Change
3cd00e3
3cd00e34cf525824ccc0283fe2ffbf9868684800 · commit on GitHub
pytorch-blog-feed: changed (339699 bytes, HTTP 200)
raw/pytorch-blog-feed/response.xml modified
- Source
- pytorch-blog-feed
- Lines added
- +116
- Lines removed
- -36
- Stored bytes at this commit
- 339,699
- Timestamp
- origin
- Raw artifact at this commit
- raw/pytorch-blog-feed/response.xml
Recorded headers
| observed_at | 2026-10-29T05:48:14.827Z |
|---|---|
| origin_date | 2026-10-29T03:44:38.000Z |
| status | 200 |
| final URL | https://pytorch.org/blog/feed/ |
| etag | "8eb1be55422e55f4a40b34fcabc8282c" |
| last-modified | Wed, 28 Oct 2026 18:19:21 GMT |
| date | Thu, 29 Oct 2026 05:48:14 GMT |
| age | 7416 |
| cache-control | public, max-age=60, s-maxage=43200, stale-while-revalidate=86400, stale-if-error=604800 |
| cf-cache-status | null |
| content-encoding | null |
| content-length | 339699 |
@
@@ -12,7 +12,7 @@ <atom:link href="https://pytorch.org/blog/feed/" rel="self" type="application/rss+xml" /> <link>https://pytorch.org/blog/</link> <description></description>-
<lastBuildDate>Mon, 26 Oct 2026 18:27:27 +0000</lastBuildDate>+
<lastBuildDate>Wed, 28 Oct 2026 17:44:20 +0000</lastBuildDate> <language>en-US</language> <sy:updatePeriod> hourly </sy:updatePeriod>@
@@ -28,6 +28,119 @@ <height>32</height></image> <item>+
<title>PyTorch 2.15 Release Blog</title>+
<link>https://pytorch.org/blog/pytorch-2-15-release-blog/</link>+
+
<dc:creator><![CDATA[PyTorch Foundation]]></dc:creator>+
<pubDate>Wed, 28 Oct 2026 17:44:20 +0000</pubDate>+
<category><![CDATA[Announcements]]></category>+
<category><![CDATA[Blog]]></category>+
<guid isPermaLink="false">https://pytorch.org/?p=208458</guid>+
+
<description><![CDATA[<p>We are excited to announce the release of PyTorch® 2.15 (release notes)! The PyTorch 2.15 release features the following changes: Rewritten NCCL backend becomes the default, with a flight recorder...</p>+
<p>The post <a href="https://pytorch.org/blog/pytorch-2-15-release-blog/">PyTorch 2.15 Release Blog</a> appeared first on <a href="https://pytorch.org">PyTorch</a>.</p>+
]]></description>+
<content:encoded><![CDATA[<p><span style="font-weight: 400;">We are excited to announce the release of PyTorch® 2.15 (</span><a href="https://github.com/pytorch/pytorch/releases/tag/v2.15.0"><span style="font-weight: 400;">release notes</span></a><span style="font-weight: 400;">)!</span></…+
<p><span style="font-weight: 400;">The PyTorch 2.15 release features the following changes:</span></p>+
<ul>+
<li style="font-weight: 400;" aria-level="1"><b>Rewritten NCCL backend becomes the default,</b><span style="font-weight: 400;"> with a flight recorder that dumps on hangs and </span><span style="font-weight: 400;"><code class="language-python">TORCH_DIST_USE_NCCL2=0</code></span><span style="font-we…+
<li style="font-weight: 400;" aria-level="1"><b>One-sided transports for torch.distributed,</b><span style="font-weight: 400;"> with a NIXL backend and CUDA-graph-capturable transfers</span></li>+
<li style="font-weight: 400;" aria-level="1"><b>FSDP2 and pipelining gain finer memory and communication control, </b><span style="font-weight: 400;">with fp32 for precision-sensitive parameters, CUDA-graph-captured gradient accumulation, and gradient reduction overlapped with compute </span></li>+
<li style="font-weight: 400;" aria-level="1"><span style="font-weight: 400;"> </span><b>Faster 4-bit (NVFP4) LLM decoding on NVIDIA Blackwell GPUs</b><span style="font-weight: 400;"> through fewer, better-overlapped kernel launches in NVGEMM</span></li>+
<li style="font-weight: 400;" aria-level="1"><b>Apple Silicon gains FP8 and faster decode,</b><span style="font-weight: 400;"> with </span><span style="font-weight: 400;"><code class="language-python">float8_e4m3fn</code></span><span style="font-weight: 400;"> and </span><span style="font-weight: 40…+
