t0-beta-q4_0-webgpu

Q4_0-quantized weights for t0-beta, packaged for browser WebGPU forecasting.

theforecastingcompany/t0-beta, by The Forecasting Company

Q4_0 quantization for WebGPU by Idle Intelligence.

Apache-2.0

Smallest, fastest quant of t0-beta: 149.5MB, 0.14x the F32 weights. Worst-case point drift against the F32 reference is 14.6%, looser than t0-alpha’s own Q4_0 (8.4%); no full 97-config GIFT-Eval run exists yet for t0-beta at any quant, only an 8-config subset.

Type
model
Runs
browser, WASM, WebGPU
Size
256M params, 149.5 MB (Q4_0)
Demo (GitHub Pages)
https://idle-intelligence.github.io/t0-web/web/
Source
idle-intelligence/t0-web
HuggingFace
idle-intelligence/t0-beta-q4_0-webgpu
Status
maintained

← Home