# torch Published articles for torch. This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles. ## High-Throughput Structure Prediction with BioNeMo Inference Runtime DevFeed: [High-Throughput Structure Prediction with BioNeMo Inference Runtime]() Original publisher: [Read original article]() Author: Elizabeth Goodman Published: 2026-09-10T15:00:00Z Content type: tutorial Language: en Sources: [NVIDIA Developer](), [NVIDIA Technical Blog]() Topics: [AI, ML & Data Engineering]() Tags: [bionemo](), [cuda](), [cuda-graphs](), [data-center-cloud](), [data-science](), [drug-discovery](), [gpu](), [healthcare-life-sciences](), [how-to](), [hpc-scientific-computing](), [inference](), [integration](), [kernels](), [model](), [models](), [node](), [nvidia](), [python](), [pytorch](), [resource](), [scale](), [simulation-modeling-design](), [tokenization](), [torch](), [tutorial](), [workflow]() ### AI overview A tutorial on using NVIDIA BioNeMo Inference Runtime to accelerate biomolecular structure-prediction models on GPUs. It covers the end-to-end Boltz2 workflow, PyTorch integration, input requirements, and Ray-based single-node throughput scaling. ### Source excerpt Biomolecular structure prediction is now often run at proteome scale, where the goal is to move an entire worklist through the pipeline efficiently. NVIDIA... ## Building a Transformer-Based Category Recommender at Thumbtack DevFeed: [Building a Transformer-Based Category Recommender at Thumbtack]() Original publisher: [Read original article]() Author: Andrew Morss Published: 2026-07-20T20:23:27Z Content type: article Language: en Sources: [Thumbtack Engineering - Medium]() Topics: [recommendation systems](), [Transformer](), [Machine learning](), [email](), [push-notification]() Tags: [catalog](), [data-science](), [email](), [engineering](), [machine-learning](), [model](), [push-notification](), [recommendations](), [recommender-systems](), [torch]() ### AI overview This engineering article describes Thumbtack's category recommender, which selects home-service categories for push and email notifications. It explains how the system addresses popularity bias through negative sampling and how Thumbtack's small category catalog and sparse user activity shape the recommendation problem. ### Source excerpt A look at compensating for popularity bias in recommender systems using negative sampling strategies By: Andrew Morss, Senior Applied Scientist Introduction A recommender system is a machine learning model that, given a user and a catalog of items, predicts which items that user is most likely to want. Recommenders set your YouTube playlist, determine what items Amazon suggests for you, push you songs on Spotify and customize your Steam store. If you're a homeowner, Thumbtack's recommender systems can suggest home projects for you such as house cleaning or lawn mowing. Thumbtack connects users with professionals spread across almost 500 different job categories. Job categories are a kind of occupational bucket that Thumbtack uses as a taxonomy to categorize local service professionals (pros). Pros have the opportunity to choose categories of jobs that they want to receive leads on; examples include things like "Electrical and Wiring Repair", "Local Moving" or "Snow Plowing". Thumbtack's Lifecycle marketing team uses a category recommender to choose which categories to surface to users in our push and email notifications. Choosing the right ones is critical. A well-chosen suggestion is how a customer starts a search for a new project, finds the right service professional, and ultimately gets work done. An example of a push notification and an in-email category recommendation. The Problem Space: Category Recommendation There are several features of category recommendations at Thumbtack that are fairly unique in the item recommender space. The number of categories that we have is relatively small (~500) compared to the number of items that are usually seen in recommender systems, which simplifies the problem considerably. Usually recommenders require a two stage approach where an initial, low compute, filtering stage reduces the number of items to some "candidate" list of items before a more complicated model ranks that smaller set of items. With a relatively small cat ## Profiling in PyTorch (Part 3): Attention is all you profile DevFeed: [Profiling in PyTorch (Part 3): Attention