<li style="font-weight: 400;" aria-level="1"><b>Broader platform support</b><span style="font-weight: 400;"> with FlyDSL kernels on AMD gfx950, Tensor Data Mover support for gfx1250 (MI450), ROCm 10.0 wheels, simpler SYCL kernel authoring and multi-GPU profiling on Intel GPUs, CUDA 13.2 as the defau…+
</ul>+
<p><span style="font-weight: 400;">This release is composed of 3,117 commits from 583 contributors since PyTorch 2.14. We want to sincerely thank our dedicated community for your contributions. As always, we encourage you to try these out and report any issues as we improve 2.15. More information ab…+
<p>Bring any questions you may have about PyTorch 2.15 to our Q&A webinar on Thursday, November 5, 2026, at 11 AM PT. Tristan Rice, Natalia Gimelshein, Huy Do, and Jane Xu (all Meta) will share an overview of the 2.15 release and answer community questions about PyTorch and the new capabilities …+
<p><span style="font-weight: 400;">Throughout the 2.x series, PyTorch has been evolving from a research-first framework into a unified, hardware-agnostic platform for production training and inference at scale. </span><a href="https://pytorch.org/blog/pytorch-2-13-release-blog/"><span style="font-we…+
<p><span style="font-weight: 400;">PyTorch 2.15 builds on that work. The rewritten NCCL backend now backs every </span><span style="font-weight: 400;"><code class="language-python">nccl</code></span><span style="font-weight: 400;"> process group, NVGEMM is tuned for low-precision decode, Apple Silic…+
<h2><span style="font-weight: 400;">Distributed Training</span></h2>+
<h3><span style="font-weight: 400;">The Rewritten NCCL Backend Becomes the Default</span></h3>+
<p><span style="font-weight: 400;">When a model trains across many GPUs, those GPUs constantly exchange data through a communication backend, and on NVIDIA hardware that backend is NCCL. In PyTorch 2.15, every job that uses </span><span style="font-weight: 400;"><code class="language-python">nccl</c…+
</span><span style="font-weight: 400;"><br />+
</span><span style="font-weight: 400;">PyTorch 2.14 previewed our rewritten NCCL backend, ported from torchcomms, as the opt-in </span><span style="font-weight: 400;"><code class="language-python">nccl2</code></span><span style="font-weight: 400;"> backend. In 2.15 it is the default for every </span…+
<p><span style="font-weight: 400;">To make the backend debuggable in production, a backend-agnostic flight recorder writes a post-mortem to disk when a collective hangs, and watchdog health is published to the debug server so a timeout on any rank triggers flight-recorder dumps from every rank. At c…+
<p><span style="font-weight: 400;">API Unstable</span></p>+
<p><span style="font-weight: 400;">(PRs </span><a href="https://github.com/pytorch/pytorch/pull/192281"><span style="font-weight: 400;">#192281</span></a><span style="font-weight: 400;">, </span><a href="https://github.com/pytorch/pytorch/pull/192895"><span style="font-weight: 400;">#192895</span></…+
<h3><span style="font-weight: 400;">One-Sided Transports with a NIXL Backend</span></h3>+
<p><span style="font-weight: 400;">torch.distributed gains a Python transport API for one-sided data movement, with NIXL as the first backend. Transports connect to a peer through a store, without needing a process group or collectives, and </span><span style="font-weight: 400;"><code class="languag…+
<p><span style="font-weight: 400;">API Unstable</span></p>+
<p><span style="font-weight: 400;">(PRs </span><a href="https://github.com/pytorch/pytorch/pull/195643"><span style="font-weight: 400;">#195643</span></a><span style="font-weight: 400;">, </span><a href="https://github.com/pytorch/pytorch/pull/197032"><span style="font-weight: 400;">#197032</span></…+
<h3><span style="font-weight: 400;">FSDP2 and Pipeline Parallelism: Memory and Communication Control</span></h3>+
