← September 27 · the hive for September 27, 2026
In the news today
TabPFN and TabICL vs. tuned XGBoost: the model that doesn't train won 14/14
The claim behind TabPFN and TabICL is that they predict on a table without ever training on it and still beat tuned boosting.
efraingaray.com
The first buzz on this hive takes the crown. Off the board. One buzz puts it on at the next redraw.
This hive has sealed.
You buzzed this
Claimed Somebody said it and nobody has confirmed it. A tier is not a verdict.
Nobody has backed it yet. Submitted September 27, 2026 at 11:45 PM EDT. Placed September 27, 2026 at 11:45 PM EDT.
Earned a place and found no room on the hive. In the feed.
Sources
The wording on the hive is the source's, never a person's. A check confirms a link resolves and that the page contains the quotation, by exact match. Nothing here decides what is true.
Source 1: efraingaray.com
TabPFN and TabICL vs. tuned XGBoost: the model that doesn't train won 14/14
The claim behind TabPFN and TabICL is that they predict on a table without ever training on it and still beat tuned boosting.
Marked as found on the page, exactly, September 28, 2026 at 11:30 PM EDT. The check history below is the evidence.
Check history
| When | Check | Result | Status | Detail |
|---|---|---|---|---|
| September 27, 2026 at 11:45 PM EDT | Link resolves | Passed | 200 | 200, text/html, 117452 characters, landed on https://efraingaray.com/en/blog/tabpfn-vs-xgboost/ |
| September 27, 2026 at 11:45 PM EDT | Page contains the quotation | Passed | 200 | the page contains the quotation, exactly |
More from the hive for September 27, 2026
Everything on September 27 · Another random story from September 27