PyTorch

Track PyTorch development. The machine learning framework powering AI research.

About this developer podcast

PyTorch turns GitHub activity into short audio updates.

Episodes cover commits, pull requests, issues, and release work.

Use this page to listen, subscribe, and find related developer podcasts.

This show page is the public hub for the podcast. It has one stable listen URL.

Use it to find RSS feeds, recent episodes, related shows, and older episode links.

If the project is quiet, new episodes may pause until there is useful activity to summarize.

Recent episodes from PyTorch

  1. Hardening Memory and Lifetime Management Across GPU Backends

    Today's activity centers on fixing resource lifetime bugs across NCCL, CUDA graphs, and the profiler, alongside a dynamo performance push to cut per-call overhead on hot paths. A cluster of typo-fix and cleanup PRs also landed,…

  2. Correctness Fixes and Reverts Take Center Stage

    A cluster of correctness fixes landed across CUDA graphs, group norm, embedding bag, and RNG code, while two contentious changes—the XNNPACK update and a ROCm CI image switch—got reverted again after breaking things. Meanwhile, a…

  3. Distributed Backends and Overflow Fixes Take Center Stage

    A major push to modernize the c10d NCCL2 backend and consolidate test infrastructure ran alongside a cluster of integer-overflow and safety fixes across CUDA kernels and type conversion utilities. Several CI reliability fixes and reverts…

  4. Packaging Fixes and Bounds-Check Sweep

    A cluster of ROCm and manywheel packaging fixes targeted broken binary distribution, while a separate wave of MPS indexing patches closed out-of-bounds memory bugs across several tensor ops. Together they point to a release cycle focused…

  5. Autograd Overhead Cuts and a New GEMM Backend Migration

    The autograd hot path and the NVGEMM inductor backend both got sustained performance surgery this cycle, while a new distributed communication backend for Apple Silicon and a fresh NCCL rewrite pushed further into review. Developers…

  6. Hardening Boundary Checks and MPS Correctness

    Today's activity centers on shape and overflow validation fixes across convolution and pooling kernels, a wave of Apple GPU correctness and performance work, and a continued push to make distributed tests hardware-agnostic.

  7. Weekly Recap - Metal Kernels, Compile Reliability, and Dynamo's Long Tail

    This week's work centered on rebuilding Apple GPU operator kernels in Metal, hardening the Inductor compile-worker pipeline against silent hangs, and clearing a long backlog of Dynamo graph-break and precompile fixes. Fifty pull requests…

  8. Apple Silicon Gets a Metal Overhaul

    A cluster of MPS backend fixes and migrations moved several operators from MPSGraph to Metal kernels, closing correctness bugs in clamp, RMSNorm, attention, and linspace along the way, while a separate stack landed the first user-facing…