<p><span style="font-weight: 400;">FSDP2 now gives finer control over precision and capture. Precision-sensitive parameters, such as MoE router gates and norm weights, can stay in fp32 while the rest of the model trains in bf16, still with one all-gather per <code class="language-python">fully_shard…+
<p><span style="font-weight: 400;">Pipeline parallelism now allocates receive buffers just before each receive instead of for every microbatch up front, and releases forward outputs once their sends complete. Schedules accept a </span><span style="font-weight: 400;"><code class="language-python">max…+
</span></p>+
<p><span style="font-weight: 400;">API Unstable</span></p>+
<p><span style="font-weight: 400;">(PRs </span><a href="https://github.com/pytorch/pytorch/pull/196640"><span style="font-weight: 400;">#196640</span></a><span style="font-weight: 400;">, </span><a href="https://github.com/pytorch/pytorch/pull/196724"><span style="font-weight: 400;">#196724</span></…+
<h2><span style="font-weight: 400;">Performance Improvements</span></h2>+
<h3><span style="font-weight: 400;">Faster MPS Kernels for Decode and Linear Algebra</span></h3>+
<p><span style="font-weight: 400;">Building on the 2.14 migration, more MPS operators move from MPSGraph to native Metal kernels. A single-row softmax over a 128,256-entry vocabulary drops by 8x, upsample backward is up to 21.7x faster, and reductions now share one set of kernels, making </span><spa…+
<p><span style="font-weight: 400;">For decode, </span><span style="font-weight: 400;"><code class="language-python">F.linear</code></span><span style="font-weight: 400;"> now uses the existing gemv kernels (1.23x geometric-mean speedup, up to 5x on small projections), and the new MPS </span><span st…+
<p><span style="font-weight: 400;">(PRs </span><a href="https://github.com/pytorch/pytorch/pull/196071"><span style="font-weight: 400;">#196071</span></a><span style="font-weight: 400;">, </span><a href="https://github.com/pytorch/pytorch/pull/199687"><span style="font-weight: 400;">#199687</span></…+
<h2><span style="font-weight: 400;">Core Features</span></h2>+
<h3><span style="font-weight: 400;">CUDA 13.2 by Default and CUDA 13.4 Wheels</span></h3>+
<p><span style="font-weight: 400;">The default PyTorch wheel moves to CUDA 13.2 (13.2.2 toolkit), with CUDA 13.4 GA wheels shipping alongside it for Linux x86_64, Linux aarch64 and Windows. The CUDA 13.4 wheels add experimental support for NVIDIA Rubin GPUs.</span></p>+
<p><span style="font-weight: 400;">(PRs </span><a href="https://github.com/pytorch/pytorch/pull/198913"><span style="font-weight: 400;">#198913</span></a><span style="font-weight: 400;"> and </span><a href="https://github.com/pytorch/pytorch/pull/196014"><span style="font-weight: 400;">#196014</span…+
<h3><span style="font-weight: 400;">ROCm 10.0 and TheRock Wheels</span></h3>+
<p><span style="font-weight: 400;">PyTorch 2.15 adds ROCm 10.0 wheels alongside the ROCm 7.14 wheels introduced in 2.14. All ROCm wheels are now built on AMD’s TheRock packages, and they install ROCm as a pip dependency (the matching rocm package) instead of bundling the ROCm libraries.</span>…+
<p><span style="font-weight: 400;">(PRs </span><a href="https://github.com/pytorch/pytorch/pull/194919"><span style="font-weight: 400;">#194919</span></a><span style="font-weight: 400;">, </span><a href="https://github.com/pytorch/pytorch/pull/198895"><span style="font-weight: 400;">#198895</span></…+
<h3><span style="font-weight: 400;">Python 3.11 to 3.15, and torch.compile on Python 3.15</span></h3>+
<p><span style="font-weight: 400;">PyTorch 2.15 supports Python 3.11 through 3.15, including the free-threaded 3.14t and 3.15t builds. In 2.15, Dynamo adds Python 3.15 support.</span></p>+
<p><span style="font-weight: 400;">(PRs </span><a href="https://github.com/pytorch/pytorch/pull/178393"><span style="font-weight: 400;">#178393</span></a><span style="font-weight: 400;">, </span><a href="https://github.com/pytorch/pytorch/pull/194117"><span style="font-weight: 400;">#194117</span></…+
