4 people dff1bde84d [Model] Qwen4Exp: fp8_e4m3 main KV cache on the QSA path (#55557) ...

Signed-off-by: Andrea Semeraro <andrea.semeraro@sezione1.it>
Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
Co-authored-by: Jiangyun Zhu <riverclouds.zhu@qq.com>

2026-09-16 15:07:52 +00:00

..

deepseek_v32

[Perf] Optimize DeepSeek V3.2 sequence parallelism (#51434)

2026-08-07 16:51:01 -07:00

fixtures

[BUGFIX] Fix Pixtral consolidated format vision weight loading (#39916)

2026-04-19 22:25:03 -07:00

glm5next

[Perf][GLM-5.3-Flash] Decode hot-path cleanups: strided KDA recurrent inputs, NoPE MQA query without concat, no duplicate router GEMM (#55736)

2026-09-10 16:28:33 +08:00

inkling

[3/N][warmup][DSv4] Migrate FA4 MLA and shared CuTeDSL kernels (#53565)

2026-09-07 09:36:35 -04:00

kimi_k3

[Bugfix][Kimi-K3] Do not classify a stateless first chunk as a decode (#51483)

2026-09-16 21:24:48 +08:00

language

[Pooling] Report actual input token usage for scoring APIs (#56573)

2026-09-13 09:20:21 -04:00

multimodal

[Bugfix] Make GPU sync checks safe under torch.compile (#56904)

2026-09-15 21:37:08 -07:00

quantization

[Kernel] Add NVFP4 support to the torch linear backend (#53319)

2026-09-08 20:30:05 -04:00

qwen4_exp

[Model] Qwen4Exp: fp8_e4m3 main KV cache on the QSA path (#55557)

2026-09-16 15:07:52 +00:00

transformers

[ROCm][CI] Use a platform-independent GEMM in the merged-column fuser test (#56130)

2026-09-09 15:08:39 -05:00

__init__.py

[CI/Build] Move test_utils.py to tests/utils.py (#4425)

2024-05-13 23:50:09 +09:00

registry.py

[Model] Add support for Nanbeige4.2 (transformers backend) (#56071)

2026-09-14 15:06:01 +00:00

test_adapters.py

[BugFix][Pooling] Skip weight-prefix probe when model has WeightsMapper (#50890)

2026-08-05 21:12:33 -07:00

test_deepseek_v4_fi_moe_ep.py

[Model][MoE] DeepSeek-V4: add opt-in FlashInfer moe_ep expert backend (#49636)

2026-08-25 03:35:01 +00:00

test_deepseek_v4_mega_moe.py

[Perf][DSV4.1] Pad shared experts for native MegaMoE fusion (#56568)

2026-09-16 17:08:42 +08:00

test_deepseek_v4_rocm_compressor_gemm_fusion.py

[ROCm][DSV4][Perf] Fuse DeepSeek V4 C4 compressor GEMMs (#53838)

2026-08-26 23:36:14 +08:00

test_deepseek_v4_vl_rocm.py

[Model][ROCm] Enable DeepSeek V4 Vision (#55107)

2026-09-11 09:17:30 -07:00

test_dspark_mla.py

[Refactor] Normalize DeepSeek V4.1 model package naming (#56741)

2026-09-13 23:17:50 -07:00

test_hyv4_rocm.py

[ROCm][CI] Add HY-V4 generation coverage (#55667)

2026-09-11 17:24:28 +08:00

test_initialization.py

[Model][ROCm] Enable DeepSeek V4 Vision (#55107)

2026-09-11 09:17:30 -07:00

test_language_model_cache_is_weak.py

[Bugfix][MRV2] Release model and KV cache on in-process engine shutdown (#54162)

2026-08-28 22:09:08 -07:00

test_qwen3_5_mtp_config.py

[Bugfix][Spec Decode] Resolve n_predict from text_config for Qwen3.5 multimodal MTP (#55369)

2026-09-07 13:28:26 +00:00

test_registry.py

[Model][ROCm] Enable DeepSeek V4 Vision (#55107)

2026-09-11 09:17:30 -07:00

test_terratorch.py

[CI/Build] Skip Prithvi/Terratorch model-registry tests when terratorch is missing (#41389)

2026-04-30 12:47:55 -07:00

test_utils.py

Reduce AutoWeightsLoader kwargs (#53106)

2026-08-20 15:40:11 +00:00

test_vision.py

[CI][ROCm] Standardize AMD test job labels by device (#52976)

2026-08-19 22:33:25 -05:00

utils.py

[Hy4] support Hy4-preview model (#54160)

2026-08-29 12:07:28 +08:00