Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[torch.compile] hide slicing under custom op for inductor #8384

Merged
merged 11 commits into from
Sep 12, 2024

Conversation

youkaichao
Copy link
Member

see pytorch/pytorch#131192

when inductor sees a view being mutated, it will copy the tensor.

hiding the slicing operation under custom op solves the issue for inductor.

Copy link

👋 Hi! Thank you for contributing to the vLLM project.
Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run fastcheck CI which starts running only a small and essential subset of CI tests to quickly catch errors. You can run other CI tests on top of those by going to your fastcheck build on Buildkite UI (linked in the PR checks section) and unblock them. If you do not have permission to unblock, ping simon-mo or khluu to add you in our Buildkite org.

Once the PR is approved and ready to go, your PR reviewer(s) can run CI to test the changes comprehensively before merging.

To run CI, PR reviewers can do one of these:

  • Add ready label to the PR
  • Enable auto-merge.

🚀

@youkaichao
Copy link
Member Author

cc @laithsakka

@youkaichao
Copy link
Member Author

after this pr, the major bug of inductor ( it tries to copy kvcache) should be fixed.

however, there are still some inefficiencies w.r.t. inductor. it cannot deal with inplace custom op well.

@youkaichao youkaichao merged commit 7de49aa into vllm-project:main Sep 12, 2024
25 of 29 checks passed
@youkaichao youkaichao deleted the fix_inductor branch September 12, 2024 07:11
@bnellnm
Copy link
Contributor

bnellnm commented Sep 12, 2024

Thanks for finding and fixing this!

@laithsakka
Copy link

This is fixed on pytorch 2.5 by setting TORCHDYNAMO_AUTO_FUNCTIONALIZED_V2=1

@laithsakka
Copy link

can you elaborate more on "there are still some inefficiencies w.r.t. inductor. it cannot deal with inplace custom op well."

@youkaichao
Copy link
Member Author

you can take a look at the log when you run with inductor:

# GPU blocks: 790

when I turn on inductor, the blocks become smaller, which means inductor takes more memory and we have less memory for the kv cache.

@laithsakka
Copy link

takes more memory and we have less memory for the kv cache.

I see i will create an issue to revisit this, for the models i ran inductor on , it was slightly better than not using torch.compile. I will sync with you on that
pytorch/pytorch#136269

MengqingCao added a commit to MengqingCao/vllm that referenced this pull request Oct 10, 2024
[Bugfix] Streamed tool calls now more strictly follow OpenAI's format; ensures Vercel AI SDK compatibility (vllm-project#8272)

[Frontend] Add progress reporting to run_batch.py (vllm-project#8060)

Co-authored-by: Adam Lugowski <adam.lugowski@parasail.io>

[Bugfix] Correct adapter usage for cohere and jamba (vllm-project#8292)

[Misc] GPTQ Activation Ordering (vllm-project#8135)

[Misc] Fused MoE Marlin support for GPTQ (vllm-project#8217)

Add NVIDIA Meetup slides, announce AMD meetup, and add contact info (vllm-project#8319)

[Bugfix] Fix missing `post_layernorm` in CLIP (vllm-project#8155)

[CI/Build] enable ccache/scccache for HIP builds (vllm-project#8327)

[Frontend] Clean up type annotations for mistral tokenizer (vllm-project#8314)

[CI/Build] Enabling kernels tests for AMD, ignoring some of then that fail (vllm-project#8130)

Fix ppc64le buildkite job (vllm-project#8309)

[Spec Decode] Move ops.advance_step to flash attn advance_step (vllm-project#8224)

[Misc] remove peft as dependency for prompt models (vllm-project#8162)

[MISC] Keep chunked prefill enabled by default with long context when prefix caching is enabled (vllm-project#8342)

[Bugfix] lookahead block table with cuda graph max capture (vllm-project#8340)

[Bugfix] Ensure multistep lookahead allocation is compatible with cuda graph max capture (vllm-project#8340)

[Core/Bugfix] pass VLLM_ATTENTION_BACKEND to ray workers (vllm-project#8172)

