AllyshipContent Radar

Deep dives 738

Paper và bài viết kỹ thuật

Crawl live-1795068900000 · 13:22 19 thg 11, 2026 · Xuất bản 13:30 19 thg 11, 2026

Kỹ thuật

langgraph-checkpoint==4.3.2

<p>Changes since checkpoint==4.3.1</p> <ul> <li>release(checkpoint): 4.3.2 (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="6382972335" data-permission-text="Title is private" data-url="https://github.com/langchain-ai/langgraph/issues/9699" data-hovercard-type="pull_request" data-hovercard-url="/langchain-ai/langgraph/pull/9699/hovercard" href="https://github.com/langchain-ai/langgraph/pull/9699">#9699</a>)</li> <li>perf(checkpoint): stop InMemoryStore search at the requested page (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="6265751896" data-permission-text="Title is private" data-url="https://github.com/langchain-ai/langgraph/issues/9633" data-hovercard-type="pull_request" data-hovercard-url="/langchain-ai/langgraph/pull/9633/hovercard" href="https://github.com/langchain-ai/langgraph/pull/9633">#9633</a>)</li> <li>perf(checkpoint): dedupe batched store reads with a dict (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="6265753880" data-permission-text="Title is private" data-url="https://github.com/langchain-ai/langgraph/issues/9636" data-hovercard-type="pull_request" data-hovercard-url="/langchain-ai/langgraph/pull/9636/hovercard" href="https://github.com/langchain-ai/langgraph/pull/9636">#9636</a>)</li> <li>chore(deps): bump the minor-and-patch group in /libs/checkpoint with 6 updates (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="6143890903" data-permission-text="Title is private" data-url="https://github.com/langchain-ai/langgraph/issues/9529" data-hovercard-type="pull_request" data-hovercard-url="/langchain-ai/langgraph/pull/9529/hovercard" href="https://github.com/langchain-ai/langgraph/pull/9529">#9529</a>)</li> <li>chore(deps-dev): bump pycryptodome from 3.23.0 to 4.2.0 in /libs/checkpoint in the major group (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="6143894542" data-permission-text="Title is private" data-url="https://github.com/langchain-ai/langgraph/issues/9531" data-hovercard-type="pull_request" data-hovercard-url="/langchain-ai/langgraph/pull/9531/hovercard" href="https://github.com/langchain-ai/langgraph/pull/9531">#9531</a>)</li> </ul>

