Mohammad Miadh Angkad d7207be156 Fix startup weight load after TorchAO removal (#34869)

2026-08-14 13:18:53 -07:00

..

model_runner_components

Fix startup weight load after TorchAO removal (#34869)

2026-08-14 13:18:53 -07:00

runner

[Scheduler] Unify WAR read-done gating behind shared-read boundary declarations (#34052)

2026-08-08 03:26:36 -07:00

runner_backend

Profiling Enhancements [1/3]: cuda graph profile traces (#24370)

2026-08-06 03:19:35 -07:00

test_chunked_prefix_cache_gate.py

config: the draft runner carries its own attention backend

2026-08-05 19:32:24 -07:00

test_cuda_graph_buffer_registry.py

Fix pad-row top-k masking with custom_routing_function under DP attention (#31838)

2026-07-21 11:49:17 -07:00

test_forward_metadata_plan_record.py

Replace skip_attn_backend_init with a batch-carried attention plan marker (+ staleness re-plan) (#27193)

2026-06-04 17:13:18 -07:00

test_hisparse_pool_configurator.py

fix(hisparse): correct DSA KV memory budget (#31992)

2026-07-27 11:19:32 +08:00

test_mlp_sync_pad_unpad.py

Multiple flexibility fixes for DP attention (#33537)

2026-08-04 15:40:40 -07:00

test_pool_configurator.py

fix(dflash): account for DCP in draft KV pool sizing (#33912)

2026-08-10 17:15:01 -07:00

test_prefill_cuda_graph_runner_helpers.py

Enable multimodal prefill BCG for VL and audio models (#30872)

2026-07-29 06:47:40 +08:00

test_prefill_cuda_graph_runner.py

fix: always capture default prefill CUDA graph (#33352)

2026-08-08 19:24:49 +08:00