Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
100 commits
Select commit Hold shift + click to select a range
333a4df
feat(benchmark): restore text model qualification
chaofengw-nv Sep 17, 2026
95d7607
fix(benchmark): restore OPT performance workload
chaofengw-nv Sep 17, 2026
66136c5
feat(benchmark): qualify more text models
chaofengw-nv Sep 17, 2026
bac2b36
fix(builder): ignore unused alternate weights
chaofengw-nv Sep 17, 2026
97d3006
feat(benchmark): restore more text profiles
chaofengw-nv Sep 17, 2026
c9123f0
feat(benchmark): restore additional text profiles
chaofengw-nv Sep 17, 2026
9f85a07
fix(benchmark): align MoE accuracy references
chaofengw-nv Sep 17, 2026
3c5093a
fix(benchmark): reserve Llama BOS cache row
chaofengw-nv Sep 17, 2026
812a44b
feat(benchmark): qualify more generation models
chaofengw-nv Sep 17, 2026
9944d56
fix(benchmark): propagate trusted HF loading
chaofengw-nv Sep 17, 2026
34352cd
fix(benchmark): isolate InternLM reference environment
chaofengw-nv Sep 17, 2026
4fdcc23
fix(benchmark): inherit runtime packages in reference envs
chaofengw-nv Sep 17, 2026
3666653
fix(benchmark): support pinned Transformers references
chaofengw-nv Sep 17, 2026
996a9e8
feat(benchmark): qualify code generation models
chaofengw-nv Sep 17, 2026
79016f6
feat(benchmark): restore translation qualification
chaofengw-nv Sep 17, 2026
658534d
feat(benchmark): restore encoder qualification
chaofengw-nv Sep 17, 2026
7f20608
fix(benchmark): align embedding timing
chaofengw-nv Sep 17, 2026
d0d8f57
refactor(benchmark): keep model details family-local
chaofengw-nv Sep 17, 2026
d6153b1
fix(benchmark): prime compiled measurements
chaofengw-nv Sep 17, 2026
0b13115
test(runtime): preserve factory mismatch coverage
chaofengw-nv Sep 17, 2026
cf8ef5a
fix(benchmark): align translation qualification
chaofengw-nv Sep 18, 2026
a03cf18
feat(benchmark): add more text qualification profiles
chaofengw-nv Sep 18, 2026
bd7a030
fix(benchmark): align Llama accuracy prompts
chaofengw-nv Sep 18, 2026
b133c43
fix(llama): sanitize decoded UTF-8
chaofengw-nv Sep 18, 2026
dacf736
feat(benchmark): qualify additional text models
chaofengw-nv Sep 18, 2026
c7c058d
feat(benchmark): qualify large text models
chaofengw-nv Sep 18, 2026
441ab20
feat(benchmark): qualify current text models
chaofengw-nv Sep 18, 2026
2921b76
feat(benchmark): qualify more text models
chaofengw-nv Sep 18, 2026
8863bf2
feat(benchmark): qualify time-series models
chaofengw-nv Sep 18, 2026
5e987e9
feat(benchmark): qualify TIMM classifiers
chaofengw-nv Sep 18, 2026
edbe986
feat(benchmark): qualify speech recognition
chaofengw-nv Sep 18, 2026
88b3f2b
feat(benchmark): qualify DINOv3 image features
chaofengw-nv Sep 18, 2026
dab17d2
feat(benchmark): qualify diffusion text generation
chaofengw-nv Sep 18, 2026
84a1cdd
feat(benchmark): qualify detection and segmentation
chaofengw-nv Sep 18, 2026
c6c85e2
feat(benchmark): qualify YOLO detection
chaofengw-nv Sep 18, 2026
5a54262
feat(benchmark): qualify YOLOv10 detection
chaofengw-nv Sep 18, 2026
cf9a385
feat(benchmark): qualify YOLOv5 detection
chaofengw-nv Sep 18, 2026
1ac5bc0
feat(benchmark): qualify SAM segmentation
chaofengw-nv Sep 18, 2026
d1159aa
feat(benchmark): qualify Qwen vision language
chaofengw-nv Sep 18, 2026
c49e769
feat(benchmark): qualify InternVL generation
chaofengw-nv Sep 18, 2026
3cb78f8
feat(benchmark): qualify Phi-4 multimodal
chaofengw-nv Sep 18, 2026
1936b70
feat(benchmark): qualify Nemotron VL embedding
chaofengw-nv Sep 18, 2026
d7cd3eb
feat(benchmark): qualify Nemotron VL reranking
chaofengw-nv Sep 18, 2026
c7543d5
feat(benchmark): qualify SAM3 segmentation
chaofengw-nv Sep 18, 2026
473b095
feat(benchmark): qualify K2 Horizon Uno
chaofengw-nv Sep 18, 2026
3843c17
feat(benchmark): qualify DeepSeek OCR
chaofengw-nv Sep 18, 2026
5859fc3
feat(benchmark): qualify LocateAnything
chaofengw-nv Sep 18, 2026
20759e2
fix(benchmark): run vision qualification inputs
chaofengw-nv Sep 18, 2026
b7a12b5
fix(benchmark): accept qualification reranking inputs
chaofengw-nv Sep 18, 2026
8aaa0a2
fix(k2_horizon_uno): accept snapshot metadata files
chaofengw-nv Sep 18, 2026
4ed46d6
fix(benchmark): run vision qualification end to end
chaofengw-nv Sep 18, 2026
36f4993
fix(benchmark): support extension build environments
chaofengw-nv Sep 18, 2026
9b19e3d
fix(benchmark): load PatchTSMixer references
chaofengw-nv Sep 18, 2026
caeca45
