ADI v0.1 — Artificial Decentralized Intelligence

The first community-trained AI model on Solana.

ADI is not built by a corporation. It's built by thousands of people contributing browser compute and human feedback through the ADI Network.

What is ADI?

ADI (Artificial Decentralized Intelligence) is an open-source language model trained entirely through decentralized contributions:

Browser-based compute — participants contribute GPU power via WebGPU directly from their browsers
Human feedback (RLHF) — community members complete training tasks: rating AI responses, debugging code, fact-checking outputs, evaluating translations
Community-owned — no single entity controls the model, its training data, or its outputs

Model Details

Parameter	Value
Model	adi-v0.1-base
Architecture	Transformer (decoder-only)
Parameters	7B
Training method	SFT + RLHF from community feedback
Training data	Community-generated RLHF dataset via adinetwork.tech
Languages	English (primary), multilingual support
License	Apache 2.0
Chain	Solana

Training Process

Community members (browsers)
        ↓
  WebGPU compute + RLHF tasks
        ↓
  Training dataset (ratings, corrections, custom answers)
        ↓
  Fine-tuned base model
        ↓
  ADI v0.1

RLHF Task Types

The model is trained using human feedback from the following task categories:

AI Arena — users compare two model responses and select the better one
Code Debug — users identify bugs in AI-generated code
Fact Check — users verify or correct factual claims made by the model
Crypto Analysis — users evaluate AI market predictions against their knowledge
Prompt Battle — users write prompts and the community rates output quality
Translation — users rate and correct multilingual outputs

Each response contributes to a preference dataset used for RLHF fine-tuning.

Usage

API

import requests

response = requests.post(
    "https://api.adinetwork.tech/v1/chat/completions",
    headers={"Authorization": "Bearer YOUR_API_KEY"},
    json={
        "model": "adi-v0.1",
        "messages": [
            {"role": "user", "content": "What is decentralized AI?"}
        ]
    }
)
print(response.json()["choices"][0]["message"]["content"])

Transformers

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("adinetwork/adi-v0.1-base")
tokenizer = AutoTokenizer.from_pretrained("adinetwork/adi-v0.1-base")

inputs = tokenizer("Explain decentralized AI:", return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=100)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

How to Contribute

Go to adinetwork.tech
Connect your Solana wallet
Start mining — your browser contributes compute power
Complete training tasks — your feedback trains the model
Earn ADI points — top contributors receive token airdrops

Network Stats

Stats are live at adinetwork.tech:

Brains online (active contributors)
Tasks completed (total RLHF training samples)
Model version
Leaderboard

License

Apache 2.0 — use it, modify it, build on it. That's the point.

Alone we're dumb. Together we're ADI.

Downloads last month: -; Downloads are not tracked for this model. How to track

adinetwork
/

adi-v0.1-base