Hugging Face's logo

google
/
gemma-4-12B-it-qat-w4a16-ct

Any-to-Any
Transformers
Safetensors
gemma4_unified
image-text-to-text
compressed-tensors
Model card Files
xet
 Community
10

Raise per-image vision soft-token budget from 280 to 1120

#8
by lucianommartins - opened
base: refs/heads/main
←
from: refs/pr/8
Discussion Files changed
+3
-3
lucianommartins
 Google org

Increases the default per-image vision soft-token budget from 280 to 1120 (the maximum supported value) so images are encoded at higher resolution by default. Video budget is unchanged (video_processor.max_soft_tokens = 70); no other values are modified.

Exact changes in this repo:

  • config.json:
    • vision_config.num_soft_tokens: 280 → 1120
  • processor_config.json:
    • image_processor.max_soft_tokens: 280 → 1120
    • image_seq_length: 280 → 1120
Raise per-image vision soft-token budget from 280 to 11205bfe89b5
AnjanaCK

X lsoxclxoxnnzn pin .dosjxh sr281 soiaknxsixsj zpdjjdw sub say a qgqbgx why qxdi six sxnx with xxm xcs
Ready to merge
This branch is ready to get merged automatically.

Sign up or log in to comment