[CI/Build][Kernel] Update CUTLASS to 3.5.1 tag (vllm-project#8043)

[Misc] Skip loading extra bias for Qwen2-MOE GPTQ models (vllm-project#8329)

[Bugfix] Fix InternVL2 vision embeddings process with pipeline parallel (vllm-project#8299)

[Hardware][NV] Add support for ModelOpt static scaling checkpoints. (vllm-project#6112)

[model] Support for Llava-Next-Video model (vllm-project#7559)

Co-authored-by: Roger Wang <ywang@roblox.com>
Co-authored-by: Cyrus Leung <tlleungac@connect.ust.hk>
Co-authored-by: Cyrus Leung <cyrus.tl.leung@gmail.com>

[Frontend] Create ErrorResponse instead of raising exceptions in run_batch (vllm-project#8347)

[Model][VLM] Add Qwen2-VL model support (vllm-project#7905)

Co-authored-by: Roger Wang <136131678+ywang96@users.noreply.github.com>
Co-authored-by: DarkLight1337 <tlleungac@connect.ust.hk>

[Hardware][Intel] Support compressed-tensor W8A8 for CPU backend (vllm-project#7257)

[CI/Build] Excluding test_moe.py from AMD Kernels tests for investigation (vllm-project#8373)

[Bugfix] Add missing attributes in mistral tokenizer (vllm-project#8364)

[Kernel][Misc] register ops to prevent graph breaks (vllm-project#6917)

Co-authored-by: Sage Moore <sage@neuralmagic.com>

[Misc] Move device options to a single place (vllm-project#8322)

[Speculative Decoding] Test refactor (vllm-project#8317)

Co-authored-by: youkaichao <youkaichao@126.com>

Pixtral (vllm-project#8377)

Co-authored-by: Roger Wang <ywang@roblox.com>

Bump version to v0.6.1 (vllm-project#8379)

[MISC] Dump model runner inputs when crashing (vllm-project#8305)

[misc] remove engine_use_ray (vllm-project#8126)

[TPU] Use Ray for default distributed backend (vllm-project#8389)

Fix the AMD weight loading tests (vllm-project#8390)

[Bugfix]: Fix the logic for deciding if tool parsing is used (vllm-project#8366)

[Gemma2] add bitsandbytes support for Gemma2 (vllm-project#8338)

[Misc] Raise error when using encoder/decoder model with cpu backend (vllm-project#8355)

[Misc] Use RoPE cache for MRoPE (vllm-project#8396)

[torch.compile] hide slicing under custom op for inductor (vllm-project#8384)

[Hotfix][VLM] Fixing max position embeddings for Pixtral (vllm-project#8399)

[Bugfix] Fix InternVL2 inference with various num_patches (vllm-project#8375)

Co-authored-by: DarkLight1337 <tlleungac@connect.ust.hk>

[Model] Support multiple images for qwen-vl (vllm-project#8247)

Signed-off-by: Alex-Brooks <Alex.Brooks@ibm.com>
Co-authored-by: Cyrus Leung <cyrus.tl.leung@gmail.com>
Co-authored-by: DarkLight1337 <tlleungac@connect.ust.hk>

[BugFix] lazy init _copy_stream to avoid torch init wrong gpu instance (vllm-project#8403)

[BugFix] Fix Duplicate Assignment in Hermes2ProToolParser (vllm-project#8423)

[Bugfix] Offline mode fix (vllm-project#8376)

Signed-off-by: Joe Runde <Joseph.Runde@ibm.com>

[multi-step] add flashinfer backend (vllm-project#7928)

[Core] Add engine option to return only deltas or final output (vllm-project#7381)

[Bugfix] multi-step + flashinfer: ensure cuda graph compatible  (vllm-project#8427)

[Hotfix][Core][VLM] Disable chunked prefill by default and prefix caching for multimodal models (vllm-project#8425)

[CI/Build] Disable multi-node test for InternVL2 (vllm-project#8428)

[Hotfix][Pixtral] Fix multiple images bugs (vllm-project#8415)

[Bugfix] Fix weight loading issue by rename variable. (vllm-project#8293)

[Misc] Update Pixtral example (vllm-project#8431)

[BugFix] fix group_topk (vllm-project#8430)

[Core] Factor out input preprocessing to a separate class (vllm-project#7329)

[Bugfix] Mapping physical device indices for e2e test utils (vllm-project#8290)

[Bugfix] Bump fastapi and pydantic version (vllm-project#8435)

[CI/Build] Update pixtral tests to use JSON (vllm-project#8436)

[Bugfix] Fix async log stats (vllm-project#8417)

[bugfix] torch profiler bug for single gpu with GPUExecutor (vllm-project#8354)

bump version to v0.6.1.post1 (vllm-project#8440)

[CI/Build] Enable InternVL2 PP test only on single node (vllm-project#8437)

[doc] recommend pip instead of conda (vllm-project#8446)

[Misc] Skip loading extra bias for Qwen2-VL GPTQ-Int8 (vllm-project#8442)

[misc][ci] fix quant test (vllm-project#8449)

[Installation] Gate FastAPI version for Python 3.8 (vllm-project#8456)

[plugin][torch.compile] allow to add custom compile backend (vllm-project#8445)

[CI/Build] Reorganize models tests (vllm-project#7820)

[Doc] Add oneDNN installation to CPU backend documentation (vllm-project#8467)

[HotFix] Fix final output truncation with stop string + streaming (vllm-project#8468)

bump version to v0.6.1.post2 (vllm-project#8473)

[Hardware][intel GPU] bump up ipex version to 2.3 (vllm-project#8365)

Co-authored-by: Yan Ma <yan.ma@intel.com>

[Kernel][Hardware][Amd]Custom paged attention kernel for rocm (vllm-project#8310)

[Model] support minicpm3 (vllm-project#8297)

Co-authored-by: DarkLight1337 <tlleungac@connect.ust.hk>

[torch.compile] fix functionalization (vllm-project#8480)

[torch.compile] add a flag to disable custom op (vllm-project#8488)

[TPU] Implement multi-step scheduling (vllm-project#8489)

[Bugfix][Model] Fix Python 3.8 compatibility in Pixtral model by updating type annotations (vllm-project#8490)

[Bugfix][Kernel] Add `IQ1_M` quantization implementation to GGUF kernel (vllm-project#8357)

[Kernel] Enable 8-bit weights in Fused Marlin MoE (vllm-project#8032)

Co-authored-by: Dipika <dipikasikka1@gmail.com>

[Frontend] Expose revision arg in OpenAI server (vllm-project#8501)

[BugFix] Fix clean shutdown issues (vllm-project#8492)

[Bugfix][Kernel] Fix build for sm_60 in GGUF kernel (vllm-project#8506)

[Kernel] AQ AZP 3/4: Asymmetric quantization kernels (vllm-project#7270)

[doc] update doc on testing and debugging (vllm-project#8514)

[Bugfix] Bind api server port before starting engine (vllm-project#8491)

[perf bench] set timeout to debug hanging (vllm-project#8516)

[misc] small qol fixes for release process (vllm-project#8517)

[Bugfix] Fix 3.12 builds on main (vllm-project#8510)

Signed-off-by: Joe Runde <Joseph.Runde@ibm.com>

[refactor] remove triton based sampler (vllm-project#8524)

[Frontend] Improve Nullable kv Arg Parsing (vllm-project#8525)

Signed-off-by: Alex-Brooks <Alex.Brooks@ibm.com>

[Misc][Bugfix] Disable guided decoding for mistral tokenizer (vllm-project#8521)

[torch.compile] register allreduce operations as custom ops (vllm-project#8526)

[Misc] Limit to ray[adag] 2.35 to avoid backward incompatible change (vllm-project#8509)

Signed-off-by: Rui Qiao <ruisearch42@gmail.com>

[Benchmark] Support sample from HF datasets and image input for benchmark_serving (vllm-project#8495)

[Encoder decoder] Add cuda graph support during decoding for encoder-decoder models (vllm-project#7631)

[Feature][kernel] tensor parallelism with bitsandbytes quantization (vllm-project#8434)

[Model] Add mistral function calling format to all models loaded with "mistral" format (vllm-project#8515)

Co-authored-by: Cyrus Leung <cyrus.tl.leung@gmail.com>

[Misc] Don't dump contents of kvcache tensors on errors (vllm-project#8527)

[Bugfix] Fix TP > 1 for new granite (vllm-project#8544)

Signed-off-by: Joe Runde <Joseph.Runde@ibm.com>

[doc] improve installation doc (vllm-project#8550)

Co-authored-by: Andy Dai <76841985+Imss27@users.noreply.github.com>

[CI/Build] Excluding kernels/test_gguf.py from ROCm (vllm-project#8520)

[Kernel] Change interface to Mamba causal_conv1d_update for continuous batching (vllm-project#8012)

[CI/Build] fix Dockerfile.cpu on podman (vllm-project#8540)

[Misc] Add argument to disable FastAPI docs (vllm-project#8554)

[CI/Build] Avoid CUDA initialization (vllm-project#8534)

[CI/Build] Update Ruff version (vllm-project#8469)

Signed-off-by: Aaron Pham <contact@aarnphm.xyz>
Co-authored-by: Cyrus Leung <cyrus.tl.leung@gmail.com>

[Core][Bugfix][Perf] Introduce `MQLLMEngine` to avoid `asyncio` OH (vllm-project#8157)

Co-authored-by: Nick Hill <nickhill@us.ibm.com>
Co-authored-by: rshaw@neuralmagic.com <rshaw@neuralmagic.com>
Co-authored-by: Robert Shaw <114415538+robertgshaw2-neuralmagic@users.noreply.github.com>
Co-authored-by: Simon Mo <simon.mo@hey.com>

[Core] *Prompt* logprobs support in Multi-step (vllm-project#8199)

[Core] zmq: bind only to 127.0.0.1 for local-only usage (vllm-project#8543)

Signed-off-by: Russell Bryant <rbryant@redhat.com>

[Model] Support Solar Model (vllm-project#8386)

Co-authored-by: Michael Goin <michael@neuralmagic.com>

[AMD][ROCm]Quantization methods on ROCm; Fix _scaled_mm call (vllm-project#8380)

Co-authored-by: Alexei-V-Ivanov-AMD <156011006+Alexei-V-Ivanov-AMD@users.noreply.github.com>
Co-authored-by: Michael Goin <michael@neuralmagic.com>

[Kernel] Change interface to Mamba selective_state_update for continuous batching (vllm-project#8039)

[BugFix] Nonzero exit code if MQLLMEngine startup fails (vllm-project#8572)

[Bugfix] add `dead_error` property to engine client (vllm-project#8574)

Signed-off-by: Joe Runde <Joseph.Runde@ibm.com>

[Kernel] Remove marlin moe templating on thread_m_blocks (vllm-project#8573)

Co-authored-by: lwilkinson@neuralmagic.com

[Bugfix] [Encoder-Decoder] Bugfix for encoder specific metadata construction during decode of encoder-decoder models.  (vllm-project#8545)

Revert "[Misc][Bugfix] Disable guided decoding for mistral tokenizer" (vllm-project#8593)

[Bugfix] fixing sonnet benchmark bug in benchmark_serving.py (vllm-project#8616)

[MISC] remove engine_use_ray in benchmark_throughput.py (vllm-project#8615)

[Frontend] Use MQLLMEngine for embeddings models too (vllm-project#8584)

[Kernel][Amd] Add fp8 kv cache support for rocm custom paged attention (vllm-project#8577)

[Core] simplify logits resort in _apply_top_k_top_p (vllm-project#8619)

[Doc] Add documentation for GGUF quantization (vllm-project#8618)

Create SECURITY.md (vllm-project#8642)

[CI/Build] Re-enabling Entrypoints tests on ROCm, excluding ones that fail (vllm-project#8551)

[Misc] guard against change in cuda library name (vllm-project#8609)

[Bugfix] Fix Phi3.5 mini and MoE LoRA inference (vllm-project#8571)

[bugfix] [AMD] add multi-step advance_step to ROCmFlashAttentionMetadata (vllm-project#8474)

[Core] Support Lora lineage and base model metadata management (vllm-project#6315)

[Model] Add OLMoE (vllm-project#7922)

[CI/Build] Removing entrypoints/openai/test_embedding.py test from ROCm build (vllm-project#8670)

[Bugfix] Validate SamplingParam n is an int (vllm-project#8548)

[Misc] Show AMD GPU topology in `collect_env.py` (vllm-project#8649)

[Bugfix] Config got an unexpected keyword argument 'engine' (vllm-project#8556)

[Bugfix][Core] Fix tekken edge case for mistral tokenizer (vllm-project#8640)

[Doc] neuron documentation update (vllm-project#8671)

Signed-off-by: omrishiv <327609+omrishiv@users.noreply.github.com>

[Hardware][AWS] update neuron to 2.20 (vllm-project#8676)

Signed-off-by: omrishiv <327609+omrishiv@users.noreply.github.com>

[Bugfix] Fix incorrect llava next feature size calculation (vllm-project#8496)

[Core] Rename `PromptInputs` and `inputs`(vllm-project#8673)

[MISC] add support custom_op check (vllm-project#8557)

Co-authored-by: youkaichao <youkaichao@126.com>

[Core] Factor out common code in `SequenceData` and `Sequence` (vllm-project#8675)

[beam search] add output for manually checking the correctness (vllm-project#8684)

[Kernel] Build flash-attn from source (vllm-project#8245)

[VLM] Use `SequenceData.from_token_counts` to create dummy data (vllm-project#8687)

[Doc] Fix typo in AMD installation guide (vllm-project#8689)

[Kernel][Triton][AMD] Remove tl.atomic_add from awq_gemm_kernel, 2-5x speedup MI300, minor improvement for MI250 (vllm-project#8646)

[dbrx] refactor dbrx experts to extend FusedMoe class (vllm-project#8518)

[Kernel][Bugfix] Delete some more useless code in marlin_moe_ops.cu (vllm-project#8643)

[Bugfix] Refactor composite weight loading logic (vllm-project#8656)

[ci][build] fix vllm-flash-attn (vllm-project#8699)

[Model] Refactor BLIP/BLIP-2 to support composite model loading (vllm-project#8407)

[Misc] Use NamedTuple in Multi-image example (vllm-project#8705)

Signed-off-by: Alex-Brooks <Alex.Brooks@ibm.com>

[MISC] rename CudaMemoryProfiler to DeviceMemoryProfiler (vllm-project#8703)

[Model][VLM] Add LLaVA-Onevision model support (vllm-project#8486)

Co-authored-by: litianjian <litianjian@bytedance.com>
Co-authored-by: Cyrus Leung <cyrus.tl.leung@gmail.com>
Co-authored-by: Roger Wang <ywang@roblox.com>
Co-authored-by: DarkLight1337 <tlleungac@connect.ust.hk>

[SpecDec][Misc] Cleanup, remove bonus token logic. (vllm-project#8701)

[build] enable existing pytorch (for GH200, aarch64, nightly) (vllm-project#8713)

[misc] upgrade mistral-common (vllm-project#8715)

[Bugfix] Avoid some bogus messages RE CUTLASS's revision when building (vllm-project#8702)

[Bugfix] Fix CPU CMake build (vllm-project#8723)

Co-authored-by: Yuan <yuan.zhou@intel.com>

[Bugfix] fix docker build for xpu (vllm-project#8652)

[Core][Frontend] Support Passing Multimodal Processor Kwargs (vllm-project#8657)

Signed-off-by: Alex-Brooks <Alex.Brooks@ibm.com>

[Hardware][CPU] Refactor CPU model runner (vllm-project#8729)

[Bugfix][CPU] fix missing input intermediate_tensors in the cpu_model_runner (vllm-project#8733)

[Model] Support pp for qwen2-vl (vllm-project#8696)

[VLM] Fix paligemma, fuyu and persimmon with transformers 4.45 : use config.text_config.vocab_size (vllm-project#8707)

[CI/Build] use setuptools-scm to set __version__ (vllm-project#4738)

Co-authored-by: youkaichao <youkaichao@126.com>

[Kernel] (2/N) Machete - Integrate into CompressedTensorsWNA16 and GPTQMarlin (vllm-project#7701)

Co-authored-by: mgoin <michael@neuralmagic.com>
Co-authored-by: Divakar Verma <137818590+divakar-amd@users.noreply.github.com>
Co-authored-by: Tyler Michael Smith <tyler@neuralmagic.com>

[Kernel][LoRA]  Add assertion for punica sgmv kernels (vllm-project#7585)

[Core] Allow IPv6 in VLLM_HOST_IP with zmq (vllm-project#8575)

Signed-off-by: Russell Bryant <rbryant@redhat.com>

Fix typical acceptance sampler with correct recovered token ids (vllm-project#8562)

Add output streaming support to multi-step + async while ensuring RequestOutput obj reuse (vllm-project#8335)

[Hardware][AMD] ROCm6.2 upgrade (vllm-project#8674)

Fix tests in test_scheduler.py that fail with BlockManager V2 (vllm-project#8728)

re-implement beam search on top of vllm core (vllm-project#8726)

Co-authored-by: Brendan Wong <bjwpokemon@gmail.com>

Revert "[Core] Rename `PromptInputs` to `PromptType`, and `inputs` to `prompt`" (vllm-project#8750)

[MISC] Skip dumping inputs when unpicklable (vllm-project#8744)

[Core][Model] Support loading weights by ID within models (vllm-project#7931)

[Model] Expose Phi3v num_crops as a mm_processor_kwarg (vllm-project#8658)

Signed-off-by: Alex-Brooks <Alex.Brooks@ibm.com>
Co-authored-by: Cyrus Leung <cyrus.tl.leung@gmail.com>
Co-authored-by: DarkLight1337 <tlleungac@connect.ust.hk>

[Bugfix] Fix potentially unsafe custom allreduce synchronization (vllm-project#8558)

[Kernel] Split Marlin MoE kernels into multiple files (vllm-project#8661)

Co-authored-by: mgoin <michael@neuralmagic.com>

[Frontend] Batch inference for llm.chat() API  (vllm-project#8648)

Co-authored-by: Cyrus Leung <cyrus.tl.leung@gmail.com>
Co-authored-by: Cyrus Leung <tlleungac@connect.ust.hk>
Co-authored-by: Roger Wang <ywang@roblox.com>
Co-authored-by: Roger Wang <136131678+ywang96@users.noreply.github.com>

[Bugfix] Fix torch dynamo fixes caused by `replace_parameters` (vllm-project#8748)

[CI/Build] fix setuptools-scm usage (vllm-project#8771)

[misc] soft drop beam search (vllm-project#8763)

[[Misc]Upgrade bitsandbytes to the latest version 0.44.0 (vllm-project#8768)

[Core][Bugfix] Support prompt_logprobs returned with speculative decoding (vllm-project#8047)

Signed-off-by: Travis Johnson <tsjohnso@us.ibm.com>

[Core] Adding Priority Scheduling (vllm-project#5958)

[Bugfix] Use heartbeats instead of health checks (vllm-project#8583)

Fix test_schedule_swapped_simple in test_scheduler.py (vllm-project#8780)

[Bugfix][Kernel] Implement acquire/release polyfill for Pascal (vllm-project#8776)

Fix tests in test_chunked_prefill_scheduler which fail with BlockManager V2 (vllm-project#8752)

[BugFix] Propagate 'trust_remote_code' setting in