is all you profile]() Original publisher: [Read original article]() Author: Aritra Roy Gosthipaty; Sergio Paniego; Sayak Paul; Rémi Ouazan Reboul Published: 2026-07-10T00:00:00Z Content type: tutorial Language: en Sources: [Hugging Face - Blog]() Topics: [AI, ML & Data Engineering](), [jobs](), [cpu]() Tags: [artificial-intelligence](), [cudnn](), [flash](), [gpu](), [jobs](), [kernel](), [profile](), [pytorch](), [torch](), [traces](), [xformers]() ### AI overview A PyTorch profiling tutorial examines naive attention, identifying its primitive operations and the CPU and GPU kernels they launch. ### Source excerpt We're on a journey to advance and democratize artificial intelligence through open source and open science. ## Profiling in PyTorch (Part 2): From nn.Linear to a Fused MLP DevFeed: [Profiling in PyTorch (Part 2): From nn.Linear to a Fused MLP]() Original publisher: [Read original article]() Author: Aritra Roy Gosthipaty; Rémi Ouazan Reboul; Sergio Paniego; Pedro Cuenca; Sayak Paul Published: 2026-06-11T00:00:00Z Content type: tutorial Language: en Sources: [Hugging Face - Blog]() Topics: [AI, ML & Data Engineering](), [cpu](), [jobs]() Tags: [cpu](), [cuda](), [deep-learning](), [gpu](), [jobs](), [kernels](), [liger](), [mlp](), [open-source](), [profile](), [pytorch](), [torch](), [traces]() ### AI overview A PyTorch profiling tutorial that moves from nn.Linear to a fused multilayer perceptron, explaining GPU kernels, CPU launch overhead, and tensor-transpose behavior in profiler traces. ### Source excerpt We're on a journey to advance and democratize artificial intelligence through open source and open science. ## Profiling in PyTorch (Part 1): A Beginner's Guide to torch.profiler DevFeed: [Profiling in PyTorch (Part 1): A Beginner's Guide to torch.profiler]() Original publisher: [Read original article]() Author: Aritra Roy Gosthipaty; Sayak Paul; Sergio Paniego; Rémi Ouazan Reboul; Pedro Cuenca Published: 2026-05-29T00:00:00Z Content type: tutorial Language: en Sources: [Hugging Face - Blog]() Topics: [GPU](), [AI, ML & Data Engineering](), [cpu](), [jobs]() Tags: [cpu](), [cuda](), [gpu](), [guide](), [how-to](), [jobs](), [kernel](), [kernels](), [llm](), [open-source](), [profile](), [python](), [pytorch](), [torch](), [traces](), [training](), [tutorials]() ### AI overview A beginner-focused tutorial on using torch.profiler to inspect PyTorch execution, interpret CPU and GPU traces, and connect Python operations to CUDA kernels. ### Source excerpt We're on a journey to advance and democratize artificial intelligence through open source and open science. ## You could have designed state of the art positional encoding DevFeed: [You could have designed state of the art positional encoding]() Original publisher: [Read original article]() Author: Christopher Fleetwood Published: 2024-11-25T00:00:00Z Content type: tutorial Language: en Sources: [Hugging Face - Blog]() Topics: [AI, ML & Data Engineering]() Tags: [artificial-intelligence](), [embeddings](), [llama](), [multimodal](), [research](), [torch](), [transformers](), [tutorial]() ### AI overview A tutorial on deriving positional encodings for transformer self-attention, progressing toward Rotary Positional Encoding (RoPE). ### Source excerpt We're on a journey to advance and democratize artificial intelligence through open source and open science. ## From DeepSpeed to FSDP and Back Again with Hugging Face Accelerate DevFeed: [From DeepSpeed to FSDP and Back Again with Hugging Face Accelerate]() Original publisher: [Read original article]() Author: Yu Chin Fabian Lim; aldo pareja; Zachary Mueller; Stas Bekman Published: 2024-06-13T00:00:00Z Content type: article Language: en Sources: [Hugging Face - Blog]() Topics: [AI, ML & Data Engineering]() Tags: [accelerate](), [collaboration](), [frameworks](), [gpu](), [guide](), [hugging-face](), [memory](), [model](), [open-source](), [pytorch](), [research](), [torch](), [training]() ### AI overview The article compares DeepSpeed and PyTorch FSDP mixed-precision training behavior. It attributes initially different convergence results to DeepSpeed's internal upcasting of master weights to full precision, then discusses the memory and optimizer implications of this design versus FSDP. ### Source excerpt We're on a journey to advance and democratize artificial intelligence through open source and open science. ## Happy New Year, the keyboard and cases are here! DevFeed: [Happy New Year, the keyboard and cases are here!]