Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

how much conflict? #10

Open
wants to merge 131 commits into
base: umerge
Choose a base branch
from
Open

how much conflict? #10

wants to merge 131 commits into from

Conversation

Flasew
Copy link
Owner

@Flasew Flasew commented Dec 27, 2021

No description provided.

Efrainq07 and others added 30 commits March 29, 2021 20:25
…ting structure. (flexflow#158)

* [FX] Deleted previously added 'intermedia_values' dict as it was redundant

* [FX] Refactores floor_divide into floor_divide and scalar_floor_divide and added floor_divide between scalars.

* [FX] Changed reshape so it accounts for polymorphism of shape.
* [FX] Deleted previously added 'intermedia_values' dict as it was redundant

* [FX] Refactores floor_divide into floor_divide and scalar_floor_divide and added floor_divide between scalars.

* [FX] Changed reshape so it accounts for polymorphism of shape.

* [Op] Added support for scalar addition operator on FF source

* [FX] Added parser for scalar addition operator

* [FX] Scalar addition operator can now be read from .ff file

* [Op] Added upport for more scalar operators on FF source

* [FX] Added parsers for more scalar operators

* [FX] FF can now open scalar operators from .ff file
…#150)

* [CMake]: use file(FLOB_RECURSE ...) to collect source files

* [CMake]: replace FILTER with REMOVE_ITEM

* [CMake]: clean up tanh and embedding_avx2

* [CMake]: enable assertions even if in release mode

* [CMake]: fix FF_BUILD_ALL_EXAMPLES
Fixes the following error:

CMake Error at CMakeLists.txt:103:
  Parse error.  Expected "(", got newline with text "

  ".
* aggregate and group_by, initial

* [Group_by] cpp impl

* [Group_by] Output list of Tensors

* [Group_by] dont pass k

* hot

* [Aggregate] bug fixes

* [TopK] output print

* [MoE] Jacobs '91 first model

* [MoE] minor

* delete examples/try_out

* [agg_spec] hot

* [agg_spec] Specification loss, first

* [MoE] add balance term to loss

Co-authored-by: Zhihao Jia <[email protected]>
Now that March release is done, make pybind11 the default.
* aggregate and group_by, initial

* [Group_by] cpp impl

* [Group_by] Output list of Tensors

* [Group_by] dont pass k

* hot

* [Aggregate] bug fixes

* [TopK] output print

* [MoE] Jacobs '91 first model

* [MoE] minor

* delete examples/try_out

* [agg_spec] hot

* [agg_spec] Specification loss, first

* [MoE] add balance term to loss

* [Agg] CUDA implementation

* [AggSpec] CUDA implementation

* [GroupBy] CUDA implementation

* [MoE] Only call malloc/free on construct and destruct

* [MoE] minor

Co-authored-by: Zhihao Jia <[email protected]>
* aggregate and group_by, initial

* [Group_by] cpp impl

* [Group_by] Output list of Tensors

* [Group_by] dont pass k

* hot

* [Aggregate] bug fixes

* [TopK] output print

* [MoE] Jacobs '91 first model

* [MoE] minor

* delete examples/try_out

* [agg_spec] hot

* [agg_spec] Specification loss, first

* [MoE] add balance term to loss

* [Agg] CUDA implementation

* [AggSpec] CUDA implementation

* [GroupBy] CUDA implementation

* [MoE] Only call malloc/free on construct and destruct

* [MoE] minor

* [Cache] first impl

* [Cache] trigger funtion for switching modes

* [TopK] fix grads set to 0

* minor

* [Agg] Minor gating net grads

* [MoE] profiling

* [Cache] recompile first

Co-authored-by: Zhihao Jia <[email protected]>
* add supoort for hip backend on NVIDIA GPU

* minor fix

* try fix cudaEventRecord

* put set cudnn cublas stream into backward/forward_kernel functions

* clean up DISABLE_LEGION_CUDA_HIJACK

* rename create_stream to get_legion_stream
* python: create a pybind wrapper

* [pybind]: add compute_metrics
[CMake] change default build mode to Release
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

8 participants