45 Pull requests merged by 15 people

• Softmin

  #2358 merged Oct 15, 2024

• Bump wasm-bindgen-futures from 0.4.43 to 0.4.45

  #2369 merged Oct 15, 2024

• Contributor Book: Fix the link of primitive types in the "Serialization" page

  #2362 merged Oct 15, 2024

• Fix typo

  #2363 merged Oct 15, 2024

• Combined PRs

  #2370 merged Oct 15, 2024

• Windows/WindowsIterator/WindowsDataset

  #2338 merged Oct 15, 2024

• Make {Cosine,Exponential,Linear} LR schedulers return the initial LR at the first step

  #2337 merged Oct 15, 2024

• Add bounds checking to implicit GEMM to allow arbitrary input shapes

  #2354 merged Oct 11, 2024

• Bump github/combine-prs from 5.1.0 to 5.2.0

  #2355 merged Oct 11, 2024

• Refactor pooling kernels

  #2356 merged Oct 11, 2024

• Refactor Adaptive Avg Pool to CubeCL

  #2351 merged Oct 9, 2024

• Chore/update cubecl

  #2352 merged Oct 9, 2024

• Update cubecl

  #2350 merged Oct 9, 2024

• enable doc_auto_cfg to show feature-req-hint in docs.rs

  #2271 merged Oct 9, 2024

• Fix autodiff memory leak

  #2347 merged Oct 9, 2024

• Refactor/elemwise fusion

  #2344 merged Oct 9, 2024

• Add onnx op trilu

  #2323 merged Oct 8, 2024

• Combined PRs

  #2343 merged Oct 7, 2024

• Add utility methods to split gradients to GradientParams

  #2311 merged Oct 4, 2024

• Add dim checks on output rank for unsqueeze_dim and stack

  #2331 merged Oct 4, 2024

• Move conv autotune under feature flag (except key)

  #2330 merged Oct 2, 2024

• Update rusqlite and associated libraries

  #2328 merged Oct 2, 2024

• Modify url links to book in crates/burn-import/README.md

  #2327 merged Oct 2, 2024

• Bump flate2 from 1.0.33 to 1.0.34

  #2319 merged Sep 30, 2024

• Combined PRs

  #2322 merged Sep 30, 2024

• Further data locality optimizations for implicit GEMM

  #2300 merged Sep 26, 2024

• Update CubeCL

  #2310 merged Sep 26, 2024

• Add fusion quantization ops

  #2301 merged Sep 26, 2024

• Move LrSchedule generic to make it easier to use

  #2309 merged Sep 26, 2024

• Use CliMetricsRenderer when not in a terminal

  #2307 merged Sep 26, 2024

• Add candle CudaDevice and MetalDevice to avoid creating a new unique device each time

  #2290 merged Sep 25, 2024

• Fix burn book links

  #2303 merged Sep 25, 2024

• Remove const D generic

  #2298 merged Sep 24, 2024

• Introduce autotuning to conv2d and conv_transpose2d with a new im2col/GEMM algorithm

  #2287 merged Sep 23, 2024

• Add deform_conv2d as implemented in torchvision

  #2147 merged Sep 23, 2024

• Combined PRs

  #2297 merged Sep 23, 2024

• doc: improve doc for burn-tch

  #2288 merged Sep 23, 2024

• Chore: Update CubeCL

  #2292 merged Sep 21, 2024

• Add more quantization support for burn-jit

  #2275 merged Sep 17, 2024

• Update rev

  #2283 merged Sep 17, 2024

• Modify unresolved import regression

  #2285 merged Sep 17, 2024

• Refactor: Split conv backward to allow conditional gradient computation

  #2278 merged Sep 16, 2024

• Add ResNet benchmark

  #1534 merged Sep 16, 2024

• Combined PRs

  #2282 merged Sep 16, 2024

• Adding some comments to burn fusion

  #2130 merged Sep 16, 2024

11 Pull requests opened by 10 people

• Add precision classification metric

  #2293 opened Sep 21, 2024

• Infer ONNX conv output shapes

  #2304 opened Sep 25, 2024

• Onnx op topk

  #2305 opened Sep 25, 2024

• Add tester

  #2335 opened Oct 5, 2024

• Add functionality to ensure ONNX node graph is sorted

  #2336 opened Oct 5, 2024

• Use core::error::Error for no-std

  #2346 opened Oct 8, 2024

• Add ConvTranspose1d ONNX op

  #2349 opened Oct 9, 2024

• Add `BackendRouter` to handle multiple backends

  #2353 opened Oct 9, 2024

• Fix xtask args which are unmodified when upgrading xtask commands

  #2364 opened Oct 14, 2024

• Add `round`, `floor`, `ceil` for float tensor

  #2372 opened Oct 14, 2024

• Update cubecl

  #2376 opened Oct 15, 2024

24 Issues closed by 9 people

• num_workers and iteration

  #2373 closed Oct 16, 2024

• Implement softmin

  #2357 closed Oct 15, 2024

• sgd optimizer

  #2359 closed Oct 14, 2024

• multi threads support

  #2345 closed Oct 11, 2024

• Add the ability to access the underlying wgpu::Buffer of a Tensor when using the Wgpu backend.

