Model card: lead with the shipped run-2 model, 4-image GenEval table, files, history

Move to ML-Intern-lab: new repo names, card metadata, Space links

run2: s4p section (paired 4-img GenEval table, gate metrics, changelog, v1 backup)

run2: s4p EMA weights (fine-tuned s4, 1125 steps)

run2: v1 backup of the shipped 4-step weights (pre-s4p)

run2: webgpu generator.onnx from s4p weights

run2: manifest for regenerated generator.onnx (s4p)

README: rounded table + student latency rows

README: images + training sections

README asset: s4_this

README asset: s4_prev

README asset: s4_teacher50

README asset: loss_curves.png

ship: 8step/generator.safetensors

ship: generator.safetensors

ship: README.md

ship: agate/marking.py

ship: agate/pipeline.py

ship: config.json

ship: webgpu/manifest.json

student 4-step webgpu onnx

remove teacher extras

remove teacher webgpu v2 format

remove teacher extras

remove teacher extras

Duplicate from Logolabs/agate-preview-002