NO/FOMO

Independent AI signal, once a day

The AI briefing worth opening.

ISSUE DATE2026-04-03DEFAULT EDITION
This issue
—
All time
—

Hacker News

3 stories
01

April 2026 TLDR Setup for Ollama and Gemma 4 26B on a Mac mini

This document, dated April 2026, presents a concise 'Too Long; Didn't Read' (TLDR) setup guide for deploying advanced large language models (LLMs) on a Mac mini. The core focus is on integrating Ollama, an open-source framework designed to simplify running LLMs locally, with the Gemma 4 26B model. This guide is crafted to offer a streamlined and efficient pathway for users to harness sophisticated AI capabilities directly on their personal hardware, thereby democratizing access to powerful generative AI. By detailing essential installation and configuration steps, the resource empowers developers and enthusiasts to experiment with and integrate LLMs without dependence on cloud-based infrastructure. The specific mention of a Mac mini underscores an emphasis on optimizing performance for Apple Silicon architectures, reflecting a significant trend in bringing demanding AI workloads to consumer-grade devices. Such local setups are increasingly vital for secure development environments, privacy-sensitive applications, and scenarios necessitating offline access to AI models, facilitating rapid prototyping and cost-effective AI exploration within a controlled, personal computing context.

OllamaGemma 4 26BMac miniLarge Language ModelLocal AIArtificial IntelligenceLarge Language ModelMachine Learning
02

Show HN: Apfel – The free AI already on your Mac

Apfel is introduced as an innovative, free artificial intelligence application specifically engineered to operate directly on Apple Mac computers, harnessing the device's inherent processing power to deliver AI capabilities locally. This approach distinguishes Apfel from conventional cloud-based AI services by eliminating the need for internet connectivity for core operations, thereby enhancing user privacy and reducing latency. The project, featured on Hacker News, underscores the growing trend of on-device AI, making advanced computational intelligence more accessible and efficient for everyday users. By leveraging the Mac's existing hardware, Apfel aims to democratize AI, enabling a range of intelligent functionalities without subscription fees or reliance on external servers. The initiative's commitment to open-source development is evident through its GitHub repository (github.com/Arthur-Ficial/apfel), inviting community involvement and fostering transparency in its growth. This signifies a move towards empowering users with self-contained AI solutions on their personal devices.

On-device AILocal AIMac softwareArtificial IntelligenceOpen SourceArtificial IntelligenceMachine LearningLarge Language Model
03

Colorado's New Speed Camera System Makes Waze Nearly Useless

The article highlights the introduction of Colorado's advanced speed camera system, which is significantly impacting the utility of real-time navigation applications such as Waze. For years, Waze has empowered drivers by providing crowdsourced alerts on speed traps and camera locations, enabling them to anticipate enforcement and adjust their driving habits. The new system in Colorado, however, appears to circumvent these traditional detection mechanisms, potentially through sophisticated deployment strategies, variable camera activation, or integration with broader traffic management infrastructure. This technological evolution in traffic enforcement challenges the effectiveness of crowdsourced warning systems, leading to a substantial decrease in Waze's ability to provide timely and accurate speed camera alerts to its users. Consequently, drivers may find themselves less informed about potential enforcement zones, potentially leading to increased compliance with speed limits and a reevaluation of reliance on such navigation features for avoiding citations. This shift underscores a broader trend where traffic enforcement agencies are adopting more robust technological solutions to enhance road safety and compliance.

Speed CamerasTraffic EnforcementWazeNavigation TechnologyCrowdsourced DataArtificial IntelligenceAI AgentComputer Vision

huggingface

6 stories
01

DataFlex: A Unified Framework for Data-Centric Dynamic Training of Large Language Models