  9. Dynamo's Great Type-System Cleanup
  10. Dynamo's CPython Test Push and a Rocky Exception-Handling Rewrite
  11. Distributed Backend Overhaul and a Wave of Cleanup
  12. The Great Runner Migration, and a Distributed Rollback
  13. Distributed Cleanup and CI Infrastructure Overhaul
  14. Correctness Fixes Sweep the Compiler Stack
  15. Weekly Recap - Release Hardening and Correctness Cleanup
  16. Determinism and Correctness Sweep
  17. Silent Correctness Bugs Take Center Stage
  18. FlexAttention's Rough Week
  19. Accelerator Backends and Memory Management
  20. Weekly Recap - Release Stabilization & Core Improvements
  21. GEMM Optimization and Core Stability Fixes
  22. C++20 Migration and Infrastructure Improvements
  23. Release Infrastructure and Compiler Stability Fixes
  24. Header Reorganization and Dynamo Improvements
  25. Infrastructure Organization and Core API Cleanup
  26. Dynamo Correctness and Build Infrastructure
  27. Performance Optimizations and Backend Improvements
  28. Weekly Recap - Compiler Optimization & Reliability
  29. Dynamo Compilation Stability and Performance
  30. Dynamo Stability and Inductor Optimizations
  31. Compiler Performance and Distributed Computing Infrastructure
  32. Distributed Naming Overhaul and Compiler Fixes
  33. Dynamo Stability and Compilation Robustness
  34. Compiler Numerical Accuracy and Index Expression Fixes
  35. Compiler Error Handling and Stability Fixes
  36. Weekly Recap - Stability and Error Handling Improvements
  37. Weekly Recap - Code Quality & Infrastructure Improvements
  38. Weekly Recap - Dynamo Enhancements & Critical Fixes
  39. Optimization Engine Revamp and Hash System Breakthrough
  40. Operator Improvements and Build System Updates
  41. Weekly Recap - Reverts and Infrastructure Updates
  42. Inductor Improvements and Bug Fixes
  43. Optimization Improvements and Cache Unification
  44. Weekly Recap - Performance Optimizations and Code Cleanup
  45. Weekly Recap - Infrastructure Stability & Mixed Precision
  46. Dynamo Improvements and Distributed Computing Fixes
  47. Weekly Recap - Performance Optimization and Infrastructure
  48. The Debugging Detective Story
  49. The Great Rollback and Recovery
  50. Exception Handling Revolution & Stateless RNG Arrives
  51. Cross-Platform Expansion and Developer Experience Wins
  52. Building Stronger Foundations
  53. Spring Cleaning & Major Infrastructure Upgrades
  54. AOT AutoGrad Fixes and Cross-Platform Polish
  55. Building Bridges - Distributed Computing Gets a Major Upgrade
  56. Profiling Power-Ups and Infrastructure Smoothing
  57. Fixes, Reverts, and Moving Forward
  58. The Infrastructure Acceleration Edition
  59. Lanczos Interpolation Breakthrough
  60. Stream Management Mastery & RNG Fixes
  61. Matrix Math Gets a Speed Boost
  62. Under the Hood Improvements and Future-Proofing
  63. Complex Math Gets Smarter & Build Improvements
  64. Memory Optimization Revolution
  65. Testing Gets Smarter and Graphs Go Universal
  66. Polish & Performance Day
  67. Distributed Computing Gets Real - Compilation, Clustering, and Convolutions
  68. Performance Revolution and Developer Experience Upgrades
  69. Windows Testing Gets Flexible & Dynamic Shapes Take Flight
  70. Metal Shaders Get a Precision Fix
  71. The Testing & Error Handling Polish Episode
  72. Stream Safety and Performance Wins
  73. Subclass Evolution and Memory Management Improvements
  74. Performance Tuning and Code Health Day
  75. Variable-Length Attention Gets Supercharged
  76. Spring Cleaning and Performance Boosts
  77. Stream Wizardry and Symbolic Shapes Magic
  78. CI Optimizations and Cross-Platform Fixes
  79. Spring Cleaning and Precision Fixes
  80. Memory Safety Fixes and Development Velocity
  81. Speed Wins and Better Error Messages
  82. Distributed Computing Gets Smarter
  83. Distributed Computing Gets Smarter & Vision Models Get Lightning Fast
  84. Release Dance and Rapid Recovery
  85. Performance Wins and Stability Fixes
  86. Distributed Computing Gets Smarter
  87. Valentine's Day Cleanup and Distributed Computing Love
  88. The Day of Rollbacks and Second Chances
  89. TPU Integration and the Dance of Reverts
  90. The Performance Optimization Sprint
  91. The Great Performance Revolution - Tests Run 70% Faster!
  92. Bug Fixes and Performance Wins
  93. The Great Test Speed Revolution
  94. Cleanup and Optimization Day
  95. Testing Cleanup and Pattern Matching Progress
  96. Type Safety Revolution and Infrastructure Cleanup
  97. Backend Flexibility Revolution
  98. The Great Configuration Cleanup & XPU Expansion
  99. Hardware Expansion and Developer Experience Polish
  100. Backend Harmony and Memory Magic

Episode archive for PyTorch

  1. Spring Cleaning and Building Blocks
  2. Bytecode Magic and Buffer Management Mastery
  3. Kernel Optimization and Clean Code Victory
  4. FMA Optimization Focus and Debugging Improvements
  5. Developer Tooling Revolution
  6. Deep Dive into PyTorch's Core - Opaque Objects and Performance Wins

More public developer podcasts

Browse other public Podlog show hubs with crawler-readable RSS and episode links.

  1. Next Js Daily 8 episodes
  2. Headroom Daily 17 episodes
  3. Agent Of Empires Daily 9 episodes
  4. Frigate NVR Updates 148 episodes
  5. Linux Kernel 135 episodes
  6. Homebrew 156 episodes
  7. TypeScript 62 episodes
  8. Ruby on Rails 149 episodes
  9. Redis 125 episodes
  10. Next.js 158 episodes
  11. Vue.js 102 episodes
  12. Rust 128 episodes
  13. Ruby Core Updates 151 episodes
  14. Python 141 episodes
  15. Go 142 episodes
  16. VS Code 159 episodes
  17. TailwindCSS 89 episodes
  18. Node.js 153 episodes
  19. Django 145 episodes
  20. PostgreSQL 146 episodes
  21. Kubernetes 154 episodes
  22. LangChain 139 episodes
  23. Ollama 143 episodes
  24. Maestro Daily 7 episodes
  25. Videorc Daily 11 episodes
  26. Navidrome Daily 22 episodes
  27. Next.js Daily 160 episodes
  28. OpenClaw 117 episodes
  29. RuView 86 episodes
  30. Godot Daily 105 episodes
  31. Agora Next Updates 175 episodes
  32. Rails Daily 159 episodes
  33. Linux Kernel Daily 140 episodes
  34. Home Assistant Daily 164 episodes
  35. BlocksBeyondTheStars Daily 6 episodes
  36. Jabref Daily 19 episodes
  37. Jabref Daily 19 episodes
  38. Buzz Transcription 55 episodes
  39. Shannon 55 episodes
  40. Agora Next Daily 7 episodes
  41. Onlook Design Updates 26 episodes
  42. iCloud Photos Downloader 21 episodes
  43. tiny-gpu Daily 18 episodes
  44. NativeScript iOS Daily 4 episodes
  45. AtomVM Daily 8 episodes
  46. OpenAI Skills 52 episodes
  47. Crush Daily 1 episode
  48. React Native 125 episodes
  49. React Daily 79 episodes
  50. Pi Mono 63 episodes
  51. The Algorithm Daily 13 episodes
  52. ControlNet Daily 0 episodes
  53. FinceptTerminal Daily 0 episodes