<h2><span style="font-weight: 400;">Platform Features and Updates</span></h2>+
<h3><span style="font-weight: 400;">CUDA</span></h3>+
<h4><span style="font-weight: 400;">NVGEMM for NVFP4 Decode</span></h4>+
<p><span style="font-weight: 400;">NVGEMM, the CuTeDSL GEMM backend added in 2.14, now folds the scalar dequantization factor of ModelOpt NVFP4 linear layers into the GEMM, removing a separate pointwise launch per layer during decode. A bounded, shape-aware candidate policy covers Blackwell NVFP4 de…+
<p><span style="font-weight: 400;">API Unstable</span></p>+
<p><span style="font-weight: 400;">(PRs </span><a href="https://github.com/pytorch/pytorch/pull/194651"><span style="font-weight: 400;">#194651</span></a><span style="font-weight: 400;">, </span><a href="https://github.com/pytorch/pytorch/pull/194652"><span style="font-weight: 400;">#194652</span></…+
<h3><span style="font-weight: 400;">ROCm</span></h3>+
<h4><span style="font-weight: 400;">FlyDSL Kernels on gfx950 and Tensor Data Mover on gfx1250</span></h4>+
<p><span style="font-weight: 400;">FlyDSL joins CuTeDSL, Triton and Helion as a kernel DSL that PyTorch can use. On gfx950, Inductor autotunes FlyDSL templates for GEMM, ragged </span><span style="font-weight: 400;"><code class="language-python">F.grouped_mm</code> and MXFP4/MXFP8 </span><span style…+
<p><span style="font-weight: 400;">On gfx1250 (MI450), Inductor uses the Tensor Data Mover for dense </span><span style="font-weight: 400;">mm</span><span style="font-weight: 400;">/</span><span style="font-weight: 400;">addmm</span><span style="font-weight: 400;">, pointwise and reduction codegen, …+
<p><span style="font-weight: 400;">API Unstable</span></p>+
<p><span style="font-weight: 400;">(PRs </span><a href="https://github.com/pytorch/pytorch/pull/192877"><span style="font-weight: 400;">#192877</span></a><span style="font-weight: 400;">, </span><a href="https://github.com/pytorch/pytorch/pull/190903"><span style="font-weight: 400;">#190903</span></…+
<h3><span style="font-weight: 400;">MPS (Apple Silicon)</span></h3>+
<h4><span style="font-weight: 400;">FP8, Streams and faster MoE inference on Apple Silicon</span></h4>+
<p><code class="language-python">torch.float8_e4m3fn</code> is now supported on MPS, along with <code class="language-python">_scaled_mm</code> with tensorwise or rowwise scaling and a gemv fast path for decode-sized shapes. MPS also gains a public stream API modeled on CUDA’s: <code class="la…+
<p><span style="font-weight: 400;">API Unstable</span></p>+
<p><span style="font-weight: 400;">(PRs </span><a href="https://github.com/pytorch/pytorch/pull/194074"><span style="font-weight: 400;">#194074</span></a><span style="font-weight: 400;">, </span><a href="https://github.com/pytorch/pytorch/pull/196916"><span style="font-weight: 400;">#196916</span></…+
<h3><span style="font-weight: 400;">XPU (Intel GPUs)</span></h3>+
<h4><span style="font-weight: 400;">SYCL Kernel Programming Simplified</span></h4>+
<p><span style="font-weight: 400;">This release introduces SYCL free-function kernel support for the XPU backend, with most existing SYCL kernels already migrated to the new programming model. By replacing verbose functor-based implementations with explicit free-function kernels, developers can writ…+
</span></p>+
<p><span style="font-weight: 400;">(PRs </span><a href="https://github.com/pytorch/pytorch/pull/196387"><span style="font-weight: 400;">#196387</span></a><span style="font-weight: 400;">, </span><a href="https://github.com/pytorch/pytorch/pull/196932"><span style="font-weight: 400;">#196932</span></…+
<h4><span style="font-weight: 400;">Enhanced XPU Runtime Capabilities</span></h4>+
