<?xml version="1.0" encoding="utf-8"?>
<feed xmlns="http://www.w3.org/2005/Atom">
    <id>https://zerooneresearch.ai/</id>
    <title>Zero One Research — Research</title>
    <updated>2026-06-03T22:17:50.302Z</updated>
    <generator>https://github.com/jpmonette/feed</generator>
    <author>
        <name>Zero One Research</name>
        <email>hello@zerooneresearch.ai</email>
        <uri>https://zerooneresearch.ai</uri>
    </author>
    <link rel="alternate" href="https://zerooneresearch.ai/"/>
    <link rel="self" href="https://zerooneresearch.ai/research/feed.xml"/>
    <subtitle>Research posts, model releases, and negative results from Zero One Research.</subtitle>
    <rights>© 2026 Zero One Research s. r. o.</rights>
    <entry>
        <title type="html"><![CDATA[PredictLM v1: 0.751 cls / 0.609 reg on OpenML via test-time training]]></title>
        <id>https://zerooneresearch.ai/research/predictlm-v1-launch/</id>
        <link href="https://zerooneresearch.ai/research/predictlm-v1-launch/"/>
        <updated>2026-05-27T00:00:00.000Z</updated>
        <summary type="html"><![CDATA[Two open-weight tabular models (Mini 13M, Base 26M, Apache-2.0) and a test-time training recipe that hits 0.751 classification accuracy on a 25-dataset OpenML benchmark. Plus 11 architecture experiments that didn't move the needle.]]></summary>
        <author>
            <name>Zero One Research</name>
            <uri>https://zerooneresearch.ai</uri>
        </author>
        <category label="predictlm"/>
        <category label="release"/>
        <category label="test-time-training"/>
        <category label="negative-results"/>
        <category label="model-release"/>
    </entry>
    <entry>
        <title type="html"><![CDATA[PredictLM-Mini: a 13M-parameter tabular foundation model with calibrated uncertainty]]></title>
        <id>https://zerooneresearch.ai/research/predictlm-mini/</id>
        <link href="https://zerooneresearch.ai/research/predictlm-mini/"/>
        <updated>2026-05-26T00:00:00.000Z</updated>
        <summary type="html"><![CDATA[The smallest open-weight tabular foundation model with calibrated uncertainty out of the box. 13M parameters, 54 MB, Apache-2.0. Statistically tied with Base on classification accuracy and within ~4 pp R² on regression.]]></summary>
        <author>
            <name>Zero One Research</name>
            <uri>https://zerooneresearch.ai</uri>
        </author>
        <category label="predictlm"/>
        <category label="release"/>
        <category label="distillation"/>
        <category label="calibration"/>
        <category label="model-release"/>
    </entry>
    <entry>
        <title type="html"><![CDATA[Six architectural experiments. Five lost. Here's what we shipped.]]></title>
        <id>https://zerooneresearch.ai/research/architectural-experiments/</id>
        <link href="https://zerooneresearch.ai/research/architectural-experiments/"/>
        <updated>2026-05-25T00:00:00.000Z</updated>
        <summary type="html"><![CDATA[Before settling on Mini's architecture, we ran six experiments at the 26–57M parameter scale. Five lost to the smaller model. Here's the full writeup, including the gradient-NaN debugging.]]></summary>
        <author>
            <name>Zero One Research</name>
            <uri>https://zerooneresearch.ai</uri>
        </author>
        <category label="predictlm"/>
        <category label="ablation"/>
        <category label="negative-results"/>
        <category label="technical"/>
    </entry>
    <entry>
        <title type="html"><![CDATA[BarDistribution: why every regressor should return a distribution]]></title>
        <id>https://zerooneresearch.ai/research/bardistribution/</id>
        <link href="https://zerooneresearch.ai/research/bardistribution/"/>
        <updated>2026-05-24T00:00:00.000Z</updated>
        <summary type="html"><![CDATA[A 1024-bin quantile head is the difference between a guess and a calibrated prediction. We explain how the head works, why softmax-over-bins beats Gaussian heads for in-context regression, and what the distribution unlocks.]]></summary>
        <author>
            <name>Zero One Research</name>
            <uri>https://zerooneresearch.ai</uri>
        </author>
        <category label="calibration"/>
        <category label="uncertainty"/>
        <category label="predictlm"/>
        <category label="technical"/>
    </entry>
</feed>