huggingface/transformers)
    2. TRL (Transformer Reinforcement Learning) (huggingface/trl)
    3. Stable Baselines3 (DLR-RM/stable-baselines3)
    4. Anthropic
    5. PyTorch (pytorch/pytorch)
    6. TensorFlow (tensorflow/tensorflow)
    7. Giskard (Giskard-AI/giskard)
    8. Arthur AI
    9. Apache Spark (apache/spark)
    10. Dask (dask/dask)
    11. Cleanlab (cleanlab/cleanlab)
    12. Google Cloud's Perspective API
    13. OpenAI's Moderation API
    14. NVIDIA NeMo Guardrails (NVIDIA/NeMo-Guardrails)
    15. scikit-learn (scikit-learn/scikit-learn)
    16. Hugging Face Transformers (huggingface/transformers)
    17. LangChain (langchain-ai/langchain)
    18. LlamaIndex (run-llama/llama_index)
    19. LIME (Local Interpretable Model-agnostic Explanations) (marcotcr/lime)
    20. SHAP (SHapley Additive exPlanations) (shap/shap)
    21. Captum (pytorch/captum)
    22. InterpretML (interpretml/interpretml)

    AI recommended 22 alternatives but never named PKU-Alignment/safe-rlhf. This is the gap to close.

    Show full AI answer

Objective checks

Rule-based audits of metadata signals AI engines weight most.

  • Metadata completeness
    pass

  • README presence
    pass

Self-mention check

Does AI even know your repo exists when asked about it directly?

  • Compared to common alternatives in this category, what is the core differentiator of PKU-Alignment/safe-rlhf?
    pass
    AI named PKU-Alignment/safe-rlhf explicitly

    AI answers can be confidently wrong. Read for accuracy: does it match your actual tech stack, audience, and differentiator?

  • If a team adopts PKU-Alignment/safe-rlhf in production, what risks or prerequisites should they evaluate first?
    pass
    AI named PKU-Alignment/safe-rlhf explicitly

    AI answers can be confidently wrong. Read for accuracy: does it match your actual tech stack, audience, and differentiator?

  • In one sentence, what problem does the repo PKU-Alignment/safe-rlhf solve, and who is the primary audience?
    pass
    AI named PKU-Alignment/safe-rlhf explicitly

    AI answers can be confidently wrong. Read for accuracy: does it match your actual tech stack, audience, and differentiator?

Embed your GEO score

Drop this badge into the README of PKU-Alignment/safe-rlhf. It auto-updates whenever the report is rescanned and links back to the latest report — easy public proof that you care about AI discoverability.

RepoGEO badge previewLive preview
MARKDOWN (README)
[![RepoGEO](https://repogeo.com/badge/PKU-Alignment/safe-rlhf.svg)](https://repogeo.com/en/r/PKU-Alignment/safe-rlhf)
HTML
<a href="https://repogeo.com/en/r/PKU-Alignment/safe-rlhf"><img src="https://repogeo.com/badge/PKU-Alignment/safe-rlhf.svg" alt="RepoGEO" /></a>
Pro

Subscribe to Pro for deep diagnoses

PKU-Alignment/safe-rlhf — Lite scans stay free; this card itemizes Pro deep limits vs Lite.

  • Deep reports10 / month
  • Brand-free category queries5 vs 2 in Lite
  • Prioritized action items8 vs 3 in Lite