3 people 4ac452ad98 [Core] Release NCCL communicator memory in sleep mode (#51485) ...

Signed-off-by: aoshen02 <aoshen02@users.noreply.github.com>
Co-authored-by: aoshen02 <aoshen02@users.noreply.github.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>

2026-09-01 14:31:16 +08:00

..

__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

[Spec Decode] Enable adaptive DSpark on SM100 sparse MLA (#52783)

2026-08-25 13:31:15 -04:00

test_cp_utils.py

DCP supports hybrid attention (#40996)

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

test_dsv4_packed_zeroer_geometry.py

[6/N][KV-Cache Layout Refactor] Standardize KV cache layout (#51718)

2026-08-21 17:47:49 -07:00

test_encoder_runner.py

[ModelRunnerV2] Support prompt embeds (#42963)

2026-08-17 21:58:50 +00:00

test_gpu_autoregressive_speculator.py

[Model] Support Qwen3.8-Flash-Next (#53896)

2026-08-31 05:57:56 +00:00

test_gpu_bad_words.py

[Bugfix][Model Runner V2][Spec Decode] Fix off-by-one in bad_words draft-prefix matching (#52311)

2026-08-16 09:02:20 +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

[Bugfix][PP] Never drop a decoding request from the sampled-token broadcast (#54436)

2026-08-31 23:49:54 +00:00

test_gpu_block_table.py

[Model] Support Qwen3.8-Flash-Next (#53896)

2026-08-31 05:57:56 +00:00

test_gpu_extract_hidden_states_speculator.py

[Feature][Model Runner V2] Support extract_hidden_states speculation (#49811)

2026-08-20 11:13:43 -04:00

test_gpu_gumbel_sample.py

[Bugfix][Model Runner V2][Spec Decode] Decouple the draft's gumbel noise stream from the target's (#54282)

2026-08-29 09:58:29 -07:00

test_gpu_input_batch_v2.py

[XPU][CI] Use platform device in InputBatch V2 test (#49939)

2026-07-27 15:28:02 +08:00

test_gpu_input_batch.py

[Pooling] Improve BGE-M3 sync pooling throughput by up to 3.13% (#53464)

2026-08-24 08:33:32 -04:00

test_gpu_kv_connector.py

[Core][KV Connector] Start async KV loads after the forward launch when no sync loads are scheduled (#53333)

2026-08-28 21:58:00 +00:00

test_gpu_logit_bias.py

[Structured Output] Let terminal grammars stop under min_tokens (II) (#54218)

2026-08-31 20:39:27 +04:00

test_gpu_model_runner_mm_gather.py

Encoder cache extension hooks (#48218)

2026-07-24 02:52:00 -07:00

test_gpu_model_runner_v2_cudagraph_profiling.py

[Bugfix] Reset cached Mamba align metadata on profiling teardown (#54044)

2026-08-30 00:43:20 -07:00

test_gpu_model_runner_v2_eplb.py

[1/N][warmup][DSv4] Migrate generic MLA metadata and indexing kernels (#50175)

2026-08-31 23:26:05 -04:00

test_gpu_model_runner_v2.py

[Model] Support Qwen3.8-Flash-Next (#53896)

2026-08-31 05:57:56 +00:00

test_gpu_model_runner.py

[Nixl][PD] DCP support for MLA models (#50611)

2026-08-29 16:03:53 +00:00

test_gpu_pcp_manager.py

BugFix(PCP): use persistent input buffers for PIECEWISE CUDA graphs (#53515)

2026-08-26 05:58:13 +00:00

test_gpu_profiler.py

[Profiler] Fix start_profile permanently no-op after max_iterations auto-stop (#51839)

2026-08-25 08:13:29 -04:00

test_gpu_rejection_sampler_chunking.py

[Feature][DSpark]: Logprobs adaptive verification (#52242)

2026-08-25 10:22:33 -04:00

test_gpu_rejection_sampler_i64.py

[Bugfix][Model Runner V2][Spec Decode] Fix int32 offset overflow in block verification kernels (#47383)

2026-07-02 07:32:38 -07:00

test_gpu_sampler_flags.py

[Performance][MRV2] Cache logits-processing request state (#52329)

2026-08-17 02:36:41 +00:00

test_gpu_thinking_budget.py

[Performance][MRV2] Cache logits-processing request state (#52329)

2026-08-17 02:36:41 +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_warmup_blocks.py

[Model] Support Qwen3.8-Flash-Next (#53896)

2026-08-31 05:57:56 +00:00

test_gpu_worker_weight_transfer.py

[RL] Add rank-local IPC weight updates (#52497)

2026-08-26 15:36:57 +08:00

test_gpu_worker.py

[ROCm][CI] Keep startup profiling from aborting when free memory grows (#53591)

2026-08-28 00:02:40 -07:00

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

[6/N][KV-Cache Layout Refactor] Standardize KV cache layout (#51718)

2026-08-21 17:47:49 -07:00

test_kv_cache_allocation_scope.py

[Core] Release NCCL communicator memory in sleep mode (#51485)

2026-09-01 14:31:16 +08:00

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_mamba_hybrid_model_state.py

[K3] support recoverssm for K3 (#51855)

2026-08-17 19:34:14 +08:00

test_mamba_utils.py

[Model] Support Qwen3.8-Flash-Next (#53896)

2026-08-31 05:57:56 +00:00

test_mixed_warmup_gate.py

[BugFix] Gate MRV2 mixed sparse-MLA warmup on max_num_seqs > 1 (#47050)

2026-06-30 16:31:27 +01:00

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

[Bugfix][PP] Never drop a decoding request from the sampled-token broadcast (#54436)

2026-08-31 23:49:54 +00:00

test_prompt_embeds_state.py

[ModelRunnerV2] Support prompt embeds (#42963)

2026-08-17 21:58:50 +00:00

test_sleep_mode_backend.py

[Core] Release NCCL communicator memory in sleep mode (#51485)

2026-09-01 14:31:16 +08:00

test_utils.py

[V0 Deprecation] Refactor kv cache from list to element (#37487)

2026-03-23 20:10:11 -07:00

test_worker_memory_snapshot.py

[Hardware] Replace torch.cuda.device_count/current_device/set_device API (#36145)

2026-03-12 07:57:47 -07:00

test_workspace.py

[Bugfix] Make DSV4 sparse MLA work end-to-end for plain decode, MTP, and DSpark (#51538)

2026-08-15 07:01:32 -07:00

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