Data-centric training has emerged as a promising direction for improving large language models (LLMs) by optimizing not only model parameters but also the selection, composition, and weighting of training data during optimization. However, existing approaches to data selection, data mixture optimization, and data reweighting are often developed in isolated codebases with inconsistent interfaces, hindering reproducibility, fair comparison, and practical integration. In this paper, we present DataFlex, a unified data-centric dynamic training framework built upon LLaMA-Factory. DataFlex supports three major paradigms of dynamic data optimization: sample selection, domain mixture adjustment, and sample reweighting, while remaining fully compatible with the original training workflow. It provides extensible trainer abstractions and modular components, enabling a drop-in replacement for standard LLM training, and unifies key model-dependent operations such as embedding extraction, inference, and gradient computation, with support for large-scale settings including DeepSpeed ZeRO-3. We conduct comprehensive experiments across multiple data-centric methods. Dynamic data selection consistently outperforms static full-data training on MMLU across both Mistral-7B and Llama-3.2-3B. For data mixture, DoReMi and ODM improve both MMLU accuracy and corpus-level perplexity over default proportions when pretraining Qwen2.5-1.5B on SlimPajama at 6B and 30B token scales. DataFlex also achieves consistent runtime improvements over original implementations. These results demonstrate that DataFlex provides an effective, efficient, and reproducible infrastructure for data-centric dynamic training of LLMs.

Data-centric trainingLarge Language ModelsDynamic trainingData optimizationLLM frameworkLarge Language ModelMachine LearningNatural Language Processing
02

SKILL0: In-Context Agentic Reinforcement Learning for Skill Internalization

Agent skills, structured packages of procedural knowledge and executable resources that agents dynamically load at inference time, have become a reliable mechanism for augmenting LLM agents. Yet inference-time skill augmentation is fundamentally limited: retrieval noise introduces irrelevant guidance, injected skill content imposes substantial token overhead, and the model never truly acquires the knowledge it merely follows. We ask whether skills can instead be internalized into model parameters, enabling zero-shot autonomous behavior without any runtime skill retrieval. We introduce SKILL0, an in-context reinforcement learning framework designed for skill internalization. SKILL0 introduces a training-time curriculum that begins with full skill context and progressively withdraws it. Skills are grouped offline by category and rendered with interaction history into a compact visual context, teaching he model tool invocation and multi-turn task completion. A Dynamic Curriculum then evaluates each skill file's on-policy helpfulness, retaining only those from which the current policy still benefits within a linearly decaying budget, until the agent operates in a fully zero-shot setting. Extensive agentic experiments demonstrate that SKILL0 achieves substantial improvements over the standard RL baseline (+9.7% for ALFWorld and +6.6% for Search-QA), while maintaining a highly efficient context of fewer than 0.5k tokens per step. Our code is available at https://github.com/ZJU-REAL/SkillZero.

Agentic Reinforcement LearningSkill InternalizationLLM AgentsIn-Context LearningZero-shot LearningAI AgentLarge Language ModelMachine Learning
03

ASI-Evolve: AI Accelerates AI

Can AI accelerate the development of AI itself? While recent agentic systems have shown strong performance on well-scoped tasks with rapid feedback, it remains unclear whether they can tackle the costly, long-horizon, and weakly supervised research loops that drive real AI progress. We present ASI-Evolve, an agentic framework for AI-for-AI research that closes this loop through a learn-design-experiment-analyze cycle. ASI-Evolve augments standard evolutionary agents with two key components: a cognition base that injects accumulated human priors into each round of exploration, and a dedicated analyzer that distills complex experimental outcomes into reusable insights for future iterations. To our knowledge, ASI-Evolve is the first unified framework to demonstrate AI-driven discovery across three central components of AI development: data, architectures, and learning algorithms. In neural architecture design, it discovered 105 SOTA linear attention architectures, with the best discovered model surpassing DeltaNet by +0.97 points, nearly 3x the gain of recent human-designed improvements. In pretraining data curation, the evolved pipeline improves average benchmark performance by +3.96 points, with gains exceeding 18 points on MMLU. In reinforcement learning algorithm design, discovered algorithms outperform GRPO by up to +12.5 points on AMC32, +11.67 points on AIME24, and +5.04 points on OlympiadBench. We further provide initial evidence that this AI-for-AI paradigm can transfer beyond the AI stack through experiments in mathematics and biomedicine. Together, these results suggest that ASI-Evolve represents a promising step toward enabling AI to accelerate AI across the foundational stages of development, offering early evidence for the feasibility of closed-loop AI research.

