t0-alpha-q4_0-webgpu

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

theforecastingcompany/t0-alpha, by The Forecasting Company

Q4_0 quantization for WebGPU by Idle Intelligence.

Apache-2.0

Smallest, fastest quant of t0-alpha: 29.8 ms per forecast on an Apple M2. Costs +1.1% MASE and +0.6% CRPS against the F32 control on the full 97-config GIFT-Eval protocol, within 1.3% of the published t0-alpha card. The quant trucs.ai/t0/ loads.

Type
model
Runs
browser, WASM, WebGPU
Size
~102M params, 58.6 MB (Q4_0)
Demo (GitHub Pages)
https://idle-intelligence.github.io/t0-web/web/
Source
idle-intelligence/t0-web
HuggingFace
idle-intelligence/t0-alpha-q4_0-webgpu
Perf
GIFT-Eval 97 configs, normalized to Seasonal Naive: MASE 0.7334, CRPS 0.4973
Status
maintained

Used on: forecasting

← Home