Ecosyste.ms: Issues

An open API service for providing issue and pull request metadata for open source projects.

GitHub / coreylowman/dfdx issues and pull requests

#828 - Moving optim kernels to tensor ops

Pull Request - State: closed - Opened by coreylowman about 1 year ago

#827 - Why do gradient tapes need to start at the inputs?

Issue - State: closed - Opened by emchristiansen about 1 year ago - 2 comments

#826 - Is it possible to train models with multiple inputs?

Issue - State: closed - Opened by emchristiansen about 1 year ago - 7 comments

#824 - Fixing examples/04-gradients.rs

Pull Request - State: closed - Opened by coreylowman about 1 year ago

#822 - impl Default for Tensor?

Issue - State: open - Opened by emchristiansen about 1 year ago - 3 comments

#821 - Add support for complex tensors

Issue - State: open - Opened by vikigenius about 1 year ago - 1 comment

#820 - `dfdx/examples/04-gradients.rs` crashes

Issue - State: closed - Opened by emchristiansen about 1 year ago

#819 - Is there a function like linspace ?

Issue - State: closed - Opened by callesp about 1 year ago - 1 comment

#818 - Do you have a changelog page somewhere?

Issue - State: closed - Opened by ninjasaid2k about 1 year ago

#817 - Make `storage_traits::TensorToArray` pub

Pull Request - State: closed - Opened by AndrejOrsula about 1 year ago

#816 - Making all shape traits public

Pull Request - State: closed - Opened by coreylowman about 1 year ago

#815 - [WIP] Add Serialization for Optimizers and Gradients

Pull Request - State: open - Opened by nkoppel about 1 year ago

#814 - Various small fixes

Pull Request - State: closed - Opened by nkoppel about 1 year ago

#813 - accurate-gelu

Pull Request - State: closed - Opened by jcrist1 about 1 year ago - 9 comments

#812 - Add rmsprop/sgd/adam as tensor ops

Issue - State: closed - Opened by coreylowman about 1 year ago

#811 - [Breaking] Adds `AMP<F>` dtype

Pull Request - State: closed - Opened by coreylowman about 1 year ago - 4 comments

#810 - Re-export `half::f16` in lib.rs if f16 feature is enabled

Issue - State: closed - Opened by coreylowman about 1 year ago
Labels: good first issue

#809 - Make Gradients::get_or_alloc_mut pub

Issue - State: closed - Opened by coreylowman about 1 year ago
Labels: good first issue

#808 - Allow models to be backward compatible through #799

Pull Request - State: closed - Opened by nkoppel about 1 year ago

#807 - conv1d - added 1d convolution

Pull Request - State: closed - Opened by jcrist1 about 1 year ago - 10 comments

#806 - Using Groups in conv weight init

Pull Request - State: closed - Opened by coreylowman about 1 year ago

#805 - Bump half/cudarc versions

Pull Request - State: closed - Opened by coreylowman about 1 year ago

#804 - Support accurate GeLU as well as fast approximate GeLu

Issue - State: closed - Opened by jcrist1 about 1 year ago - 3 comments

#803 - [WIP] Adds `#[derive(Sequential)]` on nn builder structs

Pull Request - State: closed - Opened by coreylowman about 1 year ago - 1 comment

#801 - How about integrate CUDA kernal lauching with Rust async?

Issue - State: closed - Opened by npuichigo over 1 year ago - 1 comment

#800 - Discord link is invalid

Issue - State: closed - Opened by npuichigo over 1 year ago - 2 comments

#799 - Add scalar support to TensorCollection

Pull Request - State: closed - Opened by nkoppel over 1 year ago

#798 - Adding basic RNN implementation

Pull Request - State: closed - Opened by xubaiw over 1 year ago - 2 comments

#797 - Fixing weight shape for grouped Conv2D

Pull Request - State: closed - Opened by coreylowman over 1 year ago

#796 - Make more functions/modules/traits public

Issue - State: closed - Opened by M1ngXU over 1 year ago - 3 comments

#795 - is this similar to rust for diffusers?

Issue - State: closed - Opened by ninjasaid2k over 1 year ago - 1 comment

#794 - Mobilenet integration

Pull Request - State: open - Opened by Awpteamoose over 1 year ago - 1 comment

#793 - Mobilenet ops

Pull Request - State: open - Opened by Awpteamoose over 1 year ago

#792 - Conv2D IN_CHAN is wrong if GROUPS > 1

Issue - State: closed - Opened by Awpteamoose over 1 year ago - 1 comment

#791 - Exposed NumpyDType publicly

Pull Request - State: closed - Opened by jafioti over 1 year ago

#790 - Uses grid striding in fill_with

Pull Request - State: closed - Opened by coreylowman over 1 year ago

#789 - Using gridDim.x in reduction (max/min/sum) cuda kernels

Issue - State: open - Opened by coreylowman over 1 year ago
Labels: gpu, optimization, expert

#788 - Removing __hmax and __hmin compat functions

Pull Request - State: closed - Opened by coreylowman over 1 year ago

#787 - Using for loop with gridDim.x * blockDim.x as increment

Pull Request - State: closed - Opened by coreylowman over 1 year ago

#786 - Tensor `Display` Implementation?

