AllyshipContent Radar

Hands-on 193

Công cụ, thử nghiệm và hướng dẫn thực hành

Crawl live-1795068900000 · 13:22 19 thg 11, 2026 · Xuất bản 13:29 19 thg 11, 2026

7

Show HN: A decent text-to-image model trained for ~$20K (Apache 2.0 Weights)

64.2

Hướng dẫn schopra909

My brother and I run a 2-person lab that trains our own generative video models from scratch. The ultimate goal is a set of creative tools that make it easier for anyone to make animation. Video models today are too slow/expensive and lack sufficient controls. Creativity is an iterative process. You make, you tweak, you get better. In our opinion, they're not usable yet for storytelling. When we set out to build Linum v3 (our next video model) earlier this year, the primary goal was to bring the cost envelope of inference down significantly. If our model is fast and cheap, it's the first step (of many) towards useful creative tools. Typically, you see "latent diffusion models". A VAE (Variational Autoencoder) compresses the input data into a smaller "latent space" and the diffusion model learns how to generate within this subspace. Over the past few months, we've been publishing our work on efficient pixel-space diffusion models. These throw away the VAE, having the diffusion model learn a latent-space optimized for its exact task. This serves 3 purposes: 1) Simplify Architecture (no 2-stage model training) 2) Faster Convergence (11x faster convergence vs. our latent diffusion counterpart) 3) Cheaper Inference (4x more aggressive token reduction in attention context-windows, without significant loss in quality) This past month we were pressure testing our latest architecture (Pyramid-JiT) by training a text-to-image model longer than the typical one-off experiment. Specifically, we trained it for ~$20K on 660M samples. We were surprised by the quality of results, so we thought we'd release it as open weights for the community (Apache 2.0). The research itself is already public, on our website. For us, it's a big stepping stone towards cheaper video models, since pretty much all video models start as image models. We like to say these models have to learn nouns before they learn verbs.

HN: open weights · 4 thg 11, 2026

Nguyên văn nguồn
Chi tiết nguồn
8

我写的《图解 Skill》的英文翻译版本即将发售

59.8

Hướng dẫn dotey

我写的《图解 Skill》的英文翻译版本即将发售 An Illustrated Guide to Agent Skills: Turn everyday repeatable tasks into reusable agentic workflows https://a.co/d/0d9fS6G8 With over 100,000 copies sold, this bestselling illustrated guide shows you how to create, test, combine, and share powerful Agent Skills for Claude Code, Codex, OpenClaw, and other compatible AI agents – all without writing code. Key Features - Learn visually with 85 skill-generated diagrams, 40 practical tables, and 67 field-tested tips - Use three ready-to-adapt skills plus a complete five-skill writing workflow - Follow one skill from first draft to a tested, production-ready data analysis project Book Description AI agents can follow a prompt once. Agent Skills help them repeat your best way of working. This illustrated guide from an early skills adopter shows you how to turn hard-won knowledge, recurring tasks, and proven processes into reusable skills that deliver more consistent results. Start by running your first skill in Claude Code, OpenClaw, or another compatible agent. Clear diagrams explain the six building blocks of an agent, how skills differ from prompts, and how on-demand loading keeps context focused. You will then create practical skills for writing style, meeting minutes, and article illustrations before combining them into a five-skill content workflow using serial, parallel, loop, and sub-agent patterns. The focus stays on outcomes. Learn when a task should become a skill, what the model should handle, when scripts are safer, and how to protect data with practical security gates. Follow a complete skill engineering process from requirements and design to testing, scoring, iteration, release, and team sharing. A three-stage data analysis project takes you from a minimal SKILL.md to a production-ready package that generates professional visual reports. No programming background is required. What you will learn - Run and inspect Agent Skills across compatible AI agents - Turn repeatable tasks and personal expertise into reusable SKILL.md files - Write descriptions that help agents select the right skill - Manage context with on-demand loading for more reliable results - Combine skills in serial, parallel, loop, and sub-agent workflows - Set secure boundaries between models, scripts, and tools - Test, score, debug, version, and share skills for personal or team use Who this book is for For knowledge workers, creators, AI power users, and developers who want agents to handle repeatable work more consistently. It suits readers using Claude Code, Codex, OpenClaw, or similar AI agents, as well as teams building reusable workflows for writing, meetings, visual content, research, and data analysis. No programming experience is required; technical readers will also value the practical guidance on scripts, testing, security, deployment, and multi-agent workflows.

