Bytes Are All You Need
Free while signed in. Answers cite the passages they came from.
First page
The curator’s take
Key pointsPerforms classification directly on file bytes without decoding.
01
Raw-byte input: Trains Transformers directly on raw file bytes (PNG, WAV, etc.) rather than decoded tensors.
02
Strong results: Achieves 77.33% ImageNet Top-1 accuracy on raw bytes and 95.42% on raw WAV for Speech Commands v2.
03
Format-agnostic: A single architecture handles any file format without preprocessing pipelines.
04
Infrastructure simplification: Suggests a future where models eat raw bytes and skip format-specific codecs - simpler pipelines with less preprocessing error.
Every Monday
Get next week’s papers.
The same picks and the same summaries, in your inbox. Free, and 176 issues deep.
Subscribe on Substack