scima

scima

·

AI & ML interests

None yet

Recent Activity

liked a model about 2 months ago

logic65/whittle-dev

updated a model 2 months ago

scima/whittle-14.7b-scima

published a model 2 months ago

scima/whittle-14.7b-scima

View all activity

Organizations

None yet

upvoted 3 collections 3 months ago

Whittle: research and dev

The compression ladder, recipes, adapters and unrepaired variants behind the family. • 2 items • Updated 24 days ago • 2

Whittle: models you can run

Compressed Qwen3.8-27B variants with weights published and tested. Start here. • 4 items • Updated 24 days ago • 2

Qwen3.8 Whittle family

A 27B whittled on two 8GB GPUs: measured cuts, heals, and restorations. 48L = best untrained, Whittle-16B = best overall. • 1 item • Updated 24 days ago • 1