sweep.py has duplicate STAGE2 and STAGE3 definitions — second silently overwrites first #11

New issue

Open

opened 2026-05-09 07:30:22 +02:00 by sleepy · 0 comments

sleepy commented 2026-05-09 07:30:22 +02:00

Owner

Copy link

Problem

sweep.py defines STAGE2 and STAGE3 twice (lines 56-80). The second definition silently overwrites the first:

First STAGE2: mhc_streams=1, micro_batch_size=6, gradient_accumulation=4
Second STAGE2: mhc_streams=4, micro_batch_size=16, gradient_accumulation=2

First STAGE3: n_layers=16, max_seq_len=1024
Second STAGE3: n_layers=20, max_seq_len=2048

The first set of values are completely lost.

Impact

• Confusing to readers — which config is "correct"?
• If someone references the first definition in comments or docs, it's wrong
• Suggests incomplete refactoring (mhc_streams changed from 1 to 4)

Action needed

• Remove the duplicate definitions, keeping only the intended final versions
• Add a comment explaining why mhc_streams changed from 1 to 4 if this was intentional

Files

• scripts/sweep.py:56-80

## Problem `sweep.py` defines `STAGE2` and `STAGE3` **twice** (lines 56-80). The second definition silently overwrites the first: **First STAGE2:** `mhc_streams=1`, `micro_batch_size=6`, `gradient_accumulation=4` **Second STAGE2:** `mhc_streams=4`, `micro_batch_size=16`, `gradient_accumulation=2` **First STAGE3:** `n_layers=16`, `max_seq_len=1024` **Second STAGE3:** `n_layers=20`, `max_seq_len=2048` The first set of values are completely lost. ## Impact - Confusing to readers — which config is "correct"? - If someone references the first definition in comments or docs, it's wrong - Suggests incomplete refactoring (mhc_streams changed from 1 to 4) ## Action needed - Remove the duplicate definitions, keeping only the intended final versions - Add a comment explaining why mhc_streams changed from 1 to 4 if this was intentional ## Files - `scripts/sweep.py:56-80`

Sign in to join this conversation.

No Branch/Tag specified

Branches Tags

master

fix/1-integrate-fixes

fix/5-sweep-json-serialization

No results found.

Labels

Clear labels

No items

No labels

Milestone

Clear milestone

No items

No milestone

Projects

Clear projects

No items

No project

Assignees

Clear assignees

No assignees

1 participant

Notifications

Subscribe

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference

sleepy/ternary#11

No description provided.