X · 19 thg 11, 2026

Nguyên văn nguồn
Chi tiết nguồn
9

How to own a keyword on YouTube with Grok Bot:

59.5

Hướng dẫn gregisenberg

How to own a keyword on YouTube with Grok Bot: 1. Set up a weekly routine where your Grok Bot finds a keyword in your space that’s rising month over month (Louise connects hers to vidIQ) 2. Have it check whether anyone has made a good video on it yet 3. If they haven’t, record the video this week 4. Let Grok Bot write the title, description and chapters around that keyword, then publish Media

X · 19 thg 11, 2026

Nguyên văn nguồn
Chi tiết nguồn
10

Full Fine-Tuning (FFT) on SageMaker AI Serverless Model Customization

59.5

Hướng dẫn Sandeep Raveesh-Babu

<p>Fine-tuning is the process of continuing to train a pretrained large language model (LLM) on a company’s own data so it adapts to a specific domain, task, or style rather than relying on general-purpose knowledge alone. For businesses, the value is twofold. A fine-tuned model delivers measurably better accuracy on domain-specific work while reducing costs: a smaller specialized model can match or exceed a frontier model on your task at a fraction of the per-token inference price. Those savings compound across millions of requests. Fine-tuning has therefore become an essential process for businesses looking to get more value from LLMs at lower costs.</p> <p>Parameter-efficient fine-tuning (PEFT) methods like Low-Rank Adaptation (LoRA) have emerged as a practical solution for fine-tuning LLMs to downstream tasks, reducing GPU capacity requirements by updating only a small subset of LLM parameters. LoRA aims to strike a balance between efficiency and expressiveness, however, the low-rank nature of the weight update inherently limits the representation power of LoRA fine-tuned models, potentially compromising performance on complex tasks. This trade-off hampers the ability of LoRA methods to capture complex patterns needed for diverse tasks. To address these challenges, Full Fine-Tuning (FFT) updates model parameters simultaneously, overcoming the low-rank limitations so LLMs can fully absorb complex domain-specific knowledge.</p> <p><a href="https://aws.amazon.com/sagemaker/ai/model-customization/" target="_blank" rel="noopener">Amazon SageMaker AI Serverless Model Customization</a> now supports <a href="http://docs.aws.amazon.com/sagemaker/latest/dg/customizing-models-fft.html" target="_blank" rel="noopener">Full Fine-Tuning (FFT)</a>, so you can achieve superior model quality without managing infrastructure. There is no GPU provisioning, cluster setup, or distributed-training configuration required. In this post, we explore when and where FFT consistently outperforms LoRA across key benchmarks, examine the spectral properties of model weight matrices that explain this quality gap, provide practical hyperparameter tuning guidelines backed by controlled experiments using SageMaker AI Serverless Model Customization, and show you how to get started with FFT on SageMaker AI.