Spaces:
Running
Running
Commit History
6 days ago
-
Update leaderboard data + version history (TabArena-v0.1.9.7) Running
0bc60d4
8 days ago
-
Update leaderboard data + version history (TabArena-v0.1.9.6)
New verified models PrismBoost and LightPFN, and the first closed-source API system, Chakra-Tab, in its medium and full presets. Systems are colored by category, closed-source API systems in a rust shade, and the Type pill of the cross-subset overview carries the tag chip. A system that fits inside its prediction call counts that call as its train time, marked with a dagger. The closed-source API warning and the Systems panel say why such results should not be trusted by default. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>0aac376
13 days ago
-
Update leaderboard data + version history (TabArena-v0.1.9.5)
Add the verified Kumo-Tabular, Kumo-Tabular-Medium and Kumo-Tabular-Small (suites tabarena-2026-10-06 and beyondarena-2026-10-06) to TabArena and BeyondArena. BeyondArena moves from the 507-split core protocol to core2k (2,000 splits from the cost-weighted allocation): every BeyondArena cell is regenerated on core2k (autogluon/tabarena#647), and the page, API docstrings and Metrics & Protocol tab now name core2k. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>fa159b9
21 days ago
-
Update leaderboard data + version history (TabArena-v0.1.9.3)
TabArena: the Linear re-run with L1 honored for classification. BeyondArena: the nine foundation models of the 2026-09-22 run and the Linear re-run, in the new layout with TabArena's explorers. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>0f16a23
27 days ago
-
Update leaderboard data + version history (TabArena-v0.1.9.2)
2b3adbd
-
Update leaderboard data + version history (TabArena-v0.1.9.1)
CatBoost returns to its July run (the re-run's fit times are not comparable yet), the per-dataset browser shows the Linear and KNN tuning trajectories again and marks methods without one, and the explorers keep a metric picked inside them across the paper-view toggle. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>9d41c20 -
Update leaderboard data + version history (TabArena-v0.1.9)
Re-run of the hosted foundation models and CatBoost under the timing and warm-up pipeline, TabPFN-3.5 and TabPFN-3.5-Fast (fit times re-measured), LimiX-2, the balanced / imbalanced / extreme dataset subsets, per-method licenses with the non-commercial toggle, and the explorers keeping a metric picked inside them when the host echoes its own. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>e24f3a7
-
Update leaderboard data + version history for ChimeraBoost 0.30.0
9b53220
-
Per-dataset results leave out the models that could not run on a dataset
An imputed score stands in for a model that could not run on that dataset, so the per-dataset view no longer shows it there -- the model keeps its place on every dataset where it did run. The size of the field now varies by dataset, and the pane says how many could not run. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>74c6c71 -
Charts no longer draw themselves before they can be measured
A chart drawn while its container was still unlaid-out pinned itself to its minimum width and stacked its labels, and nothing later healed it. It now waits for a measurable box, never caches a zero text measurement, and watches its own box rather than the viewport. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>5cf9f1a -
Systems as their own class of entrant, and per-dataset results
Systems (AutoGluon, TabFM+, agents, hosted APIs) are now their own kind of entrant rather than a reference pipeline alongside the models. Every leaderboard number is relative to the field, so each combination of system categories is its own published evaluation rather than a row filter, picked with the new "Who's competing?" control; models-only is the default and is now genuinely models-only. New "Per-dataset results" section: the benchmark one dataset at a time. Pick a contender and each row reports its rank and its signed margin on that dataset, with a star placing it in the field; select a dataset for its own tuning trajectories and its full ranking. Also: an "I care about" control that leads with the figure answering your question, the win-rate matrix carrying every tuning variant again, and the pinned overlays staying put inside the Space. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>9a668d8
-
Refresh leaderboard data with the re-run EXAONE-Tabular results
EXAONE-Tabular now covers regression as well as classification, so its regression numbers are measured instead of imputed. Its classification runs were redone on the updated released checkpoint and on the classifier default n_svd=0 that its authors published, which moves those numbers too. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01CJAJDEkj2fiLTtjeLo5APw3e8e748 -
