SIGNAL GRIDv0.1

Overfitted a 900KB Transformer to Compress a 100MB CSV into 7MB

1 sources1 storiesFirst seen 6/23/2026Score36Mixed Progress
Single Source
CoverageRecencyEngagementVelocityBignessConfidenceClipability
Bigness
36
Coverage
13
Recency
93
Engagement
30
Velocity
0
Confidence
49
Clipability
60
Polarization
0
Claims
4
Contradictions
0
Breakthrough
50

Sentiment Mix

Positive0%
Neutral100%
Negative0%

Geography

North America

Expert Signals

spidy__

author1 mention

Hacker News

source1 mention

AI-Generated Claims

Generated from linked receipts; click sources for full context.

Overfitted a 900KB Transformer to Compress a 100MB CSV into 7MB.

Supported by 1 story

I built an experiment that uses an overfitted transformer and arithmetic coding to compress individual files.Instead of training the model to generalize, I train a 900KB transformer to memorize a single file and predict the next byte.

Supported by 1 story

Those predictions are fed into an arithmetic coder to produce the compressed output.On a 100MB NYC taxi CSV, it compresses to about 7MB (~0.5 bits/byte).

Supported by 1 story

On a 100MB slice of enwik9, it compresses to about 21MB (~1.68 bits/byte).It's pretty slow right now (roughly 20–30 minutes of training and 45 minutes each for compression and decompression on my AMD 7800XT).Checkout the repo - https://github.com/samyak112/pym-particles

Supported by 1 story

Related Events

Timeline (1 stories)

Jun 26 03:22 AMFirst
Overfitted a 900KB Transformer to Compress a 100MB CSV into 7MB
Hacker News90 engagement

Receipts (1)

Bias Snapshot

Center
Left 0%Center 100%Right 0%