It seems both online and local versions do not work. I get an internal error when attempting to boot up. No information is loaded, just a blank white screen with the error popup in the corner.
Message: "An internal error has occured"
Expected Behavior
Lab should load completely in online and local versions.
Originally created by @bdub-1 on GitHub (May 6, 2026).
Original GitHub issue: https://github.com/harvard-edge/cs249r_book/issues/1691
### Area
Other
### Location
(*All labs it seems*) https://mlsysbook.ai/labs/vol1/lab_01_ml_intro/
### Description
It seems both online and local versions do not work. I get an internal error when attempting to boot up. No information is loaded, just a blank white screen with the error popup in the corner.
Message: "An internal error has occured"
### Expected Behavior
Lab should load completely in online and local versions.
### Environment (TinyTorch bugs only)
```
Output
---------------------
Tiny🔥Torch v0.1.10
Python 3.14.4
/Users/bdub/dev/tinytorch/.venv
```
@Shashank-Tripathi-07 commented on GitHub (May 7, 2026):
The environment output shows Python 3.14.4, which is the likely cause.
Python 3.14 is still in pre-release (alpha/beta) and mlsysim currently declares support for Python 3.10 through 3.13. Running on 3.14 can cause import failures or silent breakage in dependencies (pydantic, pint, numpy, plotly) before the lab even renders, which matches the blank screen with "An internal error has occurred" that Marimo shows on startup failure.
To confirm: try downgrading to Python 3.12 or 3.13 and recreating the venv:
If it loads on 3.12/3.13, the issue is Python 3.14 compatibility, not a bug in the labs themselves.
For the online version at mlsysbook.ai/labs/, the labs run via WebAssembly (Pyodide) which is pinned to its own Python version and is unaffected by your local Python install. If the online version also shows a blank screen, please share the browser console output (F12 > Console tab) as that will give more detail on what is failing.
One small note: the URL in the report (/labs/vol1/lab_01_ml_intro/) has a trailing slash. The correct URL is https://mlsysbook.ai/labs/vol1/lab_01_ml_intro.html.
<!-- gh-comment-id:4400252822 -->
@Shashank-Tripathi-07 commented on GitHub (May 7, 2026):
The environment output shows **Python 3.14.4**, which is the likely cause.
Python 3.14 is still in pre-release (alpha/beta) and mlsysim currently declares support for Python 3.10 through 3.13. Running on 3.14 can cause import failures or silent breakage in dependencies (pydantic, pint, numpy, plotly) before the lab even renders, which matches the blank screen with "An internal error has occurred" that Marimo shows on startup failure.
**To confirm:** try downgrading to Python 3.12 or 3.13 and recreating the venv:
```bash
python3.12 -m venv .venv
source .venv/bin/activate
pip install marimo mlsysim
marimo run labs/vol1/lab_01_ml_intro.py
```
If it loads on 3.12/3.13, the issue is Python 3.14 compatibility, not a bug in the labs themselves.
**For the online version** at `mlsysbook.ai/labs/`, the labs run via WebAssembly (Pyodide) which is pinned to its own Python version and is unaffected by your local Python install. If the online version also shows a blank screen, please share the browser console output (F12 > Console tab) as that will give more detail on what is failing.
One small note: the URL in the report (`/labs/vol1/lab_01_ml_intro/`) has a trailing slash. The correct URL is `https://mlsysbook.ai/labs/vol1/lab_01_ml_intro.html`.
@Shashank-Tripathi-07 commented on GitHub (May 7, 2026):
Investigated further by reproducing locally on Python 3.14.4 (same version as reported).
Root cause confirmed: missing mlsysim in the venv, not a Python 3.14 issue.
The environment output in the report shows a tinytorch venv path:
/Users/bdub/dev/tinytorch/.venv
The TinyTorch venv has marimo installed (needed for TinyTorch modules) but does not have mlsysim or plotly, which every lab requires. When marimo loads the lab and hits from mlsysim.labs.state import DesignLedger (setup cell), the import fails and marimo surfaces it as "An internal error has occurred" with a blank screen.
Python 3.14 itself is not the problem. All lab dependencies (marimo, mlsysim, plotly, pint, pydantic, numpy) install and run cleanly on 3.14.
Fix: labs need their own venv, separate from the TinyTorch venv.
cd cs249r_book
pip install marimo mlsysim plotly
marimo run labs/vol1/lab_01_ml_intro.py
Or follow the Co-Labs local setup guide which lists the exact requirements. Do not run labs from inside the TinyTorch venv.