</p> <h2 id="full-fine-tuning-why-it-matters">Full Fine-Tuning: Why it matters</h2> <p>Model customization types like LoRA reduce trainable parameters by freezing the pre-trained model’s weights and injecting small, separate pairs of replacement matrices (called low-rank matrices) into the model layers. This works well for simple adaptations, but the low-rank structure inherently limits the representation power of the fine-tuned model. As tasks grow more complex and datasets scale beyond tens of thousands of examples, LoRA hits an accuracy plateau where increasing rank (and thus trainable parameters) fails to bridge the accuracy gap with standard fine-tuning. This plateau reveals that the performance gap isn’t simply because of fewer parameters. It’s caused by rank deficiency itself. The low-rank update can’t capture the full complexity of the weight changes needed for deep task adaptation.</p> <p>However, for complex tasks and domains where large, good-quality datasets exist, full-rank fine-tuning excels by updating model parameters simultaneously. FFT modifies the model’s weight matrix uniformly, using the full representational capacity of the model’s weight matrices rather than restricting updates to a small subspace, and without introducing the abrupt, localized distortions characteristic of low-rank methods. The result is stronger model convergence, better activation alignment across model layers, and a model that faithfully captures both the target task distribution and the richness of its pre-trained knowledge. FFT’s quality advantage grows with dataset scale: at 100K+ training examples, FFT achieved 90.0 percent on LLM-as-Judge semantic correctness across code generation tasks, as shown in the following chart.</p> <div style="width: 810px" class="wp-caption alignnone"> <a href="https://d2908q01vomqb2.cloudfront.net/f1f836cb4ea6efb2a0b1b99f41ad8b103eff4b59/2026/11/16/ML-21644-1.png" target="_blank" rel="noopener"><img src="https://d2908q01vomqb2.cloudfront.net/f1f836cb4ea6efb2a0b1b99f41ad8b103eff4b59/2026/11/16/ML-21644-1.png" alt="LLM-as-Judge semantic correctness by training method as training data scales to 100K-plus examples" width="800"></a> <p class="wp-caption-text">Figure 1: LLM-as-Judge semantic correctness by training method as training data scales to 100K+ examples</p> </div> <p>Furthermore, the following radar chart provides a holistic view: the base model (inner polygon) shows consistently low scores across all axes. LoRA (middle polygon) shows strong, balanced improvement. FFT (outer polygon) extends further on every dimension, with particularly pronounced leads on BLEU and ROUGE-2 scores. This multi-dimensional quality lead confirms that it isn’t a metric-specific artifact but a fundamental advantage of full-rank optimization.</p> <div style="width: 810px" class="wp-caption alignnone"> <a href="https://d2908q01vomqb2.cloudfront.net/f1f836cb4ea6efb2a0b1b99f41ad8b103eff4b59/2026/11/16/ML-21644-2.png" target="_blank" rel="noopener"><img src="https://d2908q01vomqb2.cloudfront.net/f1f836cb4ea6efb2a0b1b99f41ad8b103eff4b59/2026/11/16/ML-21644-2.png" alt="Radar chart comparing the base model, LoRA, and FFT across BLEU, ROUGE, and correctness metrics" width="800"></a> <p class="wp-caption-text">Figure 2: Radar chart comparing the base model, LoRA, and FFT across quality dimensions</p> </div> <p>The details of how we arrived at these results, the coding datasets used for model customization, and the experimentation process are explained in detail in the following sections.</p> <h2 id="experimentation-and-benchmark-results">Experimentation and benchmark results</h2> <p>In this post we used the <a href="https:/

