Spaces:
Sleeping
Sleeping
Commit History
-
One tab per dataset domain after Overall
Verified signature970eece -
GIFT-Eval style absolute metrics, Overall/By Domain tabs
Verified signature77c60e3 -
GIFT-Eval style absolute metrics, Overall/By Domain tabs
Verified signatureb0dcb71 -
GIFT-Eval style absolute metrics, Overall/By Domain tabs
Verified signature60bd556 -
Keep progress bar only, remove countdown text and refresh button
Verified signature36a9390 -
Remove Refresh now button, keep countdown only
Verified signature63a5ff9 -
Fix custom_css import on Space startup
Verified signature37745ed -
Fix Space startup: sync full app source
Verified signature7f951ea -
Fix Space startup: sync full app source
Verified signature633eed0 -
Fix Space startup: sync full app source
Verified signature66928b3 -
Format last eval timestamp as UTC+8
Verified signature1b19ebd -
Fix black screen: remove MutationObserver freeze
Verified signatureb35c571 -
UI-only mode: update src/eval_scheduler.py
Verified signature8250dbb -
UI-only mode: update src/benchmark_config.py
Verified signature034b73b -
UI-only mode: update src/about.py
Verified signature6c6e508 -
Fix eval countdown to use persisted next_eval_at deadline.
zhouziyu02 +1: CursorStore next_eval_at in eval_state.json after each run and derive the progress bar from the absolute deadline so page reloads do not reset the 1-hour cycle to zero. Co-authored-by: Cursor <cursoragent@cursor.com>99ee5de -
UI: eval countdown bar, refresh leaderboard only on eval complete.
zhouziyu02 +1: CursorRemove 60s display reload timer; poll eval state internally and refresh tables when a run starts or finishes. Progress bar shows time until next hourly eval cycle. Co-authored-by: Cursor <cursoragent@cursor.com>62f5d66 -
Deploy live 1h benchmark: scheduler, tsfm_bench vendor, latest results.
zhouziyu02 +1: CursorSpace auto-collects TS-Bench data, runs TSFM.ai eval hourly, and reloads leaderboard. Co-authored-by: Cursor <cursoragent@cursor.com>5063745
-
Refresh live eval results and show raw metric aggregation.
zhouziyu02 +1: CursorRe-run all 6 TSFM models on 10 TS-Bench tasks with fresh data; remove baseline normalization from leaderboard display. Co-authored-by: Cursor <cursoragent@cursor.com>1d5b407 -
Fix auto-refresh countdown via Blocks(head=) for HF Spaces
zhouziyu028fcb36a -
Leaderboard: per-domain tables and auto-refresh countdown
zhouziyu023d6dd75 -
Leaderboard: per-domain tables and auto-refresh countdown
zhouziyu021d3f6c0 -
Update TS-Bench live eval results with domain breakdown
zhouziyu023e8d868 -
Update online eval results (6 models)
zhouziyu026ae0ddf -
Update online eval results
zhouziyu02d5f690f -
Update online Open-Meteo eval results
zhouziyu02baee854
-
Update demo leaderboard with TSFM1, TSFM2, and TSFM3.
zhouziyu02 +1: CursorUse gr.Dataframe UI and sync synthetic demo results. Co-authored-by: Cursor <cursoragent@cursor.com>3da9290
-
Add TSFM Realworld Bench leaderboard
zhouziyu02c29fb7f