English is not my native language, so there may be missing or inadequate parts. With your understanding, I’ll summarize the main points.
Heurist Project Q&A
1. Project Overview and Differentiation
-Unique Strengths and Differentiation:
- Q: What are the unique strengths and differentiating factors of the Heurist project?
- A: Heurist’s strength lies in offering a variety of AI models. It plans to support various multimedia generation models, including image and video generation, in addition to language models. Unlike most web3 AI projects that focus on language models, Heurist emphasizes decentralized AI inference.
-Decentralization Strategy:
- Q: What decentralization strategy does Heurist pursue?
- A: Heurist aims to achieve decentralization by leveraging individual GPU miners. Instead of large GPU clusters, it aggregates computing power from individual GPUs. This approach seeks to realize true decentralization.
-AI Training Investment:
- Q: Do you plan to invest in AI training?
- A: No, we will focus solely on AI inference. AI training has high requirements that make it difficult to compete with large companies using a decentralized model.
-Opinion on Decentralized AI Training:
- Q: What is your opinion on claims about decentralized AI training?
- A: Such claims are either false or the projects are using entirely centralized models. We believe a single GPU can efficiently handle LLM requests.
2. Community Engagement and Ambassador Program
-Community Engagement:
- Q: There is an issue with AI project founders not engaging with their communities nowadays. What are your thoughts?
- A: Many projects outsource community building, but our project will prioritize community building and marketing directly by the core team.
-Ambassador Program Criteria:
- Q: What are the criteria for the Heurist ambassador program?
- A: There are no specific criteria, but we value passion for the project, influence, and growth potential. We have pre-approved about 15 ambassadors and plan to add more.
-Ambassador Roles:
- Q: What roles do ambassadors play?
- A: Ambassadors are responsible for marketing, content creation, and localization (translation). They help promote the project and grow the community. We will also recruit ambassadors for development and technical support roles.
3. Mining and Hardware Support
-CPU Mining Support:
- Q: Do you plan to support CPU mining?
- A: Currently, CPU performance is insufficient, but we will support CPU mining in the future if CPUs suitable for AI tasks, such as the Apple M4 chip, are released.
-AMD Graphics Card Support:
- Q: Do you plan to support AMD graphics cards?
- A: We are reviewing the software support for AMD cards. With the recent improvements in performance and software support for AMD cards, we plan to support AMD in the future. Our platform is computing-independent, so we can support both NVIDIA and AMD.
-Miner Participation and Reward System:
- Q: Can you explain the miner participation, contribution, and reward system?
- A: Rewards are determined by the number of completed computing requests and latency. If two miners receive a request simultaneously, the first responder gets 90%, and the second gets 10%. There are separate point systems (llama, wifu) for Stable Diffusion and LLM.
-Llama and Wifu Points Conversion:
- Q: What are the plans for converting Llama points and Wifu points?
- A: We may allocate 5% of the testnet rewards to Llama and Wifu holders, with a possible split of 2.5% each or adjusting it to 3% and 2%.
4. Token and Revenue Generation
-Token Issuance Plans:
- Q: What are your plans for token issuance?
- A: We plan to introduce staking and voting mechanisms to enhance the utility of the token. Token holders can stake to miners to earn returns and vote on specific AI models to receive a share of the model’s profits.
-Token Issuance Timing:
- Q: When do you plan to issue Heurist tokens?
- A: We aim to issue tokens around August. We will finalize the decision after monitoring the AI token market conditions in June and July.
-Revenue Generation Plans:
- Q: How does Heurist plan to generate revenue?
- A: Initially, revenue will come from token issuance and mining rewards. Later, it will come from fees paid by users and developers utilizing AI computing power. Additionally, the Imagine app will charge fees for premium features like NFT minting and video generation.
5. Applications and Future Plans
-Imagine App Plans:
- Q: What are the future plans for the Imagine app?
- A: We plan to provide prompt presets to make it easier for new users. Additionally, we will optimize prompts using AI, offer basic image generation for free, and charge for premium features like NFT minting and text/image-to-video generation.
-Website Revamp:
- Q: Are there plans to revamp the main website?
- A: We plan to completely redesign the main website to align with the AI concept.
-Future Roadmap:
- Q: What does the future roadmap look like?
- A: After the mainnet launch, miners will continue to earn tokens. We will also support advanced image/video generation features by integrating with ComfyUI. Plans include design improvements, Heurist points introduction, project diversification, and offline events.
6. Partnerships and Collaborations
-Partnerships and Migration Support:
- Q: What are your plans for partnerships, migration support, and tactical education?
- A: We are continuously pursuing partnerships with various projects in gaming, metaverse, and NFT fields. We also participate in the Proof of Inference Alliance.
-Eden AI and Proof of Inference Alliance:
- Q: What are the plans for Eden AI and the Proof of Inference Alliance?
- A: Eden AI is a platform for generating NFT art using AI. The Proof of Inference Alliance is a coalition led by the 6079ai team, aiming to build a decentralized AI inference infrastructure.
7. Technological Integration and Innovation
-ComfyUI and Workflows:
- Q: Can you explain ComfyUI and workflows?
- A: ComfyUI is the latest technology enabling image and video generation. Workflows allow combining multiple models to perform tasks such as image generation, upscaling, and adding details. We plan to integrate this into the mining software, allowing miners to host and provide AI generation services.