AWS Machine Learning Blog · 19 thg 11, 2026

Nguyên văn nguồn
Chi tiết nguồn
11

Introducing workload recovery and expanded infrastructure checks on Amazon SageMaker HyperPod

59.4

Hướng dẫn Roy Allela

<p>Large training jobs can fail even when every node looks healthy. A GPU with a subtle fault can pass its stress tests and still return wrong values. When a job stalls on an NCCL timeout, it keeps its GPUs, and Slurm still shows it as running. If a device plugin crashes on one node, the job fails with an error that never mentions the plugin. Teams usually find these problems only after the loss diverges or the job fails, and by then the compute is spent. Each of these failures lowers goodput, the share of your cluster’s time that goes to useful training. Today, we’re introducing workload recovery and infrastructure checks on Amazon SageMaker HyperPod that catch these failures sooner and raise the goodput of your training workloads.</p> <p>SageMaker HyperPod already handles nodes that report hardware faults. The health monitoring agent (HMA) watches each node for GPU errors, deep health checks test nodes during provisioning, and automatic node recovery reboots or replaces faulty hardware. These features look at one node’s hardware at a time. They can’t tell whether the GPUs train correctly, whether a job is still making progress, or whether the Kubernetes components a job depends on are working.</p> <p>In this post, we walk through the three new capabilities that catch these failures. For each one, we show how to set it up and what it reports when it finds a problem:</p> <ol type="1"> <li><code>MultiNodeTrainingValidation</code> is a new deep health check for HyperPod clusters orchestrated by Amazon Elastic Kubernetes Service (Amazon EKS). When HyperPod provisions an instance, or when you run the check on demand, it trains a small model on the instance and then across instances of the same type. The instance fails if the loss becomes NaN or infinite, or training doesn’t finish in time.</li> <li>The HyperPod Kubernetes Resource Monitoring Agent watches the Kubernetes layer of an EKS cluster for problems such as crash-looping device plugins, missing DaemonSets, and tainted nodes. When a job fails, it reports the pods and nodes involved as Kubernetes events and Amazon CloudWatch logs.</li> <li>Hung job detection finds Slurm jobs that have stopped making progress. An agent on each node watches the job’s output and GPU activity, or log patterns that you define. It writes an event to CloudWatch and can cancel the job so that Slurm requeues it.</li> </ol> <h2 id="why-node-health-alone-isnt-enough">Why node health alone isn’t enough</h2> <p>The existing HyperPod resiliency features check a single node at a time. The health monitoring agent (HMA) watches each node for GPU errors such as Xid and NVLink faults, Elastic Fabric Adapter (EFA) errors, and kernel errors. The InstanceStress deep health check runs DCGM level 4 diagnostics and tests each instance’s GPUs, GPU memory, NVLink, and EFA devices. The InstanceConnectivity check runs NCCL tests across GPU instances, or NCCOM tests on AWS Trainium. Together, these confirm that each node’s hardware works. They don’t show whether the GPUs compute correctly, whether a job is still making progress, or whether the cluster’s Kubernetes components are healthy. Three kinds of failure get past them:</p> <ul> <li><strong>Numerical faults –</strong>&nbsp;A GPU can pass every hardware test and still compute wrong results. The first sign is often NaN or infinite loss in a production run. Meta attributed 6 of the 419 unexpected interruptions in its 54-day Llama 3 405B pre-training run to silent data corruption (<a href="https://arxiv.org/abs/2407.21783" target="_blank" rel="noopener">The Llama 3 Herd of Models</a>).</li> <li><strong>Hung jobs –</strong>&nbsp;An NCCL timeout, a stalled data pipeline, or a deadlock can stop a job without ending it. The job keeps its GPUs and their memory, and Slurm reports it as running until someone checks the loss curve or the log.</li> <li><strong>Cluster-level failures –</strong>&nbsp;Some failures happen in Kubernetes, not in a node’s hardware. A device plugin crash-loops, a required DaemonSet never starts on a new node, or ranks on different nodes fall out of step in an NCCL collective. HMA checks each node on its own, so it doesn’t see these problems. When the job fails, the error often doesn’t say which pod or node caused it.</li> </ul> <h2 id="solution-overview">Solution overview</h2> <p>Each capability covers a different point in a training run. <code>MultiNodeTrainingValidation</code> runs when instances are provisioned, or on demand, and confirms that their GPUs train correctly. Hung job detection runs alongside each Slurm job and catches a job that has stopped making progress. The HyperPod Kubernetes Resource Monitoring Agent watches the cluster continuously and, when a job fails, reports the pods and nodes that were involved. The following table summarizes the three.</p> <table border="1px" width="100%" cellpadding="10px"> <tbody> <tr> <td><strong>Capability</strong></td> <td><strong>What it catches</strong></td> <td><strong>When it runs</strong></td> <td><strong>Available on</strong></td> </tr> <tr> <td><code>MultiNodeTrainingValidation</code> deep health check</td> <td>NaN or infinite loss, or training that doesn’t finish in time, on GPUs that pass hardware checks</td> <td>When instances are created or updated, or on demand</td> <td>EKS clusters, on supported EFA-enabled NVIDIA GPU instances</td> </tr> <tr> <td>HyperPod Kubernetes Resource Monitoring Agent</td> <td>Node condition changes, pod and DaemonSet failures, and failed training jobs, with the pods and nodes involved. NCCL communication state for jobs that opt in</td> <td>Continuously</td> <td>EKS clusters</td> </tr> <tr> <td>Hung job detection</td> <td>Jobs that stop making progress but don’t exit</td> <td>While each job runs</td> <td>Slurm clusters</td> </tr> </tbody> </table> <p>Hung job detection is new for Slurm clusters. On EKS clusters, HyperPod already detects hung jobs that run through the <a href="https://docs.aws.amazon.com/sagemaker/latest/dg/sag

AWS Machine Learning Blog · 19 thg 11, 2026

Nguyên văn nguồn
Chi tiết nguồn
12

Introducing the Amazon SageMaker JumpStart Model Leaderboard

59.3

Hướng dẫn Dan Ferguson

<p>Amazon SageMaker JumpStart gives you hundreds of open-weight foundation models you can deploy quickly. That choice is powerful, but it also creates a real question: Which model should you start with? Comparing models usually means reading model cards one at a time and hunting for benchmark scores that different sources published under different conditions. Then you still must guess how each model will perform, and what it will cost, for your use case.</p> <p>Today we are launching the <strong>Amazon SageMaker JumpStart Model Leaderboard</strong>, a new way to compare <a href="https://console.aws.amazon.com/sagemaker/home#/launch?source=jumpstart&amp;page=jumpstart" target="_blank" rel="noopener">JumpStart models</a> side by side on quality, speed, and cost, directly in Amazon SageMaker Studio. In this post, we explain what the leaderboard shows, define the key terms in plain language, and walk through how to go from a shortlist to a deployed model.</p> <h2 id="what-is-the-jumpstart-model-leaderboard">What is the JumpStart Model Leaderboard?</h2> <p>The leaderboard is a new view of the JumpStart base models page in SageMaker Studio, right next to the Categories and All models views. It uses the same search bar and filters you already know, and it opens on a leaderboard of large language models (LLMs). You see rankings immediately without answering setup questions first. For each model, it shows two kinds of information:</p> <ol type="1"> <li><strong>Quality scores</strong> on three public benchmarks (BIG-Bench Hard, GPQA Diamond, and MMLU-Pro) that AWS runs itself, using the same evaluation setup for every model so the numbers are comparable.</li> <li><strong>Performance and cost metrics</strong> that AWS measures by sending a standardized test workload to an endpoint using each JumpStart deployment configuration, so you see how fast a model responds and what it costs to run.</li> </ol> <p>Because both kinds of numbers come from the same place and the same method, you can make a like-for-like comparison instead of stitching together data from different websites. At launch, the leaderboard covers <strong>70+ models</strong> from providers including Meta, Google, Qwen, IBM, and Arcee, with more added over time.</p> <p>The following screenshot shows the leaderboard in SageMaker Studio, with benchmark scores, performance metrics, and estimated cost for each model.</p> <div style="width: 810px" class="wp-caption alignnone"> <a href="https://d2908q01vomqb2.cloudfront.net/f1f836cb4ea6efb2a0b1b99f41ad8b103eff4b59/2026/11/18/ML-22496-1-1.png" target="_blank" rel="noopener"><img src="https://d2908q01vomqb2.cloudfront.net/f1f836cb4ea6efb2a0b1b99f41ad8b103eff4b59/2026/11/18/ML-22496-1-1.png" alt="SageMaker JumpStart Model Leaderboard in SageMaker Studio showing benchmark scores, performance metrics, and estimated cost per model" width="800"></a> <p class="wp-caption-text">Figure 1: Leaderboard overview in SageMaker Studio</p> </div> <h2 id="key-concepts">Key concepts</h2> <p>This section defines the terms the leaderboard uses: the quality benchmarks, the use case profiles and optimization goals that shape the performance numbers, and the performance metrics themselves.</p> <h3 id="quality-benchmarks">Quality benchmarks</h3> <p>A benchmark is a standardized test: a fixed set of questions with known answers. Running every model through the same test gives you a score you can compare. At launch, the leaderboard reports three reasoning benchmarks for LLMs:</p> <table border="1px" width="100%" cellpadding="10px"> <tbody> <tr> <td><strong>Benchmark</strong></td> <td><strong>What it measures, in plain terms</strong></td> </tr> <tr> <td><strong>BIG-Bench Hard (BBH)</strong></td> <td>A set of challenging tasks that require multi-step reasoning, such as logical deduction, tracking objects, and multi-step arithmetic.</td> </tr> <tr> <td><strong>GPQA Diamond</strong></td> <td>Graduate-level science questions that are difficult to answer by searching the web. A good signal of deep reasoning.</td> </tr> <tr> <td><strong>MMLU-Pro</strong></td> <td>Broad knowledge and reasoning across many subjects, such as science, law, and engineering, using harder multiple-choice questions.</td> </tr> </tbody> </table> <p>Higher is better for all three. The leaderboard shows each benchmark separately rather than blending them into one number. That keeps the results transparent: You can weigh the benchmarks that matter for your application and check each score against the public definition of the benchmark.</p> <h3 id="use-case-profiles-and-optimization-goals">Use case profiles and optimization goals</h3> <p>Speed and cost depend on your workload and on how the model is deployed. Two menus above the leaderboard control which performance and cost numbers you see:</p> <ol type="1"> <li><strong>Use case profile.</strong> Select the profile closest to your workload from the View benchmarks for your use case menu. The default is Max Context. The profile changes only the performance and cost metrics. Quality benchmark scores are the same for every profile.</li> <li><strong>Optimization goal.</strong> For each profile, a model can have several deployment configurations, and the leaderboard shows metrics for one at a time. Choose which one with the Optimized for menu. The default is Balanced.</li> </ol> <p>Each optimization goal picks a different configuration:</p> <table border="1px" width="100%" cellpadding="10px"> <tbody> <tr> <td><strong>Optimization goal</strong></td> <td><strong>Configuration shown</strong></td> </tr> <tr> <td><strong>Balanced</strong></td> <td>The configuration with the lowest estimated cost per token. Because cost per token depends on both instance price and throughput, this balances what you pay per hour against how fast the instance generates tokens.</td> </tr> <tr> <td><strong>Cost optimized</strong></td> <td>The configuration with the lowest hourly instance price. Latency can be higher and throughput lower

AWS Machine Learning Blog · 19 thg 11, 2026

Nguyên văn nguồn
Chi tiết nguồn
7–12 / 106
Trang 2 / 18