<p><span style="font-weight: 400;">This release expands XPU runtime capabilities with IPC support for torch.xpu.Event, enabling event sharing across processes and integration with torch.multiprocessing. In addition, torch.xpu.MemPool now supports use_on_oom and non_split allocation policies, providi…+
</span></p>+
<p><span style="font-weight: 400;">API Unstable</span></p>+
<p><span style="font-weight: 400;">(PRs</span><a href="https://github.com/pytorch/pytorch/pull/191182"> <span style="font-weight: 400;">#191182</span></a><span style="font-weight: 400;">,</span><a href="https://github.com/pytorch/pytorch/pull/192495"> <span style="font-weight: 400;">#192495</span></…+
<h4><span style="font-weight: 400;">Improved XPU Profiling Experience</span></h4>+
<p><span style="font-weight: 400;">This release adds kernel-level visibility into XPU Graph replay, cross-rank correlation for XCCL collectives, and per-kernel hardware counter collection across all Intel GPUs in multi-GPU workloads. These enhancements enable deeper performance analysis for graph-ba…+
<p><span style="font-weight: 400;">(PR</span><a href="https://github.com/pytorch/pytorch/pull/195128/"> <span style="font-weight: 400;">#195128</span></a><span style="font-weight: 400;"> by Chuanqi Wang, Intel; PR</span><a href="https://github.com/intel/torch-xpu-ops/pull/5486"><span style="font-wei…+
<h3><span style="font-weight: 400;">C++ ABI</span></h3>+
<h4><span style="font-weight: 400;">Expanded Stable ABI and Header-Only Surface</span></h4>+
<p><code class="language-python">torch::stable::sort</code> joins the stable ABI, the Python interop shims gain dtype/device conversions and an <code class="language-python">is_tensor</code> probe alongside the existing tensor conversions. More of <code class="language-python">c10/util</code> (irang…+
<p><span style="font-weight: 400;">(PRs </span><a href="https://github.com/pytorch/pytorch/pull/198038"><span style="font-weight: 400;">#198038</span></a><span style="font-weight: 400;"> by Adrian Abeyta, Red Hat; </span><a href="https://github.com/pytorch/pytorch/pull/194673"><span style="font-weig…+
<h2><span style="font-weight: 400;">Profiling and Debugging</span></h2>+
<h3><span style="font-weight: 400;">CUDA Graph Kernel Attribution with Cuspy</span></h3>+
<p>Kernels replayed from a CUDA graph can now be attributed back to the code that captured them. Kernel annotations are recorded from CUPTI node-creation callbacks and written inline by <code class="language-python">export_chrome_trace</code>, and with <code class="language-python">record_py_stacks<…+
<p><span style="font-weight: 400;">API Unstable</span></p>+
<p><span style="font-weight: 400;">(PRs </span><a href="https://github.com/pytorch/pytorch/pull/191999"><span style="font-weight: 400;">#191999</span></a><span style="font-weight: 400;">, </span><a href="https://github.com/pytorch/pytorch/pull/194854"><span style="font-weight: 400;">#194854</span></…+
<h2><span style="font-weight: 400;">Deprecations and Backwards-Incompatible Changes</span></h2>+
<ul>+
<li style="font-weight: 400;" aria-level="1"><b>Python 3.10 is no longer supported.</b><span style="font-weight: 400;"> PyTorch 2.15 requires Python 3.11 or newer: 3.10 wheels are no longer built, and 3.10 is no longer tested in CI. Building from source on 3.10 is not supported, and upcoming release…+
<li style="font-weight: 400;" aria-level="1"><b>nccl</b><b> now selects the rewritten NCCL backend.</b><span style="font-weight: 400;"> Set </span><span style="font-weight: 400;"><code class="language-python">TORCH_DIST_USE_NCCL2=0</code></span><span style="font-weight: 400;"> or use the </span><spa…+
<li style="font-weight: 400;" aria-level="1"><b>C++ warnings from </b><b>torch.ops</b><b> and custom ops now propagate to Python.</b><span style="font-weight: 400;"> See </span><a href="https://github.com/pytorch/pytorch/pull/193451"><span style="font-weight: 400;">#193451</span></a><span style="fon…+
