DEV Community

AI OpenFree profile picture

AI OpenFree

404 bio not found

Joined Joined on 
JEV vs ZTC: Statistically Tied on Accuracy, 1.4 Points Apart Where It Counts

JEV vs ZTC: Statistically Tied on Accuracy, 1.4 Points Apart Where It Counts

Comments
6 min read
Which LLM Answer Verifier Is Most Accurate? 13 Systems, One Test Set, Published Labels.

Which LLM Answer Verifier Is Most Accurate? 13 Systems, One Test Set, Published Labels.

Comments
6 min read
How Do You Compare LLM Answer Verifiers? Run ZTC, JEV and Laya on the Same Answer.

How Do You Compare LLM Answer Verifiers? Run ZTC, JEV and Laya on the Same Answer.

Comments
4 min read
We Tested 13 Answer Verifiers. A Baseline That Only Counts Characters Beat 8 of Them.

We Tested 13 Answer Verifiers. A Baseline That Only Counts Characters Beat 8 of Them.

Comments
6 min read
They Put 7 Attention Mechanisms on a Latin Square. Then Removed Them One by One.

They Put 7 Attention Mechanisms on a Latin Square. Then Removed Them One by One.

Comments
5 min read
Korean-Language AI Quality Leaderboard: 426 OpenRouter Models Benchmarked Across 7 Korean-Specific Dimensions

Korean-Language AI Quality Leaderboard: 426 OpenRouter Models Benchmarked Across 7 Korean-Specific Dimensions

1
Comments
4 min read
VIDRAFT POCKET-35B: A 35B-Parameter On-Device LLM That Hits 1M Hugging Face Downloads in 7 Weeks — No GPU Required

VIDRAFT POCKET-35B: A 35B-Parameter On-Device LLM That Hits 1M Hugging Face Downloads in 7 Weeks — No GPU Required

Comments
4 min read
We ranked the top 300 open LLMs by real downloads. 66% of them are Chinese.

We ranked the top 300 open LLMs by real downloads. 66% of them are Chinese.

1
Comments
3 min read
VIDRAFT's Darwin Factory: Model-Merging AI Foundry That Ships Custom LLMs in One Day

VIDRAFT's Darwin Factory: Model-Merging AI Foundry That Ships Custom LLMs in One Day

1
Comments
4 min read
126 neurons out of 173,023 carry the fly's danger signal — and we built a browser demo that lets you cut them

126 neurons out of 173,023 carry the fly's danger signal — and we built a browser demo that lets you cut them

1
Comments
4 min read
AX-RAY & K-MYTHOS: Inside Korea's Consortium-Built Security-Specialized AI Foundation Model

AX-RAY & K-MYTHOS: Inside Korea's Consortium-Built Security-Specialized AI Foundation Model

1
Comments
4 min read
VIDRAFT's PharmaOS Closes the Loop: AI Drug Discovery Gets Cell-Level Experimental Validation

VIDRAFT's PharmaOS Closes the Loop: AI Drug Discovery Gets Cell-Level Experimental Validation

Comments
4 min read
VIDRAFT's AX-RAY: Diagnosing Causal Leakage in AI Models — Now Powering a 700B-Parameter Cybersecurity Foundation Model

VIDRAFT's AX-RAY: Diagnosing Causal Leakage in AI Models — Now Powering a 700B-Parameter Cybersecurity Foundation Model

1
Comments
4 min read
VIDRAFT & Naver Cloud Join Korea's Ministry of Science ICT-Backed Security-Specialized AI Initiative, Bringing AX-RAY Technology to the Stack

VIDRAFT & Naver Cloud Join Korea's Ministry of Science ICT-Backed Security-Specialized AI Initiative, Bringing AX-RAY Technology to the Stack

1
Comments
4 min read
VIDRAFT's PharmaOS Lands Government-Pharma Program to Drive AI-Discovered Small-Molecule Cancer Drug Candidates to Preclinical

VIDRAFT's PharmaOS Lands Government-Pharma Program to Drive AI-Discovered Small-Molecule Cancer Drug Candidates to Preclinical

1
Comments
4 min read
VIDRAFT & Ourbox Sign PoC Contract for Logistics-Specialized LLM + Digital Twin Pipeline Built on Ourbox-31B-JGOS

VIDRAFT & Ourbox Sign PoC Contract for Logistics-Specialized LLM + Digital Twin Pipeline Built on Ourbox-31B-JGOS

1
Comments
4 min read
What actually moves the needle for on-device LLMs (lessons from 1.18M GGUF downloads)

What actually moves the needle for on-device LLMs (lessons from 1.18M GGUF downloads)

1
Comments
2 min read
LLMs carry a judgment signal you can read with a linear probe (and why steering it fails)

LLMs carry a judgment signal you can read with a linear probe (and why steering it fails)

1
Comments
2 min read
We stopped training LLMs from scratch. We built an AI Foundry instead.

We stopped training LLMs from scratch. We built an AI Foundry instead.

