AI Pulse by Inblix

AI Research

Cutting-edge AI research papers, breakthroughs, and academic developments.

595 articles

Hugging Face’s Research Tracker MCP automates the soul-crushing lit review grind — Inblix summary
Research

Hugging Face’s Research Tracker MCP automates the soul-crushing lit review grind

Hugging Face Blog · Aug 18, 2025 · 2 min read

If you’ve spent a Tuesday afternoon with 47 browser tabs open, manually cross-referencing an arXiv preprint against Git...

A tiny 0.6B model just crushed formal math proofs, topping the MiniF2F leaderboard — Inblix summary
Research

A tiny 0.6B model just crushed formal math proofs, topping the MiniF2F leaderboard

Hugging Face Blog · Aug 14, 2025 · 2 min read

A new open-source training recipe is producing absurdly strong theorem-proving models that fit on a laptop. The team be...

72% of Phones Can Now Run Llama 3.2 Locally—Even the 5-Year-Old Ones — Inblix summary
Research

72% of Phones Can Now Run Llama 3.2 Locally—Even the 5-Year-Old Ones

Hugging Face Blog · Aug 13, 2025 · 2 min read

The idea that generative AI needs the latest $1,000 phone is officially dead. Arm's latest software push, pairing its K...

Arm's AI upscaling cuts mobile GPU workload in half in 4ms test — Inblix summary
Research

Arm's AI upscaling cuts mobile GPU workload in half in 4ms test

Hugging Face Blog · Aug 12, 2025 · 2 min read

Arm just dropped Neural Super Sampling into the hands of developers, and the numbers are worth paying attention to. In...

Frontier LLMs fumble classic 80s text games, fail to climb back down cliffs — Inblix summary
Research

Frontier LLMs fumble classic 80s text games, fail to climb back down cliffs

Hugging Face Blog · Aug 12, 2025 · 2 min read

If you think today’s frontier models are on the verge of AGI, watching them get hopelessly lost in a 1980s text adventu...

GPT-4o still beats local LLMs in Filipino, but a 2-3% fine-tuning gain keeps the underdogs in the race — Inblix summary
Research

GPT-4o still beats local LLMs in Filipino, but a 2-3% fine-tuning gain keeps the underdogs in the race

Hugging Face Blog · Aug 12, 2025 · 2 min read

The new FilBench evaluation paints a clear but nuanced picture of AI's relationship with Philippine languages. After te...

Hugging Face drops an open-source spreadsheet that runs thousands of AI models on your data—no code needed — Inblix summary
Research

Hugging Face drops an open-source spreadsheet that runs thousands of AI models on your data—no code needed

Hugging Face Blog · Aug 8, 2025 · 2 min read

Hugging Face just released AI Sheets, an open-source tool that drags the spreadsheet interface into the era of large la...

Hugging Face's New Multi-GPU Recipe Cuts Through Training Complexity — Inblix summary
Research

Hugging Face's New Multi-GPU Recipe Cuts Through Training Complexity

Hugging Face Blog · Aug 8, 2025 · 2 min read

Distributed training is about to get a lot less painful. Hugging Face has pulled back the curtain on a new, unified app...

Hugging Face's TRL Library Just Added 3 New Ways to Align Vision-Language Models — Inblix summary
Research

Hugging Face's TRL Library Just Added 3 New Ways to Align Vision-Language Models

Hugging Face Blog · Aug 7, 2025 · 2 min read

Hugging Face has significantly expanded its Transformer Reinforcement Learning (TRL) library, adding native support for...

OpenAI drops GPT-OSS: a 120B open-source MoE model that fits on a single 80GB GPU — Inblix summary
Research

OpenAI drops GPT-OSS: a 120B open-source MoE model that fits on a single 80GB GPU

Hugging Face Blog · Aug 5, 2025 · 3 min read

OpenAI just did something it rarely does: it open-sourced a family of models. Meet GPT-OSS, a pair of reasoning-focused...

NVIDIA's 49B Nemotron Just Topped the DeepResearch Bench—and It Runs on a Single GPU — Inblix summary
Research

NVIDIA's 49B Nemotron Just Topped the DeepResearch Bench—and It Runs on a Single GPU

Hugging Face Blog · Aug 4, 2025 · 2 min read

The open-source agent stack just got a new leader. NVIDIA’s AI-Q blueprint, a portable deep research agent, has climbed...

Arabic LLMs finally face a real STEM test, and most fail hard — Inblix summary
Research

Arabic LLMs finally face a real STEM test, and most fail hard

Hugging Face Blog · Aug 1, 2025 · 3 min read

The press release for a new benchmark called 3LM opens with a diplomatic observation: Arabic LLMs have progressed, but...

VS Code can now dress you: Gradio turns Python scripts into AI shopping agents — Inblix summary
Research

VS Code can now dress you: Gradio turns Python scripts into AI shopping agents

Hugging Face Blog · Jul 31, 2025 · 2 min read

Shopping sucks. It's time-consuming, and frankly, trying on clothes in a cramped dressing room under fluorescent lights...

Hugging Face just open-sourced a free W&B replacement called Trackio — Inblix summary
Research

Hugging Face just open-sourced a free W&B replacement called Trackio

Hugging Face Blog · Jul 29, 2025 · 2 min read

Hugging Face's science team got tired of experiment trackers that cost money, lock up data, or require complex setups....

Hugging Face's 4 PB Parquet problem just got a 100x cheaper fix — Inblix summary
Research

Hugging Face's 4 PB Parquet problem just got a 100x cheaper fix

Hugging Face Blog · Jul 25, 2025 · 2 min read

Hugging Face hosts nearly 21 petabytes of datasets, and over 4 PB of that is Parquet files. That's a massive storage bi...

Hugging Face's CLI just got a radical, Docker-inspired redesign — Inblix summary
Research

Hugging Face's CLI just got a radical, Docker-inspired redesign

Hugging Face Blog · Jul 25, 2025 · 2 min read

The `huggingface-cli` is dead. Long live `hf`. Hugging Face has completely overhauled its command-line tool, slashing v...

Flux LoRA inference gets a 2.3x speed boost with new hot-swapping trick — Inblix summary
Research

Flux LoRA inference gets a 2.3x speed boost with new hot-swapping trick

Hugging Face Blog · Jul 23, 2025 · 2 min read

The team behind Hugging Face's Diffusers library has cracked a frustrating problem for anyone serving Flux text-to-imag...

TimeScope Exposes a Harsh Truth: Most AI Models Can't Really Understand Hour-Long Videos — Inblix summary
Research

TimeScope Exposes a Harsh Truth: Most AI Models Can't Really Understand Hour-Long Videos

Hugging Face Blog · Jul 23, 2025 · 2 min read

The AI industry is selling a fantasy. Every major model release now boasts about processing thousands of video frames,...

NVIDIA's single NIM container now deploys over 100,000 Hugging Face LLMs without manual config — Inblix summary
Research

NVIDIA's single NIM container now deploys over 100,000 Hugging Face LLMs without manual config

Hugging Face Blog · Jul 21, 2025 · 2 min read

NVIDIA is making a hard play to become the default engine for the open-source AI boom. The company announced that its N...

Arc's new challenge uses 220k cell profiles to build an AI that simulates CRISPR — Inblix summary
Research

Arc's new challenge uses 220k cell profiles to build an AI that simulates CRISPR

Hugging Face Blog · Jul 18, 2025 · 3 min read

Arc has thrown down a gauntlet that could fundamentally change how we do biology. The Virtual Cell Challenge asks a dec...

Forget Memorization: FutureBench Tests If AI Can Actually Predict Tomorrow — Inblix summary
Research

Forget Memorization: FutureBench Tests If AI Can Actually Predict Tomorrow

Hugging Face Blog · Jul 17, 2025 · 2 min read

Most AI benchmarks are glorified history exams. They test if a model memorized its training data or can search the web...

Gradio 5.38 adds local file support and live progress tracking for MCP servers — Inblix summary
Research

Gradio 5.38 adds local file support and live progress tracking for MCP servers

Hugging Face Blog · Jul 17, 2025 · 2 min read

The latest Gradio release solves one of the most annoying friction points for anyone using MCP servers with LLM agents:...

Multi-AI expert panels hit 85% diagnostic accuracy, crushing solo physicians — Inblix summary
Research

Multi-AI expert panels hit 85% diagnostic accuracy, crushing solo physicians

Hugging Face Blog · Jul 17, 2025 · 2 min read

A new open-source platform born from a hackathon is proving something Microsoft's research just confirmed: a panel of s...

Ettin benchmarks prove encoder models crush decoders at 4X the efficiency — Inblix summary
Research

Ettin benchmarks prove encoder models crush decoders at 4X the efficiency

Hugging Face Blog · Jul 16, 2025 · 2 min read

For years, the industry has guessed whether bidirectional encoders or causal decoders were better for non-generative ta...