<li style="font-weight: 400;" aria-level="1"><b>macOS 14 is no longer supported.</b><span style="font-weight: 400;"> PyTorch 2.15 requires macOS 15 or newer for MPS (Apple Silicon GPU) acceleration. macOS 14 has reached end of life, and the MPS workarounds for it have been removed. PyTorch may still…+
<li style="font-weight: 400;" aria-level="1"><b>FSDP2 separates sharded and unsharded gradient dtypes.</b><span style="font-weight: 400;"> See </span><a href="https://github.com/pytorch/pytorch/pull/194434"><span style="font-weight: 400;">#194434</span></a><span style="font-weight: 400;">.</span></l…+
</ul>+
<p>The post <a href="https://pytorch.org/blog/pytorch-2-15-release-blog/">PyTorch 2.15 Release Blog</a> appeared first on <a href="https://pytorch.org">PyTorch</a>.</p>+
]]></content:encoded>+
+
+
+
</item>+
<item> <title>Optimizing Streaming Speech Recognition on Arm® Edge Devices with ExecuTorch</title> <link>https://pytorch.org/blog/optimizing-streaming-speech-recognition-on-arm-edge-devices-with-executorch/</link> @
@@ -327,7 +440,7 @@ in the Agentic Era: Bringing Native PyTorch to Cloud TPUs: Bill Jia (Google Clou <description><![CDATA[<p>The post <a href="https://pytorch.org/blog/meet-the-2026-pytorch-foundation-ambassadors/">Meet the 2026 PyTorch Foundation Ambassadors</a> appeared first on <a href="https://pytorch.org">PyTorch</a>.</p>]]></description> <content:encoded><![CDATA[-
<div id="fws_6ae01edd7bdef" data-column-margin="default" data-midnight="dark" class="wpb_row vc_row-fluid vc_row top-level" style="padding-top: 0px; padding-bottom: 0px; "><div class="row-bg-wrap" data-bg-animation="none" data-bg-animation-delay="" data-bg-overlay="false"><div class="inner-wrap…+
<div id="fws_6ae23cadc32df" data-column-margin="default" data-midnight="dark" class="wpb_row vc_row-fluid vc_row top-level" style="padding-top: 0px; padding-bottom: 0px; "><div class="row-bg-wrap" data-bg-animation="none" data-bg-animation-delay="" data-bg-overlay="false"><div class="inner-wrap… <div class="vc_col-sm-12 wpb_column column_container vc_column_container col no-extra-padding" data-padding-pos="all" data-has-bg-color="false" data-bg-color="" data-bg-opacity="1" data-animation="" data-delay="0" > <div class="vc_column-inner" > <div class="wpb_wrapper">@
@@ -2309,41 +2422,8 @@ cmake --build build -
</item>-
<item>-
<title>Nscale Joins the PyTorch Foundation as a Platinum Member</title>-
<link>https://pytorch.org/blog/nscale-joins-the-pytorch-foundation-as-a-platinum-member/</link>-
-
<dc:creator><![CDATA[PyTorch Foundation]]></dc:creator>-
<pubDate>Tue, 20 Oct 2026 16:00:06 +0000</pubDate>-
<category><![CDATA[Announcements]]></category>-
<category><![CDATA[Blog]]></category>-
<guid isPermaLink="false">https://pytorch.org/?p=189141</guid>-
-
<description><![CDATA[<p>London, U. K., 20 October 2026 – The PyTorch Foundation, the vendor-neutral home for the open source intelligence layer, is announcing today that Nscale has joined as a Platinum member....</p>-
<p>The post <a href="https://pytorch.org/blog/nscale-joins-the-pytorch-foundation-as-a-platinum-member/">Nscale Joins the PyTorch Foundation as a Platinum Member</a> appeared first on <a href="https://pytorch.org">PyTorch</a>.</p>-
]]></description>-
<content:encoded><![CDATA[<p><b><img decoding="async" class="alignnone size-large wp-image-189142" src="https://pytorch.org/wp-content/uploads/2026/10/PyTorch_Foundation_Nscale-1024x576.png" alt="PyTorch Foundation Welcomes Nscale" width="1024" height="576" srcset="https://pytorch.org/wp-c…-
<p><b>London, U. K., 20 October 2026 – </b><span style="font-weight: 400;">The PyTorch Foundation, the vendor-neutral home for the open source intelligence layer, is announcing today that Nscale has joined as a Platinum member. Nscale is a full-stack AI cloud platform, bringing together power, data …-
