COMMITS
/ backends/cadence/aot/compiler.py May 8, 2025
M
Extract trace from prepare_and_convert and remove export_program
mcremon-meta committed
May 6, 2025
M
Make quantize_pt2 private and remove external call sites
mcremon-meta committed
May 5, 2025
M
Make lower_ep_to<edge, cadence, executorch> private functions and remove external call sites
mcremon-meta committed
May 3, 2025
M
Make quantize_pt2 return an ExportedProgram instead of a GraphModule
mcremon-meta committed
April 25, 2025
M
Add quantize_and_export_to_edge and quantize_and_export_to_executorch
mcremon-meta committed
April 24, 2025
M
Rename convert_pt2
mcremon-meta committed
April 11, 2025
V
Prevent decomposing RMSNorm in Jarvis (#10074)
Val Tarasyuk committed
April 2, 2025
G
executorch
gmagogsfm committed
April 1, 2025
M
Allow passing in calibration data to `convert_pt2`
mcremon-meta committed
March 31, 2025
M
Introduce missing APIs to lower ExportedProgram objects directly
mcremon-meta committed
March 11, 2025
W
Adding linalg_svd and linag_det ops to executorch
wahmed991 committed
February 26, 2025
S
Update CadenceMemoryPlanning to support per-memory alignment constraint
sabarishsnk committed
February 7, 2025
S
Add option to enforce alignment constraint when planning memory
skrtskrtfb committed
January 29, 2025
M
Improve graph logging for debug purposes
mcremon-meta committed
January 28, 2025
G
Revert D68246404
Gasoonjia committed
G
turn on dim order in cadence test
Gasoonjia committed
January 25, 2025
H
Enable constant methods in cadence export.
Hardik Sharma committed
January 24, 2025
M
Make the quantized path the main testing path, and introduce a nop quantizer for fp32 cases
mcremon-meta committed
January 9, 2025
M
Refactor CadenceQuantizer
mcremon-meta committed
January 7, 2025
M
Call eval() in `quantize_pt2`
mcremon-meta committed
January 3, 2025
M
Remove path that is not needed for SDPA decomp
mcremon-meta committed
December 23, 2024
S
Added Cadence memory planning example
skrtskrtfb committed
December 20, 2024
G
executorch/exir/program/test
gmagogsfm committed
H
Enable IR checks
Hardik Sharma committed
December 2, 2024
M
Run decompositions before the quantizer
mcremon-meta committed
November 22, 2024
November 21, 2024
November 20, 2024
J
add simply ops to oss, update fuse simply callsites
JP committed
November 19, 2024
November 12, 2024
J
migrate passes calls to pass utils
JP committed
October 12, 2024
Z
fixed empty output from runtime executor in CPU flow (#6184)
Zonglin Peng committed
October 10, 2024
T
Move cadence.aot.passes to cadence.aot._passes (#5921)
Tarun Karuturi committed
October 7, 2024
M
Cleanup export_model API calls (#5882)
Matthias Cremon committed
September 27, 2024
M
Migrate from capture_pre_autograd_graph to torch.export.export_for_training (#5730)
Matthias Cremon committed
August 26, 2024
A
typing for decorators - fx/_compatibility
Aaron Orenstein committed
August 22, 2024
M
Fix SDPA decomp problem
mcremon-meta committed
August 1, 2024
M
Split quantize_pt2 to allow calling the same APIs in testing and regular flows (#4505)
Matthias Cremon committed
July 30, 2024
M
Add exportable baby llama example (#4345)
Matthias Cremon committed
July 22, 2024
M
Fix quantized_matmul with 4D inputs (#4335)
Matthias Cremon committed
July 16, 2024
M
Migrate the quantizer to use aten ops directly (#4195)
Matthias Cremon committed
July 15, 2024
M
Move SDPA decomp pass from Qualcomm's directory to be shareable and call it for Cadence backends (#4258)
Matthias Cremon committed
July 12, 2024
M
Add support for quantized bmm (#4047)
Matthias Cremon committed
June 20, 2024
M
Move the quantization API to OSS (#3997)
Matthias Cremon committed
June 7, 2024
June 5, 2024
M
Cadence examples -- Add RNNT encoder from torchaudio (#3691)
Matthias Cremon committed
May 23, 2024
M
Clean up lowering APIs (#3690)
Matthias Cremon committed
May 21, 2024
M
Remove pt2_quant flag (#3676)
Matthias Cremon committed
May 7, 2024
M
Cadence - Move primary code to backends folder (#3353)
Matthias Cremon committed