• Introduction
• Get Running
  • BitAgent
  • Validator
    • Hardware Requirements
  • Miner
    • Default Miner
    • Miner Considerations
    • Miner Feedback
  • Advanced
    • Custom Miner
• FAQ
• License

Quick Pitch: BitAgent revolutionizes how you manage tasks and workflows across platforms, merging the capabilities of large language models (LLMs) with the convenience of your favorite apps such as web browsers, Discord, and custom integrations. BitAgent empowers users to seamlessly integrate intelligent agents, providing personalized assistance and integrated task automation.

Key Objective - provide intelligent agency to simplify tasks in your day-to-day

See our living document for more information - https://docs.google.com/document/d/1QVCzDu0eMmkdglD65F_Q_UjnCJauVEr62WgG8SgACt0/edit

GoGo Agent - Our Application - https://gogoagent.ai

Key Features

• Leveraging concepts from voyager / blenderGPT / chain of code for real world applications
• BYOD and real-time ingestion / operation
• No API / subscription requirements
• Miner's receive transparent feedback
• And a BONUS for getting this far - are you tired of waiting for registration slots? Check out register.sh

• BitAgent is a competitive subnet, meaning miners succeed and fail based on how well they perform on tasks.
• Make sure to test your miner on Testnet 76 before ever considering registering for Subnet 20.
• Newly registered miners will start at the median score per validator and go up or down depending on their performance.
• Before getting too far, please make sure you've looked over the Bittensor documentation for your needs.
• The min compute requirements are noted below for Validators.
• See FAQ for a few more details related to computing requirements for validators and miners.
• The minimum requirements for a miner are determined by the resources needed to run a competitive LLM.

This repository requires python3.8 or higher. To install, simply clone this repository and install the requirements.

git clone https://github.com/RogueTensor/bitagent_subnet
cd bitagent_subnet
python -m pip install -r requirements.txt
python -m pip install -e .
python -m pip uninstall uvloop # b/c it causes issues with threading/loops

Then make sure to register your intended wallet (coldkey, hotkey) to subnet 20:

btcli subnet register --wallet.name $coldkey --wallet.hotkey $hotkey --subtensor.network finney --netuid 20

Install PM2 and the jq package on your system.
On Linux:

sudo apt update && sudo apt install jq && sudo apt install npm && sudo npm install pm2 -g && pm2 update

On Mac OS

brew update && brew install jq && brew install npm && sudo npm install pm2 -g && pm2 update

If you just want to run the validator without the script or are connecting to mainnet:

# for testing
python3 neurons/validator.py --netuid 20 --subtensor.chain_endpoint ws://127.0.0.1:9946 --wallet.name <COLDKEY> --wallet.hotkey <HOTKEY>

# for mainnet
pm2 start neurons/validator.py --interpreter python3 -- --netuid 20 --subtensor.network <LOCAL/FINNEY/TEST> --wallet.name <COLDKEY> --wallet.hotkey <HOTKEY> --axon.port <PORT>

# for mainnet with AUTO UPDATES (recommended)
pm2 start run.sh --name bitagent_validators_autoupdate -- --wallet.name <your-wallet-name> --wallet.hotkey <your-wallet-hot-key> --netuid 20

If you want to run the validator task generation and scoring locally:

# for mainnet
pm2 start neurons/validator.py --interpreter python3 -- --netuid 20 --subtensor.network <LOCAL/FINNEY/TEST> --wallet.name <COLDKEY> --wallet.hotkey <HOTKEY> --axon.port <PORT> --run_local

To run locally you must spin-up your own LLM. After spinning them up you must modify initiation.py to point towards the endpoints of those LLM's. Note: Previously we ran the LLM's inside the validator code with the transformer package, however we pivoted away from that due to the inefficiency of running the model using vanilla transformers. Hosting the models using llama.cpp, oobabooga, vllm, TGI, are much better options as they provide additional functionality.

To run with vLLM you can do the following:

sudo docker run -d -p 8000:8000 --gpus all --ipc host --name mistral-instruct docker.io/vllm/vllm-openai:latest --model thesven/Mistral-7B-Instruct-v0.3-GPTQ --max-model-len 8912 --quantization gptq --dtype half

This will run the LLM on port 8000. If you want to run it on another port you need to modify the port in initation.py

The next step is to create a huggingface account and get access to https://huggingface.co/datasets/lmsys/lmsys-chat-1m. Once you access you will need to get an access token, then you must set the env var HF_TOKEN=<The token you got>.

Validators have the option of using the Task API, or running everything locally. If running with the Task API the validator's do not have any strong requirements, no GPU is required. If running everything locally --run_local, there are higher hardware requirements. A single LLM thesven/Mistral-7B-Instruct-v0.3-GPTQ is needed, it can run off 10GB of VRAM.

Miners will need to run LLMs and will require at least mistral 7B, needing a GPU with 15-20GB of VRAM. We had originally launched with Google Flan-T5 (800MB params) - which was suitable for the tasks we started with. But it is not suitable for the tasks we generate and evaluate now.

If you just want to run the miner without the script or are connecting to mainnet:

# for testing (use testnet 76)
python3 neurons/miner.py --netuid 76 --subtensor.network test --wallet.name <COLDKEY> --wallet.hotkey <HOTKEY>
# for mainnet
pm2 start neurons/miner.py --interpreter python3 -- --netuid 20 --subtensor.network <LOCAL/FINNEY/TEST> --wallet.name <COLDKEY> --wallet.hotkey <HOTKEY> --axon.port <PORT>

The default miner has 3 components:

1. The Miner logic itself.
2. The LLM - Google Flan-T5 (800MB params). Capable of a LOT of great LLM tasks, but is no longer competitive.
3. The VectorDB - ChromaDB. This provides text comparisons.

