* Add Llama 3.1 rope
* Clippy
* Format
* Clippy
* Add support for multiple eos tokens:
* Untagged either
* Remove either dep and fix settings.json
* Make the max positional embeddings configurable
* Add gradient test for conv_transpose2d with stride of 2.
* Swap dilation and stride in ConvTranspose2D backpropagation.
Without this, a shape mismatch occurs with a stride of 2 and dilation of 1.
* Add further tests of the ConvTranspose2D gradient.
Values calculated with torch, minor numerical errors adjusted and commented.
* Add the layernorm cuda kernels.
* Dedicated layer norm op.
* Add the slower variant.
* Plug the cuda implementation.
* Add the metal variant.
* Add a dedicated test.
* Bugfix.
* Add a slice_set op.
* Add some testing.
* Add the dedicated kv-cache module.
* Derive debug and clone.
* Expose more kv-cache functions.
* Return the current data when appending.
* Use the new cache in the quantized phi3 model.
* Separate quantized phi-3 implementation.
* Integrate the quantized phi3 model.=
* Small fixes, get the generation to work properly.
* Keep the old llama implementation around.
* Change the default.
* When converting a tensor to a variable, clone if the tensor is already a variable.
* Add a test to ensure training a batch norm works with VarMaps
---------
Co-authored-by: Jeffrey Dallatezza <jeffreydallatezza@Jeffreys-Laptop.local>
* add sigmoid op
* small fix
* add as a method on `Tensor`
* implement gradient calculation for sigmoid
* add sigmoid tests
* we should have a specialized op for this
* fix clippy
* fix clippy 2
* Revert all previous commits in favor of a `CustomOp` based solution
* use `CustomOp1` implementation
* fix rustfmt
* experimental add metal impl
* add cuda kernel impl
* fix fmt
* Add a test + reduce some cuda duplication.
---------
Co-authored-by: laurent <laurent.mazare@gmail.com>
* Add the cuda dequantize f16 kernels.
* Expose the cuda kernels.
* Add some testing + fix.
* Test the other cases too.
* A few more tests.
* Add an environment variable to enable the dequantize f16 + matmul behavior.
* Add the argsort cuda kernels.
* CPU version of arg-sort.
* Hook the cuda kernel + rework the cpu bits.
* Add some dedicated test.
* Working cuda kernel.
* Metal kernel.
* Metal adjustments.
* Bugfix.
* Use the fast rope in qwen.
* Rework the expert selection in qwen.
* add basic unary bench for sqrt
* process unary commands in tiles of 4
* re-enable all benchmarks
* rename helper to unary
* modify approach to split up tiled and non-tiled operations
* undo bench ignore for other tests
* update tile size to 2
* only perform the optimization on the contiguous even numbered element case
* Add the mmv kernels for smaller sizes.
* Support more mmv kernels.
* Use the new kernels.
* Fix the call.
* Silly fix.
* Improve the testing.
* Fix for dmmv.
* Add another dedicated test for the batching mmv.
* Fix for the batch dim in the quantized matmul example.
* Enable more tests on cuda.
* Add a test for qmm with a batch.
* Fix the zeros-dim test on metal.
* Hook the quantized matmul cuda kernels.
* Add a (currently broken) test.
* Kernel fixes.
* Fix by transposing the rhs matrix.
* Add the q4-1 kernels.
* Proper block sizes.
* More details in the tests.
* Move the metal kernels utils in a separate module.
* Use the BufferOffset for unary ops.
* Fix clippy lints.
* Use the new BufferOffset.
* Adapt the binary ops.
* Affine.
* More ops (powf, elu, cast).
