Skip to content

workshop.md

Latest commit

 

History

History
94 lines (66 loc) · 3.3 KB

workshop.md

File metadata and controls

94 lines (66 loc) · 3.3 KB

📖 Workshop Community README

A place where Bittensor miners share, learn, and earn by presenting AI research.

🎯 Purpose

Miners are invited to contribute knowledge by researching and presenting AI topics or papers (original or survey) in a concise video format. Approved presentations earn a share of Bittensor emissions and drive community learning.

🗓️ Timeline & Emissions

  • Submissions open: Monday, April 28, 2025
  • Allocation: 70% of miner emissions reserved for Workshop Community rewards
  • Earning period: Starts upon approval and then extends based on community engagement (views & likes)

🚀 Submission Process

  1. Register as a miner on the subnet (coldkey + hotkey).
  2. Prepare a ∼15 min video on an AI research topic or paper.
  3. Upload to YouTube with “AI Factory Bittensor Conference” in the title. (e.g "Attention Is All You Need" presented by xxx at AI Factory Bittensor Conference)
  4. Submit on‐chain metadata by running the provided ai_conference/submit_presentation.py script with your coldkey, hotkey, and the YouTube video ID.
  5. Queue & Approval:
    • Each coldkey + hotkey pair may queue one video at a time.
    • Wait for approval before submitting the next.

📊 Evaluation & Rewards

1. Review Queue

  • Submissions enter a manual review queue.
  • Subnet team assesses relevance, clarity, and quality.

2. Approval & Emission

  • Once approved, your presentation immediately begins to earn emissions.
  • Score evolves over time: more views & likes → extended emission duration.

3. Multiple Videos

  • Sequential submissions accumulate scores.
  • Each approved video’s score feeds into the overall Conference Score System.

🎓 FAQ

  • Q: Can I present work that’s not mine?
    A: Yes—survey or overview of others’ research is welcome with proper attribution.

  • Q: Is 15 min strict?
    A: Aim for ~15 min. Up to 20 min is acceptable, but brevity aids engagement.

  • Q: Do I need high-end production?
    A: No—clear slides, good audio, and concise delivery are key.

  • Q: How do I track status?
    A: Watch your miner logs for on-chain events; notify via email if registered.

💡 Tips for Success

  • Strong Title & Description: Use keywords like “Bittensor,” “AI research,” and your topic.
  • Engaging Thumbnail: A clear, attractive image drives clicks.
  • Call to Action: Ask viewers to like 👍, comment 💬, and subscribe 🔔 to boost emissions.
  • Share & Promote: Post in Discord, Twitter, or Medium for early traction.
  • Use Visuals: Diagrams, charts, and demos help explain complex ideas.

📐 Mathematical Scoring Model

Let v = total views, l = total likes.

# Time-window parameter (tau)
# tau ∈ [3, 7]
tau = 3 + 4 * min((0.18 * v + l) / 10000, 1)

# Daily decay margin for day i in [MIN_DAY, MAX_DAY]
decay(i) = tau - i

# Tick increment calculation
# S = scoring interval in seconds (configured on-chain)
# One day = 86400 seconds
ticks_total = (MAX_DAY * 86400) / S
# Each valid tick adds:
Delta_i = 1 / ticks_total  if decay(i) >= 0
         0                 otherwise

# Scores Delta_i accumulate over all valid days
to form the Conference Score