  #2274 closed Oct 9, 2024

• Heap memory leak from `AutodiffTensor`

  #2333 closed Oct 9, 2024

• Addition of `Resize` operator

  #510 closed Oct 9, 2024

• Memory leaks in `burn-compute`

  #1016 closed Oct 4, 2024

• Crash when calling Tensor::nonzero() on a tensor that all values are zero.

  #2180 closed Oct 4, 2024

• Better memory management in Burn Compute

  #1752 closed Oct 4, 2024

• Error when running with tch-gpu feature on Windows

  #1247 closed Oct 4, 2024

• Error when stacking tensors (images as u8)

  #2329 closed Oct 4, 2024

• Detection issue with heterogenous cards

  #2242 closed Oct 1, 2024

• No adapter found for graphics API AutoGraphicsApi

  #1691 closed Sep 30, 2024

• Wgpu backend

  #2308 closed Sep 27, 2024

• OpenGL backend

  #2306 closed Sep 27, 2024

• MetricsRenderer for non-terminals

  #2302 closed Sep 26, 2024

• run-checks failing

  #2291 closed Sep 26, 2024

• Conv2d/Conv3d doesn't match `libtorch`

  #2286 closed Sep 24, 2024

• How to use audio data for training?

  #2294 closed Sep 23, 2024

• CI Setup: Github Action Templates

  #1061 closed Sep 18, 2024

• Use `AsRef<Path>` instead of `&str` for file operations

  #698 closed Sep 17, 2024

• Simple Regression Example has import errors.

  #2284 closed Sep 17, 2024

• Add Resnet in backend-comparison

  #1521 closed Sep 16, 2024

13 Issues opened by 10 people

• Wrong usage of unsafe `Vec::from_raw_parts` in TensorData

  #2375 opened Oct 15, 2024

• MseLoss forward input

  #2371 opened Oct 14, 2024

• Candle: better pooling op support

  #2374 opened Oct 13, 2024

• Implement SegmentationMask in image_folder.rs (dataset::vision)

  #2361 opened Oct 12, 2024

• Crate futures-util 0.3.30 is yanked

  #2348 opened Oct 8, 2024

• Add AdaptiveAvgPool3d

  #2334 opened Oct 3, 2024

• Error when deserializing Enum with mismatched weight tensor shapes across variants

  #2332 opened Oct 2, 2024

• RUSTSEC-2024-0375: `atty` is unmaintained

  #2326 opened Oct 1, 2024

• Ident cannot be a number; use Literal instead

  #2324 opened Sep 30, 2024

• Add "load pytorch tensor" section into the burn book

  #2316 opened Sep 30, 2024

• PRNG Seeding Guidance

  #2312 opened Sep 27, 2024

• WGPU backend panic on android emulators <= 10

  #2299 opened Sep 23, 2024

• Features enabled by examples unexpectedly which impact features in `crates` directory.

  #2289 opened Sep 19, 2024

13 Unresolved conversations

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• Training metrics

  #544 commented on Sep 16, 2024 • 0 new comments

• FFT Module (like torch.fft)

  #788 commented on Sep 24, 2024 • 0 new comments

• RUSTSEC-2024-0370: proc-macro-error is unmaintained

  #2270 commented on Sep 27, 2024 • 0 new comments

• No adapter found for graphics API AutoGraphicsApi (wgpu backend)

  #792 commented on Sep 28, 2024 • 0 new comments

• Add common LR schedulers

  #1198 commented on Oct 1, 2024 • 0 new comments

• Support for 0-Dimensional Tensors in Burn

  #1689 commented on Oct 8, 2024 • 0 new comments

• Load `yolov8` network into `burn-rs`

  #675 commented on Oct 9, 2024 • 0 new comments

• Help Wanted: Implementing ONNX Ops

  #1714 commented on Oct 9, 2024 • 0 new comments

• Load `edge-connect` Onnx files into `burn-rs`

  #673 commented on Oct 9, 2024 • 0 new comments

• Sparse tensors

  #1998 commented on Oct 2, 2024 • 0 new comments

• Add cumsum

  #2099 commented on Sep 23, 2024 • 0 new comments

• Adding support for the ONNX Scatter operation

  #2197 commented on Sep 18, 2024 • 0 new comments

• Feat: Add android example of MNIST inference

  #2221 commented on Oct 5, 2024 • 0 new comments