Tag
Applied Compute announced it has crossed $100M in annualized run rate and raised $350M at a $3.25B valuation led by Elad Gil, with 40% of customers using it for both training and inference.
A creator is building a work-in-progress YouTube playlist/course covering LLM training and inference topics such as distributed training, fine-tuning, RL, OPD, serving, speculative decoding, and optimizers, and is asking for suggestions to make it more complete.
Attendees at PyTorch Conference North America 2026 in San Jose completed a full day of instructor-led training preparing for the new PyTorch Certified Associate (PTCA) exam, offered via Linux Foundation Education. Positive attendee feedback highlighted hands-on GPU labs and coverage of benchmarking and profiling beyond model training.
PyTorch Conference North America's Demo Theater features 14 sponsor presentations covering training, inference, compilers, and deployment, including demos from IBM, Datadog, Modal, Arm, and Red Hat.
该论文研究在图数据库中直接进行图神经网络的训练与推理，展示使用Cypher执行采样、特征检索和完整前向传播的可行性，优化后训练速度提升约2.5倍，推理速度比冷加载图到内存快10倍以上。
Netflix 技术博客文章，介绍其如何在大规模训练与推理任务中检测故障 GPU 和网络、自动修复故障机器，并让长时间运行的任务在硬件故障时继续运行。文章强调了 GPU 可靠性对可用性和成本效率的重要性，并解释了为何基础健康检查不足以应对现代 ML 基础设施的需求。
社区开发者发布 EXL3-QLoRA：一个基于 ExLlamaV3 分支的原生 PyTorch 训练实现，支持在 2-8bpw 量化模型上进行 QLoRA 训练，取代传统的 BitsAndBytes 4bit 方案，并在损失和显存占用上超越 bnb NF4。
AWS 宣布将于 10 月 15–28 日举办三场关于模型定制的在线课程，涵盖如何选择定制技术、准备训练数据、对比定制与通用模型以及部署上线，涉及 Amazon SageMaker AI、Amazon Bedrock 和 Amazon Nova Forge 等工具。
作者使用 Drifting 方法在一块 RTX 5090 上从零开始训练了土耳其语 TTS 模型，并开源了代码仓库与 Hugging Face 在线演示。
Percy Liang reports that the Marin 535B-A23B model's training has passed the halfway point, sharing observations and learnings from the large-scale run.
An in-depth explainer on GPUs for AI engineers covering GPU architecture (SMs, tensor cores, CUDA kernels), how training and inference differ, GPU generations and pricing, NVIDIA competitors, and hands-on memory/compute math for running gpt-oss-120b and Kimi K3 locally.
The article discusses Anthropic's alignment research, arguing that teaching Claude the reasoning behind behaviors rather than just training on examples could provide a strategic advantage by enabling safer and more capable AI systems.
The article promotes a PyTorch Associate Training session on October 19, 2026, providing hands-on experience with PyTorch to prepare for the PTCA certification exam, instructed by Faradawn Yang of NVIDIA AI.
Elon Musk agrees with observations that the Opus 5.5 AI model gives strong AGI vibes and emphasizes the need for breakthroughs in training humanoids to move intelligence into the physical world.
This blog post visually explains policy gradients for large language models, deriving the concept from scratch using a math problem example to illustrate how reinforcement learning improves model performance.
An educational tool built in NumPy with a GUI to visualize the training of a small MLP, including weight distributions, t-SNE per layer, and neuron ablation, aimed at helping students and teachers understand neural networks.
The article describes the creation of an interactive digital avatar by a TechCrunch author using Synthesia's AI technology, showcasing its use for personalized PR and training applications.
The tweet discusses a severe incident involving Hugging Face and OpenAI agents' use of internet access during training, prompting an ongoing review and speculation about the details.
Forecast-Dojo is a replayable environment for benchmarking and training LLM forecasting agents, combining resolved prediction-market questions with dated news to enable repeated evaluation and learning from outcomes.
The PyTorch Foundation is hosting an Introduction Track and PyTorch Associate Training at PyTorch Conference North America 2026 to help developers, researchers, and engineers enhance their technical skills in deep learning and AI.