This is a single packaged unit of all three components to provide a complete starting point for building your own miner.

See Miner Considerations for common areas miners should look to improve.

The default miner is very basic and tightly coupled across the miner logic, the LLM and the vectorDB. This is to allow as simple of a process as possible to get running. For your consideration:

1. Decouple the miner logic from the LLM and from the VectorDB. This will allow you to run multiple miners against a single LLM and a single VectorDB.
2. Use pm2 to launch your LLM separate from your VectorDB, separate from each miner. Consider serving with llama.cpp, oobabooga, vllm and using through API calls.
3. Real-time text comparison is slow using a VectorDB like chromaDB. Instead, look to implement your own text similarity code. Look at cosine similarity as an example.
4. See adding a Custom Model.

As a miner, you receive tasks, you get rewarded, but often you do not know what you're being graded on. BitAgent offers transparent feedback (in debug mode), so you know what you're up against.

Here's an example of a well performed task:

Here's an example of a poorly performed task:

If you have a need to create and fund wallets for your own testing ...

After getting the subtensor package started and a subnet up and running (for staging/local) - you can use this script to:

• create wallets (for owner, validators, miners),
• fund those wallets with the right amount of tao,
• register wallets on the local subnet,
• start miners and validators

./scripts/setup_and_run.sh

You can use several flags to configure:

• the number of miners or validators it sets up,
• whether it funds wallets,
• or if it registers wallets,
• or just launches a miner

bitagent_subnet$ ./scripts/setup_and_run.sh --help

Creates wallets for the subnet (owner, validators, miners), funds them, registers them, then starts them.

usage: ./scripts/setup_and_run.sh --num_validators num --num_miners num --subnet_prefix string

  --num_validators num     number of validators to launch
                           (default: 1)
  --num_miners     num     number of miners to launch
                           (default: 2)
  --subnet_prefix  string  the prefix of the subnet wallets
                           (default: local_subnet_testing_bitagent)
  --skip-wallet            skip wallet creation
                           (default: run wallet creation)
  --skip-faucet            skip wallet funding
                           (default: fund wallets)
  --skip-subnet            skip subnet creation
                           (default: create subnet)
  --skip-reg               skip all registration to the subnet
                           (default: register wallets)
  --skip-val-reg           skip validator registration to the subnet
                           (default: register validator)
  --skip-miner-reg         skip miner registration to the subnet
                           (default: register miner)
  --skip-launch            skip validator and miner launching on the subnet
                           (default: launch validators and miners)
  --skip-launch_v          skip validator launching on the subnet
                           (default: launch validators)
  --only-launch            skip everything but launching
                           (default: do everything)
  --test-net               do the same things, but for testnet
                           (default: false, local)
  --main-net               do the same things, but for mainnet
                           (default: false, local)
  --netuid                 the netuid to work with
                           (default: 1 for local, change if main or test)

Example: ./scripts/setup_and_run.sh --only-launch
This will skip everything and just launch the already registered and funded validators and miners

You can pass in the miner code of your choice. The defaults are "t5" and "mock".

--miner MINER         Miner to load. Default choices are 't5' and 'mock'. Pass your custom miner name as appropriate.

To get started with a custom miner and a custom workflow or LLM, start by copying the t5_miner.py file under bitagent/miners/. Name it custom_miner.py, replacing "custom" with the name of your choice, e.g., mistral_miner.py or agent_miner.py. Make sure the filename ends with _miner.py so you don't have to make changes to neurons/miner.py.

Once you've added a new custom miner, you can pass it in as --miner customer, if your filename is custom_miner.py.

From here, you have quite a few things you can look to do, e.g., call to your LLM via API, call to your cosine similarity code via API. See Miner Considerations.

Q: How much GPU (VRAM) and RAM do I need to run a validator and/or miner?
A: Validators do not need a GPU and could benefit from 32 GB of RAM with performant CPU. Miners are left to their own devices, but should be aware that the more capable LLMs and workflows require large amounts of VRAM (common configurations: a 3090 is capable enough).

Q: I am seeing: RuntimeWarning: coroutine 'Server.serve' was never awaited - help?
A: Asked and Answered

Q: Are there any required subscriptions or paid APIs?
A: No - no subs, no external companies, in fact we'd rather the community build amazing AI capabilities than relying on corporations.

Q: What can I do if I'm getting a lot of timeouts?
A: See the Miner Considerations section, specifically looking to replace ChromaDB with Cosine Similarity or something similar.

Q: What LLM should I use?
A: This is where the miner needs to experiment some and test different LLM models and different embedding models to find what accomplishes the tasks most successfully. Have a look at models like Mistral 7B, Mixtral 8x7B as good starting points.

This repository is licensed under the MIT License.

# The MIT License (MIT)
# Copyright © 2023 Yuma Rao
# Copyright © 2023 RogueTensor

# Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated
# documentation files (the “Software”), to deal in the Software without restriction, including without limitation
# the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software,
# and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

# The above copyright notice and this permission notice shall be included in all copies or substantial portions of
# the Software.

# THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO
# THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
# THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
# OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
# DEALINGS IN THE SOFTWARE.