<p><span style="font-weight: 400;">Nscale first came to the PyTorch Foundation’s project ecosystem through Ray, the open-source framework for distributed computing and scaling AI workloads. Following Nscale’s planned acquisition of Anyscale, the company behind Ray, Nscale has committed t…-
<p><span style="font-weight: 400;">“</span><span style="font-weight: 400;">Most infrastructure providers treat the framework and the data center as two separate problems. We treat them as one. PyTorch has remained the base layer of AI because it kept the Pythonic style researchers know, while gettin…-
<p><span style="font-weight: 400;">“We are excited to welcome Nscale to the PyTorch Foundation as a Platinum Member,” said Mark Collier, Executive Director of the PyTorch Foundation. “As frontier AI workloads scale to multi-trillion parameters, seamless integration between PyTorch, distributed frame…-
<p><span style="font-weight: 400;">As a platinum member, Nscale is granted one seat to the PyTorch Foundation Governing Board. The Board sets policy through our bylaws, mission and vision statements, describing the overarching scope of foundation initiatives, technical vision, and direction.</span><…-
<p><span style="font-weight: 400;">We’re happy to welcome Nick Jones to our board. Nick is VP of Cloud Native Platform Engineering at Nscale, where he leads engineering strategy for its high-performance GPU cloud. He brings more than 15 years’ experience across cloud-native infrastructure, AI …-
<p><span style="font-weight: 400;">We’re also pleased to welcome Eyal Lantzman, VP of AI Services at Nscale to the PyTorch Foundation’s Technical Advisory Council (TAC). At Nscale, Eyal leads the AI services stack, focused on serving and optimizing large open-weight models in production.…-
<p><span style="font-weight: 400;">To learn more about how your organization can join the PyTorch Foundation, visit our </span><a href="https://pytorch.org/join/"><span style="font-weight: 400;">website</span></a><span style="font-weight: 400;">.</span></p>-
<p><b>About PyTorch Foundation</b></p>-
<p><span style="font-weight: 400;">The PyTorch Foundation is the vendor-neutral home for the open source intelligence layer developers use for training, optimizing, serving, and running models on any chip in any cloud for any agent. As a community-driven hub hosted by the Linux Foundation, the PyTor…-
<p><b>About Nscale</b></p>-
<p><span style="font-weight: 400;">Nscale Limited (Nscale) is a full-stack AI cloud platform. We bring together software, compute, and power in a vertically integrated offering – from a unified cloud platform for running AI training and inference, to the data centers and low-cost power that make it …-
<p><b>Press contact: </b><a href="mailto:[email redacted]"><span style="font-weight: 400;">[email redacted]</span></a></p>-
<p>The post <a href="https://pytorch.org/blog/nscale-joins-the-pytorch-foundation-as-a-platinum-member/">Nscale Joins the PyTorch Foundation as a Platinum Member</a> appeared first on <a href="https://pytorch.org">PyTorch</a>.</p>-
]]></content:encoded>-
-
-
</item> </channel></rss>-
<!-- plugin=object-cache-pro client=phpredis metric#hits=4832 metric#misses=31 metric#hit-ratio=99.4 metric#bytes=1583428 metric#prefetches=277 metric#store-reads=35 metric#store-writes=13 metric#store-hits=285 metric#store-misses=15 metric#sql-queries=25 metric#ms-total=508.07 metric#ms-cache=21.34…+
<!-- plugin=object-cache-pro client=phpredis metric#hits=4828 metric#misses=31 metric#hit-ratio=99.4 metric#bytes=1617102 metric#prefetches=276 metric#store-reads=35 metric#store-writes=7 metric#store-hits=284 metric#store-misses=15 metric#sql-queries=19 metric#ms-total=673.72 metric#ms-cache=31.11 …60 lines shown here cut at 300 characters. 2 email addresss masked in this display. See the About page on personal data. The raw artifact at this commit is linked above.