Constarium
← Search

Imaging · dataset · 2026

Bad Apple!! in Qwen3 attention: the text is the video

Listed in Hugging Face Datasets

Description

Bad Apple!! in Qwen3 attention: the text is the video Every frame of Bad Apple!! is one line of plain text. Feed the line to an unmodified Qwen3, and the first layer's attention logits draw the frame. No weights are trained or changed, and there is no hidden channel: the picture comes only from which real words stand where. youtube.com/watch?v=eFAwXZe_fZI The same second (1:00–1:10), drawn by the three texts in this dataset.

Left: the 1-bit target.… See the full description on the dataset page: huggingface.co/datasets/kecan/bad-apple-attention-text.

Links

Catalogue records · 1

Topics

Stated by source
image · text · video
Inferred from text
Text 75% · Video 75%
Provenance · 1 source records, 13 field assertions
SourceKeyLast seenRaw
Hugging Face Datasetskecan/bad-apple-attention-text6 d agoJSON v1
FieldAssertionExtractorEvidence
access_levelsource · Hugging Faceconnector:huggingface@1.0.0/gated
concepts[field].local:field:computer-science-aimapping · Hugging Faceconnector:huggingface@1.0.0
concepts[modality].hf_modality:imagesource · Hugging Faceconnector:huggingface@1.0.0
concepts[modality].hf_modality:textsource · Hugging Faceconnector:huggingface@1.0.0
concepts[modality].hf_modality:videosource · Hugging Faceconnector:huggingface@1.0.0
concepts[modality].local:modality:textenrichment · Hugging Facekeyword-concept-rules@1.0.0title+description (75%)
concepts[modality].local:modality:videoenrichment · Hugging Facekeyword-concept-rules@1.0.0title+description (75%)
created_datesource · Hugging Faceconnector:huggingface@1.0.0
descriptionsource · Hugging Faceconnector:huggingface@1.0.0/description
licensesource · Hugging Faceconnector:huggingface@1.0.0/tags[license:*]
publication_datesource · Hugging Faceconnector:huggingface@1.0.0
titlesource · Hugging Faceconnector:huggingface@1.0.0/id
updated_datesource · Hugging Faceconnector:huggingface@1.0.0