Revert "Refresh leaderboard data with the re-run EXAONE-Tabular results"
This reverts commit d96dd9846bbc2f9caffd31fd20055e720e08f98e.32fa5a1 -
Refresh leaderboard data with the re-run EXAONE-Tabular results
EXAONE-Tabular now covers regression as well as classification, so its regression numbers are measured instead of imputed. Its classification runs were redone on the updated released checkpoint, which moves those numbers too. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01CJAJDEkj2fiLTtjeLo5APwd96dd98
-
fix: remove medals again
454a394 -
add: add to LB
4af701c
-
fix: drop medals from the full leaderboard table, hold the dark theme
The full leaderboard table already carries a # column, so the 🥇🥈🥉 on its Elo cells were a second, competing ordering. Removed upstream in `tabarena.plot.interactive._leaderboard_table_template` and the 72 `leaderboard_table.html` artifacts re-rendered from their sibling `website_leaderboard.csv` — the same call the conversion step makes, verified by re-rendering before the change and getting all 72 back byte-identical. The app's own fallback table lost them too. The cross-subset overview keeps its medals: no # column there. Theme: `FORCE_DARK_THEME_JS` reloaded the page to append `__theme=dark`, but Gradio 6 resolves the `theme-mode` attribute and `system` (which follows the viewer's OS and re-applies itself on change) ahead of that parameter, so a light-OS or light-embed visitor beat it — and reloading only lost the same race one page load later. Force the `dark` class instead, which is what Gradio's stylesheet keys off, and hold it with a MutationObserver. `THEME` covers the fra…dd9bac5
-
add: MCP and new interactive plots
218b520
-
add: paper-view default, figure download, panel restyle
Each figure panel now opens in paper view — white background, legend and chart only — with a prominent "Edit view" button into the controls and the SVG/PDF/PNG download buttons pinned to the right of the header. Both are driven over postMessage, since the explorers run in sandboxed cross-origin frames; `allow-downloads` is added to the sandbox, without which the download is silently dropped. The panels are also easier to tell apart: each is a card with a titled header strip, and the long "Name [subset]" label splits into a heading plus a quiet qualifier. Three fixes found while testing, all in the host page rather than the explorers: - The full-bleed header was made so with negative margins, which pushed it wider than its Gradio block and gave the title its own scrollbar. - Gradio puts `margin-bottom: 4px` on every button but the last, which under centre alignment lifted them 2px above their neighbours. - `HEAD` is a plain Python string, so JS `\u` escapes need doubling or the …c63e790 -
add: interactive leaderboard overview, paper view, US decimals
Embeds the new leaderboard_overview_explorer artifact beside the static tuning-impact figure, and reworks the figure panels around it. - Every figure panel is a card with a titled header carrying two toggles: interactive/static, and paper view. The long "Name [subset]" label splits into a heading plus a quiet qualifier so the blocks are easy to tell apart. - Paper view (white background, chart and legend only, for slides and papers) is driven from the panel header via postMessage, since the explorers run in sandboxed cross-origin frames. - Numbers use a "." decimal separator whatever the browser locale: toLocaleString and Intl.NumberFormat are pinned to en-US with grouping off, so a visitor with a German browser no longer reads "0,886" for a score of 0.886 beside a figure that says 0.886. - BeyondArena deliberately stays on static figures for now. Also refreshes the shipped Pareto and tuning-trajectory explorer pages, which pick up the shared chrome, the flatter fixed…3b80af5
-
add: interactive explorer artifacts (html + csv data exports)
43646be -
add: new plot style
7c1c953 -
add: nori-30m
832ff68 -
add: rerun Pareto front
84710dc
-
add: tabswift
c7f648b
-
add: tabfm
6622a60
-
fix: model type map
2c7e176 -
add nori and chimera
f851b5b
-
Update to v0.1.4
a318804
-
Update leaderboard to v0.1.3.1
6f42195
-
Update plot visuals
8d41bef
-
Update to v0.1.3
afdd14e
-
add sap-rpt-1-oss
c6f5a2c
-
update: fixed plots
0c1ff2d
-
maint: new LB
b17d346
-
maint: update to new lb design
4c3d3df
-
update data from final artifacts, remove not used LBs
ddac93b
-
fix: lb and data update
06834bd
-
fix file endings
add: new LB version912c074
-
add: more metric cols
5251891
-
maint: make more nice looking
bb4bec3 -
add files
25c94ab
