5 people 6491f481a7 [Bugfix] Stop allocator fragmentation from shrinking the KV cache during memory profiling (#58430) ...

Signed-off-by: Robert Shaw <robertgshaw2@gmail.com>
Signed-off-by: Robert Shaw <robertgshaw2-redhat@h100-01.nemg-001.lab.rdu2.dc.redhat.com>
Signed-off-by: Robert Shaw <robshaw@redhat.com>
Signed-off-by: Andreas Karatzas <Andreas.Karatzas@amd.com>
Co-authored-by: Robert Shaw <robertgshaw2-redhat@h100-01.nemg-001.lab.rdu2.dc.redhat.com>
Co-authored-by: Robert Shaw <robshaw@redhat.com>
Co-authored-by: Andreas Karatzas <Andreas.Karatzas@amd.com>
Co-authored-by: Andreas Karatzas <akaratza@amd.com>

2026-09-25 12:16:27 -04:00

..

__init__.py

[V1] Adding min tokens/repetition/presence/frequence penalties to V1 sampler (#10681)

2024-12-26 19:02:58 +09:00

test_attn_utils.py

[Bugfix][MRV2] Match fast-prefill padding to active LoRA batches (#56456)

2026-09-18 22:13:23 +00:00

test_cp_utils.py

DCP supports hybrid attention (#40996)

2026-07-09 21:34:45 -07:00

test_dspark_utils.py

[Bugfix][Spec Decode][MoE] Avoid uninitialized EPLB state in DeepSeek V4.1 DSpark drafter (#56387)

2026-09-15 14:35:47 +00:00

test_dsv4_packed_zeroer_geometry.py

[CI] Ignore ruff D209, rejoin the docstrings it split, and silence incompatible-rule warnings (#57212)

2026-09-16 10:57:24 -07:00

test_eagle3_aux_hidden_states_pp.py

[Core][MRV2] Support eagle3 spec decode with pipeline parallel (#50514)

2026-09-05 11:45:34 +08:00

test_encoder_runner.py

[MM][V2] Enable encoder-only ViT CUDA graph capture (#56922)

2026-09-17 16:12:17 +08:00

test_gpu_autoregressive_speculator.py

[PCP] Support prefill context parallelism with data parallelism (#57075)

2026-09-23 22:32:36 -04:00

test_gpu_bad_words.py

[Model Runner V2] Support custom logits processors (#56497)

2026-09-19 17:46:49 +00:00

test_gpu_batch_ordering.py

[Spec Decode] DSpark confidence-scheduled verification (#47808)

2026-08-12 12:27:48 -07:00

test_gpu_batch_shard.py

[Model Runner V2] Support custom logits processors (#56497)

2026-09-19 17:46:49 +00:00

test_gpu_block_table.py

[Bugfix][Spec Decode] Stop dummy draft decode steps from writing KV through stale block-table rows (#56734)

2026-09-21 01:06:25 +00:00

test_gpu_extract_hidden_states_speculator.py

[Model Runner V2] Acceptance estimation for adaptive verification (#52228)

2026-09-14 15:58:48 -07:00

test_gpu_gumbel_sample.py

Add pydocstyle to the ruff rules (#52136)

2026-09-16 16:44:29 +00:00

test_gpu_input_batch_v2.py

[MRV2] Miscellaneous code cleanup (#57980)

2026-09-23 01:51:52 -07:00

test_gpu_input_batch.py

Release prompt_embeds tensor when its InputBatch slot is freed (#57988)

2026-09-25 00:00:44 +00:00

test_gpu_kpool_tail_slot_mapping.py

[Bugfix][KV Cache][GLM-5.3-Flash] Disable slot mapping kernel for the kpool tail buffer (#57317)

2026-09-17 22:42:54 -04:00

test_gpu_kv_connector.py

[Bugfix][KVConnector] Finalize saves on steps without a forward (#57775)

2026-09-25 00:06:36 +00:00

test_gpu_logit_bias.py

[Model Runner V2] Support custom logits processors (#56497)

2026-09-19 17:46:49 +00:00

test_gpu_logits_processors.py

[Model Runner V2] Support custom logits processors (#56497)

2026-09-19 17:46:49 +00:00

test_gpu_model_runner_mm_gather.py

[CI] Ignore ruff D209, rejoin the docstrings it split, and silence incompatible-rule warnings (#57212)

2026-09-16 10:57:24 -07:00

test_gpu_model_runner_v2_cudagraph_profiling.py

[CI] Ignore ruff D209, rejoin the docstrings it split, and silence incompatible-rule warnings (#57212)

2026-09-16 10:57:24 -07:00

test_gpu_model_runner_v2_eplb.py

[Model Runner V2] Support custom logits processors (#56497)

2026-09-19 17:46:49 +00:00

test_gpu_model_runner_v2.py

[MRV2] Minor model_runner.py code cleanup (#58610)

2026-09-25 07:59:27 -07:00

test_gpu_model_runner.py

[Bugfix] Pick a KV block size supported by every attention backend (#49845)

2026-09-23 16:36:35 -04:00

test_gpu_pcp_manager.py

[PCP] Support prefill context parallelism with data parallelism (#57075)

2026-09-23 22:32:36 -04:00

test_gpu_profiler.py

[5/12][ci-selector][CI] Skip the Proton GPU test when another CUPTI tool is injected (#58455)

2026-09-23 17:25:41 -07:00

test_gpu_rejection_sampler_chunking.py

[Model Runner V2] Support custom logits processors (#56497)

2026-09-19 17:46:49 +00:00

test_gpu_rejection_sampler_i64.py

[CI] Ignore ruff D209, rejoin the docstrings it split, and silence incompatible-rule warnings (#57212)

2026-09-16 10:57:24 -07:00

test_gpu_sampler_flags.py

[Model Runner V2] Support custom logits processors (#56497)

2026-09-19 17:46:49 +00:00

test_gpu_thinking_budget.py

[Model Runner V2] Support custom logits processors (#56497)

2026-09-19 17:46:49 +00:00

test_gpu_trace_replay.py

[Core][V1] Support trace_decode_token_ids for deterministic decode replay (#46701)

2026-08-20 15:12:43 +08:00

test_gpu_ubatch_slicing.py

[MRV2] Miscellaneous code cleanup (#57980)

2026-09-23 01:51:52 -07:00

test_gpu_warmup_blocks.py

[3/N] HiSparse: host-resident sparse-MLA decode hot-buffering (#53781)

2026-09-12 00:36:47 -04:00

test_gpu_worker_weight_transfer.py

[Bugfix] Use DP index for dense DP weight updates and EC CPU region (#56950)

2026-09-17 22:44:36 +08:00

test_gpu_worker.py

[Bugfix] Stop allocator fragmentation from shrinking the KV cache during memory profiling (#58430)

2026-09-25 12:16:27 -04:00

test_jit_warmup_migration.py

[3/N][Feat][Perf] Add new warmup infrastructure for JITs. Add provider registry and orchestration for JIT warmup (#50174)

2026-08-17 10:11:14 -07:00

test_kv_block_zeroer.py

[CI] Ignore ruff D209, rejoin the docstrings it split, and silence incompatible-rule warnings (#57212)

2026-09-16 10:57:24 -07:00

test_kv_cache_allocation_scope.py

[5/N] Share HiSparse host cache across TP ranks (reopens #52760) (#56629)

2026-09-12 19:48:25 -04:00

test_late_interaction_runner.py

[Core] Move MRV1 late_interaction_runner.py out of MRV2 subtree (#48014)

2026-07-08 18:30:11 +00:00

test_logits_processor_loader.py

[Model Runner V2] Support custom logits processors (#56497)

2026-09-19 17:46:49 +00:00

test_mamba_hybrid_model_state.py

[Bugfix][MRV2] Treat padded prompt tails as spec-decode rows for hybrid models (#58434)

2026-09-24 18:29:53 -07:00

test_mamba_utils.py

[CI] Ignore ruff D209, rejoin the docstrings it split, and silence incompatible-rule warnings (#57212)

2026-09-16 10:57:24 -07:00

test_mixed_warmup_gate.py

[CI] Ignore ruff D209, rejoin the docstrings it split, and silence incompatible-rule warnings (#57212)

2026-09-16 10:57:24 -07:00

test_mrope_prompt_embeds.py

[Security] Fix DoS via prompt_embeds on M-RoPE models (#45252)

2026-06-13 10:17:38 +00:00

test_pp_utils.py

Revert "[DSpark] Support pipeline-parallel targets in aggregated serving (#56956)" (#58484)

2026-09-24 08:48:36 -07:00

test_prompt_embeds_state.py

[XPU][CI] enable prompt embeds tests on XPU (#58283)

2026-09-25 06:56:59 +08:00

test_sleep_mode_backend.py

[RL][Sleep] Retain frozen weights across level-2 sleep (#57891)

2026-09-22 12:36:55 +08:00

test_spec_decode_embed_sharing_pp.py

[Core][MRV2] Support eagle3 spec decode with pipeline parallel (#50514)

2026-09-05 11:45:34 +08:00

test_utils.py

[Bugfix][HiSparse][NIXL] Import full blocks without tail prefill on D (#57049)

2026-09-17 15:26:10 -04:00

test_worker_memory_snapshot.py

Add pydocstyle to the ruff rules (#52136)

2026-09-16 16:44:29 +00:00

test_workspace.py

[Core][Kernel] Share persistent workspaces for Marlin and Humming (#57421)

2026-09-20 17:28:33 +08:00

test_xpu_model_runner.py

[XPU] Fix CUDA API shims breaking Torch Dynamo during AOT compile (#43092)

2026-07-06 08:29:20 +08:00