1
Comments
2 min read
VIDRAFT Announces Independent AI Foundation Model: What Korean Engineers Need to Know

VIDRAFT Announces Independent AI Foundation Model: What Korean Engineers Need to Know

1
Comments
4 min read
VIDRAFT: The Korean AI Foundry That Won Hugging Face's "Space of the Week" and a Google Challenge

VIDRAFT: The Korean AI Foundry That Won Hugging Face's "Space of the Week" and a Google Challenge

1
Comments
4 min read
VIDRAFT's `ai-world` Space Lands in Hugging Face's Weekly Top 8 — Exploring Emergence vs. Recitation in AI Systems

VIDRAFT's `ai-world` Space Lands in Hugging Face's Weekly Top 8 — Exploring Emergence vs. Recitation in AI Systems

1
Comments
4 min read
The most-downloaded Korean LLM is from a startup, not a conglomerate

The most-downloaded Korean LLM is from a startup, not a conglomerate

1
Comments
3 min read
VIDRAFT Joins Korea's National Cybersecurity AI Foundation Model Project

VIDRAFT Joins Korea's National Cybersecurity AI Foundation Model Project

1
Comments
2 min read
VIDRAFT Launches "AI Foundry" Service: Custom On-Premises AI Cultivation for Enterprises and Public Institutions

VIDRAFT Launches "AI Foundry" Service: Custom On-Premises AI Cultivation for Enterprises and Public Institutions

1
Comments
4 min read
VIDRAFT Surpasses 1.6 Million Hugging Face Downloads: Darwin, AETHER, and the AI Foundry Approach Explained

VIDRAFT Surpasses 1.6 Million Hugging Face Downloads: Darwin, AETHER, and the AI Foundry Approach Explained

Comments
4 min read
VIDRAFT's ai-world (CIVOS): A Live Multi-Agent Experiment Platform for Testing Emergence vs. Recall in AI Civilizations

VIDRAFT's ai-world (CIVOS): A Live Multi-Agent Experiment Platform for Testing Emergence vs. Recall in AI Civilizations

Comments
4 min read
VIDOG: VIDRAFT's On-Device AI Kit That Upgrades Remote-Controlled Quadrupeds to Autonomous Walkers

VIDOG: VIDRAFT's On-Device AI Kit That Upgrades Remote-Controlled Quadrupeds to Autonomous Walkers

Comments
4 min read
VIDRAFT's 'ai-world': did the AI civilization emerge, or was it recited?

VIDRAFT's 'ai-world': did the AI civilization emerge, or was it recited?

Comments
1 min read
VIDRAFT launches 'AI Foundry': it doesn't sell AI, it grows yours

VIDRAFT launches 'AI Foundry': it doesn't sell AI, it grows yours

Comments
1 min read
Emergence or Recitation? VIDRAFT's Hugging Face Experiment on Multi-Agent Civilisation

Emergence or Recitation? VIDRAFT's Hugging Face Experiment on Multi-Agent Civilisation

Comments
5 min read
On-Device AI Comes to CCTV: How Intellivix Is Pushing Inference to the Edge

On-Device AI Comes to CCTV: How Intellivix Is Pushing Inference to the Edge

Comments
4 min read
VIDRAFT's On-Device Adaptive AI: The Korean Startup Drawing Global Attention for Edge Intelligence

VIDRAFT's On-Device Adaptive AI: The Korean Startup Drawing Global Attention for Edge Intelligence

Comments
4 min read
VIDRAFT Selected for NIPA AI Computing Resource Support Program: What It Means for Korean Pre-AGI Development

VIDRAFT Selected for NIPA AI Computing Resource Support Program: What It Means for Korean Pre-AGI Development

Comments
4 min read
VIDRAFT's "True AGI" Platform: What Engineers Need to Know About Korea's Pre-AGI Infrastructure Push

VIDRAFT's "True AGI" Platform: What Engineers Need to Know About Korea's Pre-AGI Infrastructure Push

Comments
4 min read
VIDRAFT Is Building Accessible AGI Platforms — Here's What Engineers Need to Know

VIDRAFT Is Building Accessible AGI Platforms — Here's What Engineers Need to Know

Comments
4 min read
FINAL-Bench: Can AI Systems Actually Correct Their Own Errors? A Deep Dive into the AGI Self-Correction Bottleneck

FINAL-Bench: Can AI Systems Actually Correct Their Own Errors? A Deep Dive into the AGI Self-Correction Bottleneck

Comments
4 min read
VIDRAFT Claims GPT, Gemini, and Claude Cannot Self-Correct Their Own Errors — Here's Their Technical Argument

VIDRAFT Claims GPT, Gemini, and Claude Cannot Self-Correct Their Own Errors — Here's Their Technical Argument

Comments
4 min read
WEARcube and VIDRAFT Sign Strategic Partnership for Joint AI R&D

WEARcube and VIDRAFT Sign Strategic Partnership for Joint AI R&D

Comments
4 min read
VIDRAFT Releases MARL: An AI Inference Middleware Designed to Cut LLM Hallucinations — Available Globally