Nguyên văn nguồn
Chi tiết nguồn
Tác tử

Import AI 477: In which I tell the IMF the Singularity has begun

Welcome to Import AI, a newsletter about AI research. Import AI runs on arXiv, cappuccinos, and feedback from readers. If you’d like to support this, please subscribe. Subscribe now Here’s a speech I gave on Saturday the 17th of October at the IMF meeting in Bangkok. This speech is inspired by the megatrend readers have seen me cover this year of AI beginning to operate at the frontier of human science, sometimes autonomously. While we all may quibble over the details, I think it’s evident we’re in the early stages of the singularity, and I wanted to use my privileged position of giving the Per Jacobsson lecture to try and tell the assembled audience of economists, bankers, and policymakers, that they should expect significant, rapid change in the future. In this speech, I focused primarily on how humanity can share in the benefits of AI if we exercise proper foresight and restraint, and I’m going to be expanding on the stakes of the singularity and the risks that we’ll all face in coming issues. As I said a few times in the speech - some of which was ad libbed and not captured here - I take my own responsibility here extremely seriously. I find myself both a narrator of what is happening in AI, a principle agent behind what is happening in AI, and a translator that goes between many communities. In all of this, I am seeking to be truthful and to be clear about my own culpability. This is a forever work-in-progress. Thank you all for reading. Speeches like this draw heavily from the newsletter, and the newsletter is my attempt to provide week-by-week information to all of you. I appreciate your readership. *** Inside the Event Horizon: Why Everyone Should Take AI Progress Seriously Hello, thank you for having me. It is a great privilege to be here. To start, I should list my relevant credentials: I am a human being. There is one of me. It has taken me almost 40 years to build the experience to give this speech. I will come back to those credentials later. My whole career has been defined by finding information that seemed important to the future of the world, then telling the world about that information.Two decades ago, I started my career as a journalist, focusing on data centers. Through that work, I became so obsessed with AI that in 2016 I left journalism to work at OpenAI, then co-founded Anthropic a few years later. Ever since, I’ve been trying to convince the world that one day, AI systems might become so powerful that they could alter national security, change the economic trajectory of the planet, and perhaps eventually gain the ability to build themselves. Well, the moment has arrived. AI is now changing national security, it is altering the economic trajectory of the world - as we’ve heard in so many sessions here - and is contributing to its own development. Of course, many of you know this, and you are having to track AI as one of a multitude of destabilizing forces that is shaping your world and your work. What I am here to explain today is that AI is on a course to bring about even more change than what we are expecting and we should prepare accordingly. And I’m going to explain why. For many years, researchers and science fiction authors have forecast a future in which AI systems drastically increase the pace of overall scientific and technical progress, making things move so quickly that it is hard for humans to keep up. They call this the singularity. I believe that this year we have gained convincing evidence that it has now begun. The writer Vernor Vinge forecast this in 1993. “When greater-than-human intelligence drives progress, that progress will be much more rapid. In fact, there seems no reason why progress itself would not involve the creation of still more intelligent entities -- on a still-shorter time scale,” he wrote, describing the singularity as “a point where our models must be discarded and a new reality rules.” The singularity is here . The implications reach across every aspect of life. But the most important implication is a simple one: we should expect that in the coming years, the whole pace of progress of not just AI, but science as a whole, will speed up. This will feel confusing and disorienting, and it will bring with it great wealth and potentially great peril. Today, I’m going to concentrate on the first point: the potential benefits of rapid AI progress, and what we need to do to share in these benefits together. So many of the ways that humanity will benefit from AI come from science. I believe that AI-driven scientific invention will change the course of the global economy, alter the ways countries plan and build their own infrastructure, make us all healthier and able to spend more time with our loved ones, and –– at the same time –– cause us all to ask deep philosophical questions about choices we should make about its continued development. I’ll develop my claim: For the last half century, most countries have dealt with falling research productivity while researcher headcount grows rapidly - ideas and inventions appear to be getting harder to find. But something important has changed: we’ve now built AI systems which have been able to augment, extend, and automate aspects of human labor. This has most obviously happened in software development, where AI systems are now more capable and efficient than the vast majority of the world’s software engineers. As a consequence, the field of software development is changing - A single person is now able to write software for anything they can imagine, creating things that previously would have taken entire teams of humans. We’ve seen concrete evidence of this inside Anthropic, where our researchers now produce almost 10 times the amount of code than they did for the first four years of the company’s life. But new AI capabilities bring new challenges - the coding improvements have meant AI systems have got much better at cybersecurity and cyberattacks –– leading the world to race to deploy these

  • But new AI capabilities bring new challenges - the coding improvements have meant AI systems have got much better at cybersecurity and cyberattacks –– leading the world to race to deploy these
Nguyên văn nguồn
Chi tiết nguồn
Kỹ thuật