Issue - State: open - Opened by zicklag over 1 year ago - 4 comments

#785 - Impl SaveToNpz for Optimizers

Issue - State: open - Opened by jafioti over 1 year ago - 2 comments

#784 - Adding better error messages if nvidia-smi/nvcc are not found

Pull Request - State: closed - Opened by coreylowman over 1 year ago

#783 - [Breaking] Adding dilation/groups to ConvTranspose2D

Pull Request - State: closed - Opened by coreylowman over 1 year ago

#781 - Incorrect PATH in Linux causes build to implode

Issue - State: closed - Opened by sdake over 1 year ago - 3 comments

#780 - Conv transpose upgrades

Issue - State: closed - Opened by coreylowman over 1 year ago

#779 - Moving some tests off nightly. Adding docs to conv2d op

Pull Request - State: closed - Opened by coreylowman over 1 year ago

#777 - Properly setting read_dst for gemm in forward/backward pass

Pull Request - State: closed - Opened by coreylowman over 1 year ago

#775 - impl Clone for Adam, SGD, RMSprop

Pull Request - State: closed - Opened by coreylowman over 1 year ago

#774 - Moving `cudarc?/f16` behind `f16` feature

Pull Request - State: closed - Opened by coreylowman over 1 year ago

#773 - Add integration test for mobile net

Issue - State: open - Opened by coreylowman over 1 year ago - 1 comment
Labels: test

#772 - Conv2d general

Pull Request - State: closed - Opened by M1ngXU over 1 year ago - 4 comments

#770 - Conv2D non-squared kernels/strides/padding

Issue - State: closed - Opened by M1ngXU over 1 year ago - 6 comments
Labels: new feature

#769 - Add LogSoftmax

Pull Request - State: closed - Opened by kurnevsky over 1 year ago - 2 comments

#767 - [Breaking] Adding dilation/groups to Conv2D. Adding dilation to Pool2D

Pull Request - State: closed - Opened by coreylowman over 1 year ago
Labels: breaking-change

#765 - Placeholder f16 gemm speedups

Pull Request - State: closed - Opened by coreylowman over 1 year ago

#764 - [WIP] Adds rayon support to CPU

Pull Request - State: closed - Opened by coreylowman over 1 year ago - 16 comments

#762 - __hmin/__hmax already defined on compute_cap 75 with newer driver version

Issue - State: closed - Opened by opfromthestart over 1 year ago - 7 comments
Labels: bug

#760 - Adds TriangleTensor and CmpKernel traits to Device bound

Pull Request - State: closed - Opened by coreylowman over 1 year ago

#759 - Add support for custom function backward passes

Issue - State: open - Opened by mlupo19 over 1 year ago
Labels: new feature

#757 - Fixing compatibility for compute cap 70-75

Pull Request - State: closed - Opened by coreylowman over 1 year ago

#756 - Fix cuda kernels on compute_cap 75

Issue - State: closed - Opened by coreylowman over 1 year ago
Labels: bug

#755 - Dynamic dimensions in neural network layers?

Issue - State: closed - Opened by bplevin36 over 1 year ago - 1 comment
Labels: new feature

#754 - Removing rerun build.rs for output destination

Pull Request - State: closed - Opened by coreylowman over 1 year ago

#753 - Use `gemm` package for cpu matrix multiplication

Issue - State: closed - Opened by coreylowman over 1 year ago - 4 comments
Labels: optimization

#752 - Changed CUDA_ARCH in compatibility.cuh

Pull Request - State: closed - Opened by jafioti over 1 year ago - 1 comment

#751 - Allow `broadcast_like` to accept tensors OR shapes

Pull Request - State: closed - Opened by VasanthakumarV over 1 year ago

#750 - Adds TryConcatAlong to support Concat along any axis

Pull Request - State: closed - Opened by coreylowman over 1 year ago

#749 - Support for TensorRT

Issue - State: closed - Opened by soloist-v over 1 year ago - 9 comments

#748 - NaN weights after update with half precision

Issue - State: closed - Opened by jafioti over 1 year ago - 7 comments

#747 - Wrong CUDA_ARCH check in compatibility.cuh

Issue - State: closed - Opened by jafioti over 1 year ago - 3 comments

#746 - Allow broadcast_like to accept tensors OR shapes

Issue - State: closed - Opened by coreylowman over 1 year ago
Labels: good first issue

#745 - Add bf16 Dtype

Issue - State: open - Opened by coreylowman over 1 year ago
Labels: new feature

#744 - [Breaking] Optimizer config values are now all f64

Pull Request - State: closed - Opened by coreylowman over 1 year ago

#743 - v0.12.0 tracking issue

Issue - State: closed - Opened by coreylowman over 1 year ago - 3 comments

#742 - Fixes most of the failing tests for f16 on older GPUs

Pull Request - State: closed - Opened by ViliamVadocz over 1 year ago - 1 comment

#740 - Adding example

Pull Request - State: closed - Opened by sirandreww over 1 year ago - 2 comments

#737 - Adds rust 1.65 as the minimum rust compiler version

Pull Request - State: closed - Opened by coreylowman over 1 year ago

#736 - Fixing dependencies for no-std

Pull Request - State: closed - Opened by coreylowman over 1 year ago

#735 - Work on compatibility

Pull Request - State: closed - Opened by ViliamVadocz over 1 year ago - 1 comment

#734 - Having compile errors when doing cargo test from the dfdx repo

Issue - State: closed - Opened by tayglobal over 1 year ago - 4 comments

#733 - Use atomicCAS for atomicMin/Max

Pull Request - State: closed - Opened by ViliamVadocz over 1 year ago

#732 - Untested AtomicMin and AtomicMax for f16

Pull Request - State: closed - Opened by ViliamVadocz over 1 year ago

#731 - Concat along any dimension

Issue - State: closed - Opened by coreylowman over 1 year ago
Labels: new feature

#730 - [RFC] Adding `thread_cpu()` and functions to create tensors on cpu

Pull Request - State: closed - Opened by coreylowman over 1 year ago - 2 comments

#729 - Get size of tensor

Issue - State: closed - Opened by sirandreww over 1 year ago - 2 comments