AI for AIAgentic FrameworkEvolutionary AgentsNeural Architecture DesignReinforcement LearningArtificial IntelligenceAI AgentDeep Learning
04

DynaVid: Learning to Generate Highly Dynamic Videos using Synthetic Motion Data

Despite recent progress, video diffusion models still struggle to synthesize realistic videos involving highly dynamic motions or requiring fine-grained motion controllability. A central limitation lies in the scarcity of such examples in commonly used training datasets. To address this, we introduce DynaVid, a video synthesis framework that leverages synthetic motion data in training, which is represented as optical flow and rendered using computer graphics pipelines. This approach offers two key advantages. First, synthetic motion offers diverse motion patterns and precise control signals that are difficult to obtain from real data. Second, unlike rendered videos with artificial appearances, rendered optical flow encodes only motion and is decoupled from appearance, thereby preventing models from reproducing the unnatural look of synthetic videos. Building on this idea, DynaVid adopts a two-stage generation framework: a motion generator first synthesizes motion, and then a motion-guided video generator produces video frames conditioned on that motion. This decoupled formulation enables the model to learn dynamic motion patterns from synthetic data while preserving visual realism from real-world videos. We validate our framework on two challenging scenarios, vigorous human motion generation and extreme camera motion control, where existing datasets are particularly limited. Extensive experiments demonstrate that DynaVid improves the realism and controllability in dynamic motion generation and camera motion control.

Video SynthesisDynamic Motion GenerationSynthetic Motion DataOptical FlowVideo Diffusion ModelsGenerative AIComputer VisionDeep Learning
05

Omni123: Exploring 3D Native Foundation Models with Limited 3D Data by Unifying Text to 2D and 3D Generation

Recent multimodal large language models have achieved strong performance in unified text and image understanding and generation, yet extending such native capability to 3D remains challenging due to limited data. Compared to abundant 2D imagery, high-quality 3D assets are scarce, making 3D synthesis under-constrained. Existing methods often rely on indirect pipelines that edit in 2D and lift results into 3D via optimization, sacrificing geometric consistency. We present Omni123, a 3D-native foundation model that unifies text-to-2D and text-to-3D generation within a single autoregressive framework. Our key insight is that cross-modal consistency between images and 3D can serve as an implicit structural constraint. By representing text, images, and 3D as discrete tokens in a shared sequence space, the model leverages abundant 2D data as a geometric prior to improve 3D representations. We introduce an interleaved X-to-X training paradigm that coordinates diverse cross-modal tasks over heterogeneous paired datasets without requiring fully aligned text-image-3D triplets. By traversing semantic-visual-geometric cycles (e.g., text to image to 3D to image) within autoregressive sequences, the model jointly enforces semantic alignment, appearance fidelity, and multi-view geometric consistency. Experiments show that Omni123 significantly improves text-guided 3D generation and editing, demonstrating a scalable path toward multimodal 3D world models.

3D Native Foundation ModelsText-to-3D GenerationMultimodalAutoregressive FrameworkGeometric ConsistencyGenerative AIMultimodalDeep Learning
06

GPA: Learning GUI Process Automation from Demonstrations

GUI Process Automation (GPA) is a lightweight but general vision-based Robotic Process Automation (RPA), which enables fast and stable process replay with only a single demo. Addressing the fragility of traditional RPA and the non-deterministic risks of current vision language model-based GUI agents, GPA introduces three core benefits: (1) Robustness via Sequential Monte Carlo-based localization to handle rescaling and detection uncertainty; (2) Deterministic and Reliability safeguarded by readiness calibration; and (3) Privacy through fast, fully local execution. This approach delivers the adaptability, robustness, and security required for enterprise workflows. It can also be used as an MCP/CLI tool by other agents with coding capabilities so that the agent only reasons and orchestrates while GPA handles the GUI execution. We conducted a pilot experiment to compare GPA with Gemini 3 Pro (with CUA tools) and found that GPA achieves higher success rate with 10 times faster execution speed in finishing long-horizon GUI tasks.

GUI Process AutomationRobotic Process AutomationVision-basedAI AgentSequential Monte CarloRoboticsComputer VisionAI Agent