Spaces:
Running on Zero
Running on Zero
Commit History
Today
-
UX update (rev-ask-base120-zerogpu-ux-001, Parallax 2/2, Reuben OK) Running
Verified signature0a301b4
5 days ago
-
Run Ask Base120 on ZeroGPU so free accounts can use live answers
Host the Space on ZeroGPU and run Qwen2.5-7B-Instruct locally instead of Inference Providers (which require paid credits). Each live run uses one short GPU reservation (40s Manual / 55s Auto) and 384 new tokens. Cached examples and the Operator explorer never request a GPU. README documents free-tier hosting and daily quota limits.2084d2a -
Switch default live model to Qwen2.5-7B-Instruct (7.61B)
Mozilla Human-Scale AI / small-model constraint: replace Qwen2.5-72B-Instruct with the 7.61B Apache-2.0 instruct sibling, still via Hugging Face Inference Providers. Cached one-click examples are left as the 72B runs from 15 October 2026 and the README now states both models and parameter counts honestly.5a72e94
26 days ago
-
Deploy rev-ask-base120-space-015 (route guard, timeout/budget/slots polish, README accuracy)
Verified signature007282a
-
Ask Base120: operator-fit prompts, regenerated cache with raw responses, known-mistakes README (wags rev-ask-base120-space-012 APPROVE 2/2)
Verified signatureConverged APPROVE 2/2, 0 blockers (MTSMU f4d7f37a…, Parallax Quilter 1032a67e…386a). Includes rev-008..rev-012. Deployed with Reuben's explicit OK.67f221e -
Ask Base120: sign-in (OAuth) live mode, wags rev-ask-base120-space-007 APPROVE (MTSMU + Parallax Quilter)
Verified signature7f03c90
-
Ask Base120: cached-only demo (wags rev-ask-base120-space-003 APPROVE)
Verified signature03d11a7