() Original publisher: [Read original article]() Published: 2021-12-31T00:00:00Z Content type: release Language: en Sources: [Community blog on PINE64]() Topics: [Pinephone-Pro](), [Pinephone](), [keyboard]() Tags: [battery](), [camera](), [firmware](), [keyboard](), [peripheral](), [pinephone](), [pinephone-pro](), [screen](), [speaker](), [torch](), [usb](), [wireless]() ### AI overview Pine64 announces that the PinePhone and PinePhone Pro keyboard case, along with add-on cases, are available in the Pine Store. The keyboard case features a 180-degree hinge, rearrangeable keycaps, open firmware, a 6000mAh battery, and USB-C peripheral support. ### Source excerpt The PinePhone & PinePhone Pro keyboard case and add-on cases are now available in the Pine Store! Go to store About the keyboard case The keyboard case works with both the PinePhone and PinePhone Pro and features a clam-shell design. It uses pogo pins located on the phone's midsection and attaches by replacing the default back cover. When folded, the phone's screen and the keyboard rest securely against each other. The hinge features a 180° design, which not only allows for two-hand typing on a surface but also for comfortable thumb-typing when fully extended. The etched keycaps can be easily relocated for alternate layouts such as AZERTY or QWERTZ. The keyboard case runs an open firmware, which means that anyone with the know-how can alter existing functions or add new ones. The bottom (keyboard) and top (phone) sections of the assembly are well-balanced thanks to the large, 6000mAh, internal battery capable of charging the PinePhone (Pro) during operation. The internal battery effectively triples the phone's battery life. The internal keyboard battery can be manually toggled on/off and the keyboard's battery charge level can be read in the supported OSes; the keyboard remains functional with the battery fully depleted. You do not lose access to the PinePhone (Pro)'s USB-C port, speaker, microphone, or any external features, such as volume and lock buttons, with the keyboard attached. There is also a cut-out for the camera, torch, and headphone jack. The USB-C port on the keyboard is capable of powering both the keyboard and PinePhone (Pro) simultaneously. This means that you can plug in a USB mouse, a USB-C dongle, or some other peripheral while the phone and keyboard's internal battery charge. Please keep in mind that the keyboard case transforms the PinePhone (Pro) into a PDA, which means that taking calls will likely prove awkward without a wired or wireless headset connected. More information To learn more about the keyboard and its journey please read: Februa ## Teaching an LSTM to Play Beethoven DevFeed: [Teaching an LSTM to Play Beethoven]() Original publisher: [Read original article]() Published: 2015-07-01T00:00:00Z Content type: tutorial Language: en Sources: [Alex Korbonits]() Topics: [MIDI](), [Code](), [Lua](), [GPU]() Tags: [audio](), [channel](), [code](), [experiment](), [format](), [gpu](), [lstm](), [lua](), [music](), [structure](), [timing](), [torch](), [training](), [types](), [velocity]() ### AI overview An experiment trains Karpathy's char-rnn LSTM on Beethoven piano sonata MIDI files, treating the raw MIDI bytes as a sequence. The author describes prepending a shared MIDI header to generated output so players could interpret it, and reports imperfect results containing occasional phrase-like and melodic fragments. ### Source excerpt In June 2015, I trained Karpathy's char-rnn on a corpus of Beethoven piano sonata MIDI files. Here's what it sounds like -- and how I got it to play. ## Torch for Deep Learning Research DevFeed: [Torch for Deep Learning Research]() Original publisher: [Read original article]() Published: 2015-06-28T11:49:40Z Content type: tutorial Language: en Sources: [Alex Korbonits]() Topics: [torch](), [Deep learning](), [Artificial Intelligence]() Tags: [bash](), [deep-learning](), [lua](), [research](), [torch]() ### AI overview This article introduces Torch as a Lua-based deep learning framework and discusses its use for research, including its ecosystem, flexibility, installation, and relationship to Python. ### Source excerpt An introduction to Torch, the deep learning framework favored by Facebook AI and Google DeepMind for cutting-edge research.