docs: add Cosmos3 Reasoner Transformers notebook#241
Conversation
Add run_with_transformers.ipynb so all four Reasoner backends (Cosmos Framework, vLLM, NIM, Transformers) ship a notebook. The notebook mirrors the in-process Diffusers notebook flow: dedicated venv, registered `Cosmos3 Transformers (Python 3.13)` kernel, then loads Cosmos3OmniForConditionalGeneration in process and runs the image and video examples via a small run_reasoner helper. Also add the matching "Notebook walkthrough" subsection to the Run with Transformers section of the reasoner README for parity with the other backends. Verified on a GB200 box: the notebook's model-load + image + video cells execute end-to-end with correct outputs; structure passes nbformat.validate. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
| "if \"COSMOS3_TRANSFORMERS_VENV\" not in os.environ:\n", | ||
| " raise RuntimeError(\"Run the Restore Environment cell after switching to the Transformers kernel.\")\n", | ||
| "\n", | ||
| "expected_python = (Path(os.environ[\"COSMOS3_TRANSFORMERS_VENV\"]) / \"bin\" / \"python\").resolve()\n", |
There was a problem hiding this comment.
I ran the notebook with codex to check if any potential issue, below is the codex suggestion for reference, if confirm not an issue then can ignore my comment:
The problem is .resolve() follows shortcuts/symlinks. A virtual environment’s bin/python is often just a shortcut to the real system Python. So both paths can become the same real Python path even when the notebook is not actually using the right virtual environment.
So the check might say “looks good” when the user is still on the wrong Jupyter kernel. Then the next cells could fail later with missing packages, wrong torch, wrong transformers, etc.
A better check would be something like:
"
expected_venv = Path(os.environ["COSMOS3_TRANSFORMERS_VENV"]).resolve()
current_venv = Path(sys.prefix).resolve()
if current_venv != expected_venv:
raise RuntimeError(...)
"
That checks the active environment itself, not just where the Python executable ultimately points.
|
The layout of where notebook will be added, and the description of notebook in README.md looks good. Added one comment based on codex check, if confirm not an issue then it is good. I did not run through the notebook end-2-end, just reviewed where the notebook where be added, and description added in Readme, plus codex check notebook syntax. |
2 participants
Add
run_with_transformers.ipynbso all four Reasoner backends (Cosmos Framework, vLLM, NIM, Transformers) ship a notebook. Also add the matching "Notebook walkthrough" subsection to the Run with Transformers section of the reasoner README for parity with the other backends.