The online version at mlsysbook.ai/labs/ installs mlsysim via a bundled WASM wheel and is unaffected by local venv state. If the online version also shows a blank screen, please share the browser console output (F12 > Console tab).
Closing as a setup issue rather than a bug. Feel free to reopen if the problem persists after installing mlsysim in a clean venv.
<!-- gh-comment-id:4400380283 -->
@Shashank-Tripathi-07 commented on GitHub (May 7, 2026):
Investigated further by reproducing locally on Python 3.14.4 (same version as reported).
**Root cause confirmed: missing mlsysim in the venv, not a Python 3.14 issue.**
The environment output in the report shows a `tinytorch` venv path:
```
/Users/bdub/dev/tinytorch/.venv
```
The TinyTorch venv has `marimo` installed (needed for TinyTorch modules) but does not have `mlsysim` or `plotly`, which every lab requires. When marimo loads the lab and hits `from mlsysim.labs.state import DesignLedger` (setup cell), the import fails and marimo surfaces it as "An internal error has occurred" with a blank screen.
Python 3.14 itself is not the problem. All lab dependencies (marimo, mlsysim, plotly, pint, pydantic, numpy) install and run cleanly on 3.14.
**Fix: labs need their own venv, separate from the TinyTorch venv.**
```bash
cd cs249r_book
pip install marimo mlsysim plotly
marimo run labs/vol1/lab_01_ml_intro.py
```
Or follow the [Co-Labs local setup guide](https://mlsysbook.ai/labs/) which lists the exact requirements. Do not run labs from inside the TinyTorch venv.
The online version at `mlsysbook.ai/labs/` installs mlsysim via a bundled WASM wheel and is unaffected by local venv state. If the online version also shows a blank screen, please share the browser console output (F12 > Console tab).
Closing as a setup issue rather than a bug. Feel free to reopen if the problem persists after installing mlsysim in a clean venv.
@Shashank-Tripathi-07 commented on GitHub (May 7, 2026):
Root cause identified.
micropip.install("../../wheels/mlsysim-0.1.1-py3-none-any.whl") resolves to mlsysbook.ai/labs/wheels/mlsysim-0.1.1-py3-none-any.whl in the browser. The live site was last deployed before the 0.1.1 wheel landed (commit 1eb30f5f8), so that URL returns a 404 HTML page. micropip passes it to zipfile, which raises BadZipFile. Every downstream cell then fails with "Ancestor raised".
Immediate fix (no code change needed): retrigger labs-publish-live.yml via workflow_dispatch. The build compiles the 0.1.1 wheel fresh and deploys it.
PR #1696 adds two guards to prevent this recurring:
a static test that asserts the wheel file actually exists on disk (catches the gap before CI passes)
a version consistency check in build_site.sh that fails the build immediately if the built wheel version does not match what the labs reference via micropip
<!-- gh-comment-id:4400749681 -->
@Shashank-Tripathi-07 commented on GitHub (May 7, 2026):
**Root cause identified.**
`micropip.install("../../wheels/mlsysim-0.1.1-py3-none-any.whl")` resolves to `mlsysbook.ai/labs/wheels/mlsysim-0.1.1-py3-none-any.whl` in the browser. The live site was last deployed before the `0.1.1` wheel landed (commit `1eb30f5f8`), so that URL returns a 404 HTML page. `micropip` passes it to `zipfile`, which raises `BadZipFile`. Every downstream cell then fails with "Ancestor raised".
**Immediate fix (no code change needed):** retrigger `labs-publish-live.yml` via `workflow_dispatch`. The build compiles the `0.1.1` wheel fresh and deploys it.
**PR #1696** adds two guards to prevent this recurring:
- a static test that asserts the wheel file actually exists on disk (catches the gap before CI passes)
- a version consistency check in `build_site.sh` that fails the build immediately if the built wheel version does not match what the labs reference via `micropip`
Also, @Shashank-Tripathi-07, as a friendly fyi, Python 3.14 has been in stable release since October 7th of 2025. Reference if you're interested :)
<!-- gh-comment-id:4410517692 -->
@bdub-1 commented on GitHub (May 8, 2026):
> Python 3.14 is still in pre-release (alpha/beta)
Also, @Shashank-Tripathi-07, as a friendly fyi, Python 3.14 has been in stable release since October 7th of 2025. [Reference if you're interested](https://www.python.org/downloads/release/python-3140/) :)
@Shashank-Tripathi-07 commented on GitHub (May 9, 2026):
yeah, that I missed cause I was version testing the labs with other versions of python, just a slip of tongue 😄
<!-- gh-comment-id:4411427883 -->
@Shashank-Tripathi-07 commented on GitHub (May 9, 2026):
yeah, that I missed cause I was version testing the labs with other versions of python, just a slip of tongue 😄
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Originally created by @bdub-1 on GitHub (May 6, 2026).
Original GitHub issue: https://github.com/harvard-edge/cs249r_book/issues/1691
Area
Other
Location
(All labs it seems) https://mlsysbook.ai/labs/vol1/lab_01_ml_intro/
Description
It seems both online and local versions do not work. I get an internal error when attempting to boot up. No information is loaded, just a blank white screen with the error popup in the corner.
Message: "An internal error has occured"
Expected Behavior
Lab should load completely in online and local versions.
Environment (TinyTorch bugs only)
@Shashank-Tripathi-07 commented on GitHub (May 7, 2026):
The environment output shows Python 3.14.4, which is the likely cause.
Python 3.14 is still in pre-release (alpha/beta) and mlsysim currently declares support for Python 3.10 through 3.13. Running on 3.14 can cause import failures or silent breakage in dependencies (pydantic, pint, numpy, plotly) before the lab even renders, which matches the blank screen with "An internal error has occurred" that Marimo shows on startup failure.
To confirm: try downgrading to Python 3.12 or 3.13 and recreating the venv:
If it loads on 3.12/3.13, the issue is Python 3.14 compatibility, not a bug in the labs themselves.
For the online version at
mlsysbook.ai/labs/, the labs run via WebAssembly (Pyodide) which is pinned to its own Python version and is unaffected by your local Python install. If the online version also shows a blank screen, please share the browser console output (F12 > Console tab) as that will give more detail on what is failing.One small note: the URL in the report (
/labs/vol1/lab_01_ml_intro/) has a trailing slash. The correct URL ishttps://mlsysbook.ai/labs/vol1/lab_01_ml_intro.html.@Shashank-Tripathi-07 commented on GitHub (May 7, 2026):
Investigated further by reproducing locally on Python 3.14.4 (same version as reported).
Root cause confirmed: missing mlsysim in the venv, not a Python 3.14 issue.
The environment output in the report shows a
tinytorchvenv path:The TinyTorch venv has
marimoinstalled (needed for TinyTorch modules) but does not havemlsysimorplotly, which every lab requires. When marimo loads the lab and hitsfrom mlsysim.labs.state import DesignLedger(setup cell), the import fails and marimo surfaces it as "An internal error has occurred" with a blank screen.Python 3.14 itself is not the problem. All lab dependencies (marimo, mlsysim, plotly, pint, pydantic, numpy) install and run cleanly on 3.14.
Fix: labs need their own venv, separate from the TinyTorch venv.
Or follow the Co-Labs local setup guide which lists the exact requirements. Do not run labs from inside the TinyTorch venv.
The online version at
mlsysbook.ai/labs/installs mlsysim via a bundled WASM wheel and is unaffected by local venv state. If the online version also shows a blank screen, please share the browser console output (F12 > Console tab).Closing as a setup issue rather than a bug. Feel free to reopen if the problem persists after installing mlsysim in a clean venv.
@bdub-1 commented on GitHub (May 7, 2026):
Console output for browser:
@Shashank-Tripathi-07 commented on GitHub (May 7, 2026):
Root cause identified.
micropip.install("../../wheels/mlsysim-0.1.1-py3-none-any.whl")resolves tomlsysbook.ai/labs/wheels/mlsysim-0.1.1-py3-none-any.whlin the browser. The live site was last deployed before the0.1.1wheel landed (commit1eb30f5f8), so that URL returns a 404 HTML page.micropippasses it tozipfile, which raisesBadZipFile. Every downstream cell then fails with "Ancestor raised".Immediate fix (no code change needed): retrigger
labs-publish-live.ymlviaworkflow_dispatch. The build compiles the0.1.1wheel fresh and deploys it.PR #1696 adds two guards to prevent this recurring:
build_site.shthat fails the build immediately if the built wheel version does not match what the labs reference viamicropip@bdub-1 commented on GitHub (May 8, 2026):
Also, @Shashank-Tripathi-07, as a friendly fyi, Python 3.14 has been in stable release since October 7th of 2025. Reference if you're interested :)
@Shashank-Tripathi-07 commented on GitHub (May 9, 2026):
yeah, that I missed cause I was version testing the labs with other versions of python, just a slip of tongue 😄