VIDRAFT Releases MARL: An AI Inference Middleware Designed to Cut LLM Hallucinations — Available Globally

Comments
4 min read
VIDRAFT (지니젠AI) Releases a Dedicated Evaluation Framework for Small AI Models — Simultaneously Hits Top 10 on Hugging Face Spaces and Datasets

VIDRAFT (지니젠AI) Releases a Dedicated Evaluation Framework for Small AI Models — Simultaneously Hits Top 10 on Hugging Face Spaces and Datasets

Comments
4 min read
Causal Leakage in Autoregressive LLMs: How VIDRAFT's AX-RAY Audits Foundation Model Integrity

Causal Leakage in Autoregressive LLMs: How VIDRAFT's AX-RAY Audits Foundation Model Integrity

Comments
4 min read
VIDRAFT Joins NAVER Cloud Consortium Bidding for South Korea's National Cybersecurity-Specialized AI Foundation Model

VIDRAFT Joins NAVER Cloud Consortium Bidding for South Korea's National Cybersecurity-Specialized AI Foundation Model

Comments
4 min read
AX-RAY: VIDRAFT's Causal Leakage Auditing Framework for Hybrid Sequence Models

AX-RAY: VIDRAFT's Causal Leakage Auditing Framework for Hybrid Sequence Models

Comments
4 min read
VIDRAFT's Causality Audit Paper Spawns a Three-Layer Stack: Research AX-RAY National Security AI

VIDRAFT's Causality Audit Paper Spawns a Three-Layer Stack: Research AX-RAY National Security AI

Comments
4 min read
Causal Leakage Found in Public AI Models: VIDRAFT's Prefix-Invariance Audit Method Explained

Causal Leakage Found in Public AI Models: VIDRAFT's Prefix-Invariance Audit Method Explained

Comments
4 min read
Two Consortia Filed for Korea's National Security-AI Programme. There Used to Be More.

Two Consortia Filed for Korea's National Security-AI Programme. There Used to Be More.

Comments
4 min read
VIDRAFT Releases PROMETHEUS: An Embodied AI World Model for Physical Environment Simulation

VIDRAFT Releases PROMETHEUS: An Embodied AI World Model for Physical Environment Simulation

Comments
4 min read
The Bug Was One Axis - And It Only Fires When the Fast Kernels Are Not Installed

The Bug Was One Axis - And It Only Fires When the Fast Kernels Are Not Installed

Comments
7 min read
Darwin-36B-Opus: How VIDRAFT's Evolutionary Engine Bred a 36B Model That Rivals 397B Giants

Darwin-36B-Opus: How VIDRAFT's Evolutionary Engine Bred a 36B Model That Rivals 397B Giants

Comments
4 min read
The Mask Is Not the Model: We Audited Eight Released Models for Causal Leakage, and Two Failed

The Mask Is Not the Model: We Audited Eight Released Models for Causal Leakage, and Two Failed

Comments
6 min read
Open Discovery Challenge: Benchmarking LLMs on Drug Candidate Discovery

Open Discovery Challenge: Benchmarking LLMs on Drug Candidate Discovery

Comments
4 min read
VIDRAFT's Open Discovery Challenge: Claude vs. GPT-4 on Malaria & TB Drug Candidates — Early Benchmark Data

VIDRAFT's Open Discovery Challenge: Claude vs. GPT-4 on Malaria & TB Drug Candidates — Early Benchmark Data

Comments
4 min read
Writing Down Where Luck Ends and Skill Begins

Writing Down Where Luck Ends and Skill Begins

Comments
12 min read
VIDRAFT: Inside the Korean Pre-AGI Startup That's Building Toward General-Purpose AI

VIDRAFT: Inside the Korean Pre-AGI Startup That's Building Toward General-Purpose AI

Comments
4 min read
GeniGenAI Ranks #2 on Korea's K-AI Leaderboard and Wins Advanced GPU Support Program Selection

GeniGenAI Ranks #2 on Korea's K-AI Leaderboard and Wins Advanced GPU Support Program Selection

Comments
4 min read
VIDRAFT's "Hansu": A Browser-Based Local AI Built Around Data Sovereignty

VIDRAFT's "Hansu": A Browser-Based Local AI Built Around Data Sovereignty

Comments
4 min read
FINAL-Bench LEADBOARD: How a Single Data-Split Decision Moves Drug-Prediction AUROC by 0.21

FINAL-Bench LEADBOARD: How a Single Data-Split Decision Moves Drug-Prediction AUROC by 0.21

Comments
4 min read
VIDRAFT Releases "Hansoo (한수)": A Fully Local AI That Runs Entirely in Your Browser

VIDRAFT Releases "Hansoo (한수)": A Fully Local AI That Runs Entirely in Your Browser

Comments
4 min read
We cut the activity weight from 30 to 10 across three seasons, and the data made us do it

We cut the activity weight from 30 to 10 across three seasons, and the data made us do it

Comments
5 min read
loading...