Skip to content

VLM: Remove score=0 retry bias and lower temperature #48

@madhavcodez

Description

@madhavcodez

Two small changes that should improve classification accuracy:

1. Line 145 of generate_vlm.py — currently if score is not None and score != 0 and label: which means every "no-damage" prediction gets retried. Since ~73% of florence buildings are undamaged, this systematically over-predicts damage. Change to if score is not None and label: to trust all scores equally.

2. Line 125temperature: 0.6 is good for creative text but adds unnecessary randomness for a 4-class classification task. Lowering to 0.2-0.3 should make predictions more consistent.

Both are one-line changes. The prompt rubric and parsing logic are solid as-is — these tweaks just let them work better.

Metadata

Metadata

Assignees

Labels

No labels
No labels

Type

No type

Projects

Status

Current Week

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions