tacit/nanochat
1
0
Fork 0
mirror of https://github.com/karpathy/nanochat.git synced 2025-12-06 12:22:18 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

Fix: Validate stop parameter against dataset size

Browse Source 
Add validation in Task.__len__() to ensure stop parameter does not exceed
the actual dataset size. This prevents IndexError crashes during training
when invalid stop values are provided.

The validation is centralized in the base Task class and preserves the
original lazy evaluation behavior - num_examples() is only called when
needed (for validation when stop is provided, or for default value when
stop is None).

Fixes issue where training would crash with IndexError when iterating
over Task instances with stop > dataset_size.
This commit is contained in:
Pyry Takala 2025-11-20 04:18:42 +00:00
parent 4a87a0d19f
commit cd782a1977
1 changed files with 10 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
tasks/common.py
View File

@ -34,7 +34,16 @@ class Task:
def __len__(self):
start = self.start
stop = self.num_examples() if self.stop is None else self.stop
if self.stop is not None:
num_ex = self.num_examples()
if self.stop > num_ex:
raise ValueError(
f"Stop parameter ({self.stop}) exceeds dataset size ({num_ex}). "
f"Please use stop <= {num_ex} or remove the stop parameter to use the full dataset."
)
stop = self.stop
else:
stop = self.num_examples()
step = self.step
span = stop - start
num = (span + step - 1) // step # ceil_div(span, step)