@@ -4,6 +4,7 @@ This file contains the changelog for the Deeploy project. The changelog is divid
44## Unreleased (Planned Release Target: v0.2.1)
55
66### List of Pull Requests
7+ - Refactors and fixes [ #131 ] ( https://github.com/pulp-platform/Deeploy/pull/131 )
78- Disallow shape inference [ #128 ] ( https://github.com/pulp-platform/Deeploy/pull/128 )
89- Remove memory-aware node bindings [ #123 ] ( https://github.com/pulp-platform/Deeploy/pull/123 )
910- Fix missing const's layout transformation and refactor NCHWtoNHWC passes [ #122 ] ( https://github.com/pulp-platform/Deeploy/pull/122 )
@@ -91,6 +92,8 @@ This file contains the changelog for the Deeploy project. The changelog is divid
9192- Removed Wmem variants of bindings and tile constraints from Neureka
9293- Disabled ICCT_ITA_8 MemPool test because it was using a lowering that created shapeless tensors
9394- Added missing shape annotation to the testTypeInferenceDifferentTypes
95+ - ref naming scheme in Gemm and FloatGemm templates from ${data_out}_ ${<tensor >} to ${nodeName}_ ${<tensor >}
96+ - move iNoNorm from Generic to Snitch since it uses a Snitch kernel
9497
9598### Fixed
9699- Prevent node duplication for graphs generated via GraphSurgeon
@@ -105,6 +108,8 @@ This file contains the changelog for the Deeploy project. The changelog is divid
105108- Missing layout transformation of the const's (bias, mul, add, shift in Conv/RequantizedConv)
106109- Keep mul/add rank of requantized Neureka tile constraints
107110- Fix bias hoisting in generic GEMM with no bias
111+ - formatting of test_input/output integer values
112+ - pulp rqs tile constraints now properly target the last dimension of rqs params
108113
109114### Removed
110115- Delete outdated and unused ` .gitlab-ci.yml ` file
@@ -180,9 +185,9 @@ This release containing major architectural changes, new platform support, enhan
180185
181186
182187### Added
183- - BatchNorm kernel
184- - ConvTranspose kernel
185- - MaxPool1D kernel
188+ - BatchNorm kernel
189+ - ConvTranspose kernel
190+ - MaxPool1D kernel
186191- Template for 1D Convolution
187192- Support for float32 data type in the previous kernels
188193- Float binding for Pad1D kernel
@@ -321,7 +326,7 @@ This release containing major architectural changes, new platform support, enhan
321326
322327### Changed
323328- FloatConvTemplate file
324- - Platform.py file
329+ - Platform.py file
325330- Bump the CMake version to 3.24 as required for the chimera-sdk
326331- Bump GVSoC's version and add chimera simulation target
327332- Rename the generic source util to utils to avoid name collision with chimera-sdk
0 commit comments