A developer fully fine-tuned a 270M-parameter Gemma 3 model on a laptop using the Banking77 dataset, achieving ~96% accuracy on intent classification. The project used full fine-tuning with loss maski…
A developer trained a large language model from scratch using plain PyTorch, implementing the full post-training pipeline including SFT, reward modeling, DPO, PPO, and GRPO on public datasets, all run…
Alloy, a new PyTorch backend and inference engine for Apple Silicon, has been released as a technical preview. The open-source project compiles Python GPU kernels to Metal and supports LLM serving wit…
A developer using PyTorch and Lightning AI demonstrated automated neural network training with GPU acceleration. By setting the trainer's accelerator and devices to 'auto', Lightning automatically det…
Apple introduced Core AI at WWDC 26 as the successor to Core ML, designed to run large language models and generative AI on-device using Apple Silicon. The framework segments Apple's ML ecosystem into…
Apple announced Core AI, a new framework for running generative AI models on-device, at WWDC 26. The framework supports large language models up to 70B parameters and runs exclusively on Apple Silicon…
A developer built Omni-VRAM, an open-source voice AI platform with 28 modules. The platform includes speech recognition with five Whisper backends, real-time streaming under 200ms latency, speaker dia…
Sasha Rush released GPU Puzzles, an interactive notebook teaching beginner GPU programming through coding exercises that map Python to CUDA kernels using NUMBA. The puzzles aim to build intuition for …
A Hacker News user asks whether programmers will write more efficient code during a memory shortage, sparking debate about trends in software efficiency. Commenters express skepticism, noting that abu…
Meta's LLMs have evolved from simple Transformer stacks to complex architectures with multiple attention variants, mixture-of-experts, and multimodal encoders, mirroring the complexity of recommendati…
Neo launched an MCP server that lets users offload AI tasks from Claude, reducing costs by 62% and speeding up runtime by 37% in benchmarks. The tool integrates with Claude Code and other MCP clients,…
Meta released Pyrefly 1.0, a fast Python linter and type checker for PyTorch and Django projects. Python 3.15 introduces a sampling profiler for efficient code instrumentation. A new malware called Ha…
Docker solves the 'it works on my machine' problem for AI engineers by packaging all dependencies—OS, Python, packages, code, and config—into a single portable image. A typical AI project has multiple…
Meta researchers introduced an LLM-guided autotuner for Helion, PyTorch's domain-specific language for performance portable machine learning kernels, that matches the performance of the existing Likel…
The Linux Foundation Education and PyTorch Foundation have launched the PyTorch Certified Associate (PTCA) certification, validating foundational skills in PyTorch for AI and machine learning practiti…
The schedule for KubeCon + CloudNativeCon + OpenInfra Summit + PyTorch Conference China, taking place September 7-9 in Shanghai, is now available. The event features sessions on AI agents, GPU virtual…
A developer built a high-precision food nutrition AI engine by combining Meta's Segment Anything Model (SAM) for pixel-perfect object isolation and GPT-4o Vision for multi-modal reasoning and volume e…
KerasFormers, an open-source library built entirely in Keras 3, launches with over 100 transformer models spanning vision, language, multimodal, and speech, supporting seamless execution on TensorFlow…
Tensordyne is developing AI chips that use a logarithmic number system to multiply numbers more efficiently than conventional floating-point math, aiming to improve compute efficiency for AI workloads…
By 2026, AI workloads will consume roughly 40% of enterprise Kubernetes clusters, but the default scheduler is ill-suited for GPU-intensive tasks, leading to 30-45% GPU utilization rates and wasted co…