feat(benchmark): qualify Fast Foundation Stereo
chaofengw-nv Sep 18, 2026
0af8dbc
feat(benchmark): qualify MoGe geometry
chaofengw-nv Sep 19, 2026
17c94b0
feat(benchmark): qualify LeRobot ACT control
chaofengw-nv Sep 19, 2026
e8f57be
refactor(test): simplify fake task factory
chaofengw-nv Sep 19, 2026
042d6e5
style(test): format fake segmentation signature
chaofengw-nv Sep 19, 2026
b5bcc7b
fix(benchmark): align execution contracts
chaofengw-nv Sep 19, 2026
ba49e8c
fix(benchmark): retain perception outputs
chaofengw-nv Sep 19, 2026
fc91f8b
fix(benchmark): keep YOLO NMS in inference mode
chaofengw-nv Sep 19, 2026
b68b9da
fix(canary): declare benchmark source language
chaofengw-nv Sep 19, 2026
72ed654
fix(benchmark): align qualification task inputs
chaofengw-nv Sep 19, 2026
f1ba8e6
fix(timm): use family-owned benchmark tasks
chaofengw-nv Sep 19, 2026
1a9b212
fix(fast-foundation-stereo): recognize official checkpoint
chaofengw-nv Sep 19, 2026
b047eb3
fix(benchmark): support legacy monocular geometry
chaofengw-nv Sep 19, 2026
33f284c
fix(benchmark): use prepared candidate model directories
chaofengw-nv Sep 19, 2026
ea1d7fd
fix(benchmark): separate MoGe release scope
chaofengw-nv Sep 19, 2026
ffb2b34
feat(benchmark): qualify remaining ready models
chaofengw-nv Sep 19, 2026
2a4b5bf
fix(benchmark): preserve Bark prompts
chaofengw-nv Sep 19, 2026
ec6947d
fix(ci): align benchmark contract tests
chaofengw-nv Sep 19, 2026
1f3d134
fix(benchmark): preserve scalar image requests
chaofengw-nv Sep 19, 2026
4fb90e3
fix(benchmark): retain legacy media outputs
chaofengw-nv Sep 19, 2026
c706dd3
fix(benchmark): bound reference extension builds
chaofengw-nv Sep 19, 2026
14b0d23
fix(benchmark): install PersonaPlex audio runtime
chaofengw-nv Sep 19, 2026
6e37282
fix(benchmark): route text-only image requests
chaofengw-nv Sep 19, 2026
78c5708
fix(benchmark): localize Sana reference weights
chaofengw-nv Sep 19, 2026
5c3132b
fix(benchmark): use stable Z-Image reference precision
chaofengw-nv Sep 19, 2026
c6c7b21
fix(benchmark): support native reference builds
chaofengw-nv Sep 19, 2026
dee0339
fix(benchmark): use headless OpenCV for SANA
chaofengw-nv Sep 19, 2026
d6a7adc
fix(benchmark): prepare Lance image reference
chaofengw-nv Sep 19, 2026
208134f
fix(benchmark): stabilize Qwen Image reference timing
chaofengw-nv Sep 19, 2026
eeccdd6
fix(benchmark): materialize SANA text encoder
chaofengw-nv Sep 19, 2026
b50b8a3
fix(benchmark): pin Wan reference attention
chaofengw-nv Sep 20, 2026
b8c4c9e
fix(benchmark): use prepared SANA candidate
chaofengw-nv Sep 20, 2026
84b9dc7
fix(benchmark): retain legacy image features
chaofengw-nv Sep 20, 2026
41c009e
fix(sana-wm): honor TensorRT install paths
chaofengw-nv Sep 20, 2026
e312bfe
fix(sana-wm): discover TensorRT Python library
chaofengw-nv Sep 20, 2026
bfc9272
fix(benchmark): restore report JSON filename
chaofengw-nv Sep 20, 2026
05c79a3
fix(benchmark): gate detection on COCO ground truth
chaofengw-nv Sep 21, 2026
c7234e8
fix(benchmark): diversify qualification samples
chaofengw-nv Sep 21, 2026
94c5e40
chore(benchmark): split first text batch
chaofengw-nv Sep 21, 2026
4b971a3
chore(benchmark): split two text batches
chaofengw-nv Sep 21, 2026
0136f3b
fix(benchmark): decouple family qualification policy
chaofengw-nv Sep 22, 2026
e053d61
fix(benchmark): identify TIMM reference backend
chaofengw-nv Sep 22, 2026
c7d92eb
fix(benchmark): isolate qualification policy
chaofengw-nv Sep 22, 2026
c2446d2
refactor(benchmark): separate qualification runtime
chaofengw-nv Sep 22, 2026
1285ba8
fix(ci): satisfy source quality checks
chaofengw-nv Sep 22, 2026
962335c
fix(ci): match pinned clang format
chaofengw-nv Sep 22, 2026
94b375e
fix(sana-wm): load reference encoder locally
chaofengw-nv Sep 22, 2026
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
The table of contents is too big for display.
Diff view
Diff view
  •  
  •  
  •  
216 changes: 181 additions & 35 deletions apps/benchmark/native/benchmark_worker.cpp

Large diffs are not rendered by default.

356 changes: 0 additions & 356 deletions apps/benchmark/performance/README.md

This file was deleted.

392 changes: 0 additions & 392 deletions apps/benchmark/performance/baselines/elf_reference.py

This file was deleted.

54 changes: 0 additions & 54 deletions apps/benchmark/performance/baselines/reference_support.py

This file was deleted.

Loading
Loading