datasette-cluster-map 0.19

<p><strong>Release:</strong> <a href="https://github.com/simonw/datasette-cluster-map/releases/tag/0.19">datasette-cluster-map 0.19</a></p> <p>One of the oldest Datasette plugins, and still one of my favorites. It now works with Datasette 1.0a42, and depends on <a href="https://simonwillison.net/2026/Nov/13/datasette-leaflet/">the new datasette-leaflet</a> to ensure OpenStreetMap tiles are requested with a referrer header.</p> <p>Here's <a href="https://datasette.io/global-power-plants/global-power-plants?owner=Pacific+Gas+%26+Electric+Co.">a map of power plants owned by California's PG&amp;E</a>, also demonstrating Datasette's recently <a href="https://github.com/simonw/datasette/pull/3032">redesigned facets</a>:</p> <p><img alt="Screenshot of a faceted map interface. Three facet panels at the top: &quot;country_long 1&quot; with United States of America 81 selected; &quot;owner 1&quot; with Pacific Gas &amp; Electric Co. 81 selected; &quot;primary_fuel 5&quot; listing Hydro 63, Solar 10, Gas 5, Storage 2, Nuclear 1. Each primary_fuel row has a pale blue horizontal bar behind its label, sized in proportion to its count: Hydro's bar fills most of the row width, Solar's is a short sliver, Gas's is shorter still, and Storage and Nuclear have only a thin strip at the left edge. Below is a Leaflet map with + and − zoom controls showing California and Nevada, with marker clusters across northern and central California: 9 near Redding, 21 near Chico, 2 to the west of Chico, 13 northeast of Sacramento, 7 east of Stockton, 6 at Oakland, 10 at Fresno, 9 west of Visalia, 2 north of Bakersfield, plus two individual blue pin markers on the coast, one in the far north and one north of Santa Maria. Attribution: Leaflet | © OpenStreetMap contributors" src="https://static.simonwillison.net/static/2026/power-plants-pge.webp" /></p> <p>Tags: <a href="https://simonwillison.net/tags/datasette">datasette</a>, <a href="https://simonwillison.net/tags/geospatial">geospatial</a>, <a href="https://simonwillison.net/tags/openstreetmap">openstreetmap</a></p>

Nguyên văn nguồn
Chi tiết nguồn
Kỹ thuật

v0.34.0rc3

<p>[KVConnector] Allocate the extensible KV cache exportable for EXPORTS…</p>

Nguyên văn nguồn
Chi tiết nguồn
  • v0.34.0rc3 vllm-project/vllm releases · 14 thg 11, 2026
Tác tử

v0.14.26

<h1>Release Notes</h1> <h2>[2026-11-13]</h2> <h3>llama-index-agent-agentmesh [0.3.0]</h3> <ul> <li>Python security relock: all integration uv.lock files (lock files only, no Python floor changes) (<a href="https://github.com/run-llama/llama_index/pull/24039" data-hovercard-type="pull_request" data-hovercard-url="/run-llama/llama_index/pull/24039/hovercard">#24039</a>)</li> <li>BREAKING: raise requires-python for security fixes across integrations (stacked on <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="6251082326" data-permission-text="Title is private" data-url="https://github.com/run-llama/llama_index/issues/24039" data-hovercard-type="pull_request" data-hovercard-url="/run-llama/llama_index/pull/24039/hovercard" href="https://github.com/run-llama/llama_index/pull/24039">#24039</a>) (<a href="https://github.com/run-llama/llama_index/pull/24040" data-hovercard-type="pull_request" data-hovercard-url="/run-llama/llama_index/pull/24040/hovercard">#24040</a>)</li> </ul> <h3>llama-index-callbacks-argilla [0.6.0]</h3> <ul> <li>Python security relock: all integration uv.lock files (lock files only, no Python floor changes) (<a href="https://github.com/run-llama/llama_index/pull/24039" data-hovercard-type="pull_request" data-hovercard-url="/run-llama/llama_index/pull/24039/hovercard">#24039</a>)</li> <li>BREAKING: raise requires-python for security fixes across integrations (stacked on <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="6251082326" data-permission-text="Title is private" data-url="https://github.com/run-llama/llama_index/issues/24039" data-hovercard-type="pull_request" data-hovercard-url="/run-llama/llama_index/pull/24039/hovercard" href="https://github.com/run-llama/llama_index/pull/24039">#24039</a>) (<a href="https://github.com/run-llama/llama_index/pull/24040" data-hovercard-type="pull_request" data-hovercard-url="/run-llama/llama_index/pull/24040/hovercard">#24040</a>)</li> </ul> <h3>llama-index-callbacks-arize-phoenix [0.8.0]</h3> <ul> <li>Python security relock: all integration uv.lock files (lock files only, no Python floor changes) (<a href="https://github.com/run-llama/llama_index/pull/24039" data-hovercard-type="pull_request" data-hovercard-url="/run-llama/llama_index/pull/24039/hovercard">#24039</a>)</li> <li>BREAKING: raise requires-python for security fixes across integrations (stacked on <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="6251082326" data-permission-text="Title is private" data-url="https://github.com/run-llama/llama_index/issues/24039" data-hovercard-type="pull_request" data-hovercard-url="/run-llama/llama_index/pull/24039/hovercard" href="https://github.com/run-llama/llama_index/pull/24039">#24039</a>) (<a href="https://github.com/run-llama/llama_index/pull/24040" data-hovercard-type="pull_request" data-hovercard-url="/run-llama/llama_index/pull/24040/hovercard">#24040</a>)</li> </ul> <h3>llama-index-callbacks-honeyhive [0.6.0]</h3> <ul> <li>Python security relock: all integration uv.lock files (lock files only, no Python floor changes) (<a href="https://github.com/run-llama/llama_index/pull/24039" data-hovercard-type="pull_request" data-hovercard-url="/run-llama/llama_index/pull/24039/hovercard">#24039</a>)</li> <li>BREAKING: raise requires-python for security fixes across integrations (stacked on <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="6251082326" data-permission-text="Title is private" data-url="https://github.com/run-llama/llama_index/issues/24039" data-hovercard-type="pull_request" data-hovercard-url="/run-llama/llama_index/pull/24039/hovercard" href="https://github.com/run-llama/llama_index/pull/24039">#24039</a>) (<a href="https://github.com/run-llama/llama_index/pull/24040" data-hovercard-type="pull_request" data-hovercard-url="/run-llama/llama_index/pull/24040/hovercard">#24040</a>)</li> </ul> <h3>llama-index-callbacks-langfuse [0.6.0]</h3> <ul> <li>Python security relock: all integration uv.lock files (lock files only, no Python floor changes) (<a href="https://github.com/run-llama/llama_index/pull/24039" data-hovercard-type="pull_request" data-hovercard-url="/run-llama/llama_index/pull/24039/hovercard">#24039</a>)</li> <li>BREAKING: raise requires-python for security fixes across integrations (stacked on <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="6251082326" data-permission-text="Title is private" data-url="https://github.com/run-llama/llama_index/issues/24039" data-hovercard-type="pull_request" data-hovercard-url="/run-llama/llama_index/pull/24039/hovercard" href="https://github.com/run-llama/llama_index/pull/24039">#24039</a>) (<a href="https://github.com/run-llama/llama_index/pull/24040" data-hovercard-type="pull_request" data-hovercard-url="/run-llama/llama_index/pull/24040/hovercard">#24040</a>)</li> </ul> <h3>llama-index-callbacks-literalai [1.5.0]</h3> <ul> <li>Python security relock: all integration uv.lock files (lock files only, no Python floor changes) (<a href="https://github.com/run-llama/llama_index/pull/24039" data-hovercard-type="pull_request" data-hovercard-url="/run-llama/llama_index/pull/24039/hovercard">#24039</a>)</li> <li>BREAKING: raise requires-python for security fixes across integrations (stacked on <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="6251082326" data-permission-text="Title is private" data-url="https://github.com/run-llama/llama_index/issues/24039" data-hovercard-type="pull_request" data-hovercard-url="/run-llama/llama_index/pull/24039/hovercard" href="https://github.com/run-llama/llama_index/pull/24039">#24039</a>) (<a href="https://github.com/run-llama/llama_index/pull/24040" data-hovercard-type="pull_request" data-hovercard-url="/run-llama/llama_index/pull/24040/hovercard">#24040</a>)</li> </ul> <h3>llama-index-callbacks-openinference [0.6.0]</h3> <ul> <li>Python security

Nguyên văn nguồn
Chi tiết nguồn
  • v0.14.26 run-llama/llama_index releases · 14 thg 11, 2026
Kỹ thuật

datasette-leaflet 0.3

<p><strong>Release:</strong> <a href="https://github.com/simonw/datasette-leaflet/releases/tag/0.3">datasette-leaflet 0.3</a></p> <p>OpenStreetMap have a very reasonable policy that hits to their hosted tile server <a href="https://wiki.openstreetmap.org/wiki/Referer">should send a referrer header</a>. Datasette table pages used the <code>Referrer-Policy: no-referrer</code> HTTP header, causing plugins that loaded tiles from OpenStreetMap to violate that policy.</p> <p><code>datasette-leaflet</code> is the plugin that loads a bundled Leaflet so other Datasette plugins can depend on it. It now sets the referrer policy on the tile requests themselves to <code>strict-origin-when-cross-origin</code>, passing the host but not the full path to the page.</p> <p>Tags: <a href="https://simonwillison.net/tags/leaflet">leaflet</a>, <a href="https://simonwillison.net/tags/datasette">datasette</a>, <a href="https://simonwillison.net/tags/openstreetmap">openstreetmap</a></p>

Nguyên văn nguồn
Chi tiết nguồn
Kỹ thuật

Faster Scientific Simulations with New FP64 Algorithms in NVIDIA cuBLAS

Many scientific simulations, from molecular chemistry to materials science, spend the majority of their execution time computing high-precision dense general... Many scientific simulations, from molecular chemistry to materials science, spend the majority of their execution time computing high-precision dense general matrix multiplications (GEMM). These operations are called DGEMM for real 64-bit floating-point numbers and ZGEMM for complex 64-bit floating-point numbers. cuBLAS is widely used by HPC and AI applications to accelerate GEMM and other Basic… Source

Nguyên văn nguồn
Chi tiết nguồn
Kỹ thuật

Run GPU-Powered Black Hole Interferometry with NVIDIA cuPhoton

Imaging black holes is challenging because they are distant and appear tiny in the sky—for example, M87*’s observed ring spans just 42 microarcseconds. The... Imaging black holes is challenging because they are distant and appear tiny in the sky—for example, M87*’s observed ring spans just 42 microarcseconds. The black hole itself emits no light, so telescopes capture radio emission from superheated gas swirling outside the event horizon. The black hole’s intense gravity bends the paths of this light, shaping the bright ring that frames the central… Source

Nguyên văn nguồn
Chi tiết nguồn
Kỹ thuật

The Claude Tag guide for startups

How leading startups use Claude Tag to ship faster and scale operations Startups frequently work at a fast pace with lean teams wearing many hats. In many ways, Claude Tag (beta) is suited for how startups work. You @Claude to a Slack channel where your context lives, add connectors, and have your team assign and steer tasks together. For startup teams new to using Claude Tag, one of the most frequent questions is “which channels should I add Claude to first?” To help answer it, we spoke with eight high growth startups to see how Claude Tag works inside their teams.

Nguyên văn nguồn
Chi tiết nguồn
Kỹ thuật

Adapt Human Motion for Robots with NVIDIA SOMA Retargeter

Human motion offers roboticists a great starting point for teaching humanoid robots how to move. A human demonstration captures the coordination and timing... Human motion offers roboticists a great starting point for teaching humanoid robots how to move. A human demonstration captures the coordination and timing behind natural movement, from everyday actions to complex tasks. But a robot cannot simply copy the same joint motion from a human. Its proportions, joint structure, and range of motion are different, so the movement needs to be adapted to fit… Source

Nguyên văn nguồn
Chi tiết nguồn
Kỹ thuật

When One GPU Isn’t Enough: Serving GLM-5.3 Across Two NVIDIA DGX Stations with vLLM

Running a frontier-scale model locally means keeping all its weights in GPU memory; the largest models no longer fit on one machine. Two NVIDIA DGX Stations... Running a frontier-scale model locally means keeping all its weights in GPU memory; the largest models no longer fit on one machine. Two NVIDIA DGX Stations connected over ConnectX-8 RDMA serve one model through a single endpoint, with tensor parallelism splitting the model weights and computation across both GPUs. The NVIDIA GLM-5.3 NVFP4 model, a quantized version of Z.ai’s mixture-of… Source

Nguyên văn nguồn
Chi tiết nguồn
Tác tử

Introducing ChatGPT Wallet

Introducing ChatGPT Wallet, a secure way for agents in ChatGPT Work, Codex, and dots to complete purchases with user approval and existing card protections.

Nguyên văn nguồn
Chi tiết nguồn
Kiến trúc

Release v5.23.0

<h1>Release v5.23.0</h1> <h2>New Model additions</h2> <h3>Inkling2</h3> <a target="_blank" rel="noopener noreferrer" href="https://private-user-images.githubusercontent.com/73884904/705200812-9115e142-1be3-4f79-a51d-ea0e5c25b152.png?jwt=eyJ0eXAiOiJKV1QiLCJhbGciOiJIUzI1NiJ9.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.9nW88kvLaCB-4vAwqBoE13ES0jFxwgS5MLXWBn1q4Go"><img width="1200" height="675" alt="image" src="https://private-user-images.githubusercontent.com/73884904/705200812-9115e142-1be3-4f79-a51d-ea0e5c25b152.png?jwt=eyJ0eXAiOiJKV1QiLCJhbGciOiJIUzI1NiJ9.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.9nW88kvLaCB-4vAwqBoE13ES0jFxwgS5MLXWBn1q4Go" content-type-secured-asset="image/png" style="max-width: 100%; height: auto; max-height: 675px;"></a> <p>Inkling 2 Small is a multimodal model from Thinking Machines Lab that accepts text and image inputs and generates text outputs, and it is intended for use in English and other languages and across multiple coding languages. Its language backbone has 48 mixture-of-experts decoder layers, each routing tokens to 8 of 384 experts, for 594B total and 17B active parameters. It keeps the Inkling decoder (learned relative attention, short causal convolutions, mixture-of-experts layers) and adds block attention residuals, gated attention with a learned sink, and dynamic key/value convolutions. Images are encoded by a Perception Encoder-based vision transformer and projected into the language model embeddings, one embedding per 40x40 pixel patch.</p> <p><strong>Links:</strong> <a href="https://huggingface.co/docs/transformers/main/en/model_doc/inkling2" rel="nofollow">Documentation</a> | <a href="https://thinkingmachines.ai/news/inkling-2-small/" rel="nofollow">Blog Post</a></p> <ul> <li>Add inkling2 (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="6359350142" data-permission-text="Title is private" data-url="https://github.com/huggingface/transformers/issues/50682" data-hovercard-type="pull_request" data-hovercard-url="/huggingface/transformers/pull/50682/hovercard" href="https://github.com/huggingface/transformers/pull/50682">#50682</a>) by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/guarin/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/guarin">@guarin</a> in <a href="https://github.com/huggingface/transformers/pull/50682" data-hovercard-type="pull_request" data-hovercard-url="/huggingface/transformers/pull/50682/hovercard">#50682</a></li> </ul> <h3>JEV</h3> <p>JEV is a decision head that scores single-token labels with a linear layer at the last token of the prompt, soft-capped like the logits of a language model head. <code>JevForDecision</code> is the model of the checkpoint and its head, and it reads the head files shipped next to the weights (<code>head.safetensors</code> and <code>judge_config.json</code> for JEV). The same head is also used by GEV and reads the 255 answer codes of pplx-decider and d3, and it can be run through the experimental <code>decision</code> pipeline or directly with <code>decide</code>.</p> <p><strong>Note:</strong> JEV and the decision pipeline are currently experimental. Their APIs, configuration, and behavior may change as development continues. They should be used with caution in production environments.</p> <p><strong>Links:</strong> <a href="https://huggingface.co/docs/transformers/main/en/model_doc/jev" rel="nofollow">Documentation</a></p> <ul> <li>Add Jev + decision core (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="6070865283" data-permission-text="Title is private" data-url="https://github.com/huggingface/transformers/issues/50054" data-hovercard-type="pull_request" data-hovercard-url="/huggingface/transformers/pull/50054/hovercard" href="https://github.com/huggingface/transformers/pull/50054">#50054</a>) by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/ArthurZucker/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/ArthurZucker">@ArthurZucker</a> in <a href="https://github.com/huggingface/transformers/pull/50054" data-hovercard-type="pull_request" data-hovercard-url="/huggingface/transformers/pull/50054/hovercard">#50054</a></li> </ul> <h3>LFM3</h3> <a target="_blank" rel="noopener noreferrer" href="https://private-user-images.githubusercontent.com/73884904/705205366-5a67fe26-2cee-4bbf-bc51-65e8f843b756.png?jwt=eyJ0eXAiOiJKV1QiLCJhbGciOiJIUzI1NiJ9.eyJpc3MiOiJnaXRodWIuY29tIiwiYXVkIjoicmF3LmdpdGh1YnVzZXJjb250ZW50LmNvbSIsImtleSI6ImtleTUiLCJleHAiOjE3OTUwNjk2MTcsIm5iZiI6MTc5NTA2OTMxNywicGF0aCI6Ii83Mzg4NDkwNC83MDUyMDUzNjYtNWE2N2ZlMjYtMmNlZS00YmJmLWJjNTEtNjVlOGY4NDNiNzU2LnBuZz9YLUFtei

Nguyên văn nguồn
Chi tiết nguồn
Huấn luyện

Helping robots fit into the human workplace

<p>Despite fears about robots replacing humans at work, there are still plenty of things humans can do far better than machines. That’s fine with Robust AI, a startup building robots designed to complement what humans do best.</p><p>The company’s mobile robot, named Carter, resembles a utility cart on wheels and is designed to fit seamlessly alongside humans in fast-changing warehouse and factory environments. Carter features a force-sensitive handlebar for easy guidance, an AI-powered vision system, and holonomic wheels for instant movement in any direction.</p><p>As a result, humans can cut Carter off, push and pull it in any direction with one hand, and direct it with gestures or the onboard touch screen.</p><p>The company’s robots are currently deployed in logistics warehouses and on factory floors, where they help humans while picking or stocking shelves and moving items across facilities autonomously.</p><p>“In these environments, what really determines success is how the overall system performs,” explains Anthony Jules ’92, SM ’92, who co-founded the company with MIT Professor Emeritus Rodney Brooks. “The robot does some of the work, the people do some of the work, and the backend systems do some of the work. It’s that combination that determines how well your system performs. We believe it’s really important that people have an appropriate level of agency and input so that they’re always moving the system forward.”</p><p><strong>Collaborative robots</strong></p><p>Brooks was a faculty member at MIT from 1984 until 2010 and served as the director of the Computer Science and Artificial Intelligence Laboratory (CSAIL) from 1997 until 2007. While at MIT, he co-founded iRobot, which invented the popular Roomba vacuum and went public in 2005. Later, he co-founded the company Rethink Robotics, which also designed collaborative robots.</p><p>Jules met Brooks as a graduate student at CSAIL in the 1990s. The two never worked together, but their labs were on the same floor. The pair kept in touch after Jules graduated and co-founded the company Sapient, which went public before being acquired by Publicis in 2015.</p><p>In 2019, Brooks bumped into Jules at a robotics conference. He told Jules he was thinking of starting a robotics software company, which he compared to a video game engine for training robots. Robust AI was founded later that year.</p><p>The Covid-19 pandemic caused the founders to shift to creating robots that could autonomously disinfect surfaces, but when it was discovered that Covid-19 was not transmitted via surfaces, they had to pivot again.</p><p>“That forced us to decide how to use this technology we had been building, which was really about helping robots understand the world more fully,” Jules recalls. “Long story short, I said to Rodney, ‘I have an idea. You’re going to love it and you’re going to hate it. You’re going to love it because it’s solving problems for customers that we’ve been hearing about, and you’re going to hate it because we have to build hardware.”</p><p>The founders had been interviewing potential customers for years through Robust AI’s early pivots. They finally decided on logistics and manufacturing as the industries to target early on.</p><p>“When people think about automation in factories, they think about pictures of the 5 percent that are fully automated with $50 to $200 million dollar systems,” says Jules, who now serves as Robust AI’s chief innovation officer. “They don’t think about the vast majority of factories that aren’t automated at all. These factories don’t even have a conveyor belt. They are not going to tear everything out and replace it with a $200 million system.”</p><p>Robust AI’s robots can be phased into operations as needed and offer several workflows for customers. In one workflow, the robot can guide human workers to specific spaces to retrieve items. In another it can follow humans to important spots for loading. The primary workflows involve Carter robots moving things from one place to another, often in large warehouses or factories, where they say human workers can walk on average 30,000 steps a day.</p><p>“These are huge spaces, so a person might load the robot with things and then need to bring them to another station, but that could be 400 feet away,” says, Brooks, Robust AI’s chief technology officer. “Now the person can just say ‘done’ and the robot brings it instead. How the robots are integrated into the workspace is entirely up to the customer.”</p><p><strong>Getting work done</strong></p><p>Leila Takayama, Robust AI’s vice president of human robot interaction, recently discovered a peer-reviewed study of robots in hospitals, which were used to do things like gather dishes, move sheets, and transport other equipment. The researchers discovered that, in many hospitals, a large number of robots were turned off and against the walls. The nurses said they broke.</p><p>“What really happened was the nurses were rushing with a patient on a gurney, and the robot sees them in the hallway and stops dead, and they couldn’t get the gurney around them,” Brooks says. “The robots were getting in the way of nurses’ primary tasks.”</p><p>Robust AI’s approach to technology development is to understand workers’ primary tasks deeply and build features that accelerate those tasks, whether a human or a robot is doing them.</p><p>“We let humans do what humans are good at in the physical world, which we’ve had millions of years of evolution to get good at, and have the digital system respond to that and provide us with all the things that it is good at,” Brooks explains.</p>

Nguyên văn nguồn
Chi tiết nguồn
Kiến trúc

Building always-on agents

Always-on agents run indefinitely, deciding how and when to act across many sessions. This guide walks through their core primitives and how they fit together. Over the past few months, we’ve seen a resurgence of always-on agents. Always-on agents are stateful and proactive, meaning they manage context across sessions and take actions without direct supervision. Underneath, they run on the same core loop as other agents. The difference is the surrounding architecture, which is built from primitives that control when the agents wake up, what carries over between runs, and how their work reaches the user.

Nguyên văn nguồn
Chi tiết nguồn
Tác tử

Gemini 4 Hydrogen 🚀, self-hosted Claude agents 🔒, Replit agent commerce 🛒

Fireworks Nexus gives developers access to frontier open models in the harnesses they already use. With Nexus, each request is scored and routed to the optimal model for the task, while difficult tasks can pass through to your closed provider. Set up in minutes. Keep your existing workflow. Get more control over your AI spend.

Nguyên văn nguồn
Chi tiết nguồn
341–360 / 738
Trang 18 / 37