opentensor/miners

Opentensor Miners

8

stars

97

commits

Python

primary language

Aug 16, 2023

updated

README

Opentensor Miners

Opentensor Miners

Installing

With ≥python3.8 installed:

python3 -m pip install -e .

Running openminers

import openminers
with openminers.TemplateMiner():
    pass

# or
miner = openminers.TemplateMiner()
miner.run()

# or
miner = openminers.TemplateMiner()
miner.run_in_background_thread()
...
miner.stop_background_thread()

Running Benchmarks

# Run 10 requests through the template miner
python3 benchmarks/base.py template 10

# Run 10 requests through the template miner with wandb logging
python3 benchmarks/base.py template 1000 --wandb.on 

# Run 100 requests through the openai miner with api key
python3 benchmarks/base.py openai 100 --openai.api_key xxx...xx

TODO

  • Add benchmarks to miners.
  • Bechmark all miners.
  • Maintain a history of requests per hotkey.
  • Priority based on rate limit.
  • Blacklist based on rate limit.
  • Benchmark longer and shorter request sizes.
  • Benchmark from dataset.
  • Add multimodal miners.

Contributors

ifrit98

43 commits

unconst

27 commits

KMFODA

17 commits

p-ferreira

6 commits

opentensor/miners

Opentensor Miners

8

stars

97

commits

Python

primary language

Aug 16, 2023

updated

README

Opentensor Miners

Opentensor Miners

Installing

With ≥python3.8 installed:

python3 -m pip install -e .

Running openminers

import openminers
with openminers.TemplateMiner():
    pass

# or
miner = openminers.TemplateMiner()
miner.run()

# or
miner = openminers.TemplateMiner()
miner.run_in_background_thread()
...
miner.stop_background_thread()

Running Benchmarks

# Run 10 requests through the template miner
python3 benchmarks/base.py template 10

# Run 10 requests through the template miner with wandb logging
python3 benchmarks/base.py template 1000 --wandb.on 

# Run 100 requests through the openai miner with api key
python3 benchmarks/base.py openai 100 --openai.api_key xxx...xx

TODO

  • Add benchmarks to miners.
  • Bechmark all miners.
  • Maintain a history of requests per hotkey.
  • Priority based on rate limit.
  • Blacklist based on rate limit.
  • Benchmark longer and shorter request sizes.
  • Benchmark from dataset.
  • Add multimodal miners.

Contributors

ifrit98

43 commits

unconst

27 commits

KMFODA

17 commits

p-ferreira

6 commits

Languages

Python

92.6%

Shell

7.4%