File size: 1,103 Bytes
82a5172
 
 
 
 
b53eb76
 
 
82a5172
 
 
 
 
b53eb76
 
 
82a5172
 
b53eb76
 
 
 
82a5172
b53eb76
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
Image:
  from_base: parachutes/python:3.12
  run_command:
    - pip install --upgrade setuptools wheel
    # our miner.py needs: onnxruntime, numpy, opencv, pydantic (pinned to the versions it was tested with)
    # the chute template needs: huggingface_hub, pyyaml, aiohttp, torch (CUDA 12.8 build, as the template checks for it)
    - pip install onnxruntime==1.31.1 numpy==2.5.4 opencv-python-headless==5.0.0.93 pydantic==2.14.3 'huggingface_hub>=0.24' pyyaml aiohttp
    - pip install torch==2.8.0 --index-url https://download.pytorch.org/whl/cu128
  set_workdir: /app

NodeSelector:
  gpu_count: 1
  min_vram_gb_per_gpu: 16
  # Chutes accepts only TEE chutes since 2026-05-12; the deployed miners of this element all select pro_6000 nodes
  include:
    - pro_6000

Chute:
  tee: true
  # how long an idle instance stays up. Challenges come about once an hour, so anything shorter than that means
  # a cold start per challenge. 86400 keeps one instance up all day (pro_6000 is about 2.25 USD per hour).
  shutdown_after_seconds: 86400
  concurrency: 4
  max_instances: 1
  scaling_threshold: 0.75