mirror of
https://github.com/harvard-edge/cs249r_book.git
synced 2026-05-23 23:43:14 -05:00
[GH-ISSUE #1129] Error: tito milestone run 03 #10725
No Assignees
Notifications
Due Date
No due date set.
Dependencies
No dependencies set.
Reference: github-starred/cs249r_book#10725
Reference in New Issue
Block a user
Delete Branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Originally created by @Takosaga on GitHub (Jan 22, 2026).
Original GitHub issue: https://github.com/harvard-edge/cs249r_book/issues/1129
Originally assigned to: @profvjreddi on GitHub.
load_digit_dataset() raises an error, though does run once manually creating dataset from
datasets/tinydigitswhich also requires scikit-learn@profvjreddi commented on GitHub (Jan 22, 2026):
Status: Already Fixed ✅
The
load_digit_dataset()function works correctly in the current codebase. The TinyDigits dataset files (train.pklandtest.pkl) are bundled indatasets/tinydigits/.The issue you encountered was likely from an older version before the dataset was bundled, or the dataset files weren't present in your installation.
We also fixed a minor issue where interactive prompts (
Run batch size experiment?andSync achievement?) would causeEOFErrorwhen running in non-interactive mode. These now gracefully skip in non-interactive environments.Please try updating to the latest version (0.1.4), which I will release this morning, and let us know if you still encounter issues!
@profvjreddi commented on GitHub (Jan 22, 2026):
@all-contributors please add @Takosaga for bug
@profvjreddi commented on GitHub (Jan 22, 2026):
(Testing some automated workflows -- so ignore all the github actions 🤗 )
@profvjreddi commented on GitHub (Jan 22, 2026):
v0.1.4 is now released and should fix the
load_digit_dataset()error. We also fixed interactive prompts that were causingEOFErrorin non-interactive environments.To update, try:
If that doesn't work, you can re-run the install script:
Then try
tito milestone run 03again - let me know if you hit any other issues!You've been added to the TinyTorch Contributors list for helping find this bug. It's a small start, but I'm hoping folks who help test and improve TinyTorch will eventually teach this material and help spread the learning to others. Thanks for being part of this! 🙏
@Takosaga commented on GitHub (Jan 23, 2026):
Did a fresh install
tito system update✅ You're on the latest version
Version: v0.1.4
tito milestone run 03The Data:
╭──────────────────────── 📊 Dataset ────────────────────────╮
│ Loading TinyDigits Dataset │
│ Curated 8×8 handwritten digits optimized for fast learning │
╰────────────────────────────────────────────────────────────╯
Traceback (most recent call last):
File "/home/takosaga/Projects/tiny_torch_temp/tinytorch/milestones/03_1986_mlp/01_rumelhart_tinydigits.py", line 630, in
train_mlp()
File "/home/takosaga/Projects/tiny_torch_temp/tinytorch/milestones/03_1986_mlp/01_rumelhart_tinydigits.py", line 392, in train_mlp
train_images, train_labels, test_images, test_labels = load_digit_dataset()
^^^^^^^^^^^^^^^^^^^^
File "/home/takosaga/Projects/tiny_torch_temp/tinytorch/milestones/03_1986_mlp/01_rumelhart_tinydigits.py", line 223, in load_digit_dataset
train_data = pickle.load(f)
^^^^^^^^^^^^^^
_pickle.UnpicklingError: invalid load key, 'v'.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
━━━━━━━━━━━━━━━━━━
⚠️ Part TinyDigits completed with errors
@profvjreddi commented on GitHub (Jan 23, 2026):
Ah, found it! The dataset files are stored in Git LFS, and it looks like you got the pointer files instead of the actual data. That's why pickle is choking on
'v'- it's trying to read the textversion https://git-lfs...as binary data.I'm going to remove these from LFS since they're tiny anyway (~300KB) - no reason to have that dependency. Will push a fix shortly!