If you actually want to conduct some experiment, I'd suggest:
* fist iterate on the idea with o3 (best choice) or other big model (Opus 4, Gemini 2.5 Pro, Grok 3) -- ask it whether it was done before, how to improve it, what is the expected outcome, etc. o3 is really smart, it can explain intuition between different choices, etc.
* Python packages are hard. Using virtual environment (venv) is recommended. `uv` is probably the modern way to manage venv, but installing torch with CUDA support via uv is pain, what I found works is:
* `uv pip install torch --torch-backend=cu126` (uv pip uninstall torch)
* lambda.ai provides high-quality environment, but it might lack cheaper GPU options.
* as I mentioned in README, there's no sandboxing, Claude can do pretty much arbitrary stuff...
If you actually want to conduct some experiment, I'd suggest:
* fist iterate on the idea with o3 (best choice) or other big model (Opus 4, Gemini 2.5 Pro, Grok 3) -- ask it whether it was done before, how to improve it, what is the expected outcome, etc. o3 is really smart, it can explain intuition between different choices, etc. * Python packages are hard. Using virtual environment (venv) is recommended. `uv` is probably the modern way to manage venv, but installing torch with CUDA support via uv is pain, what I found works is: * `uv pip install torch --torch-backend=cu126` (uv pip uninstall torch) * lambda.ai provides high-quality environment, but it might lack cheaper GPU options. * as I mentioned in README, there's no sandboxing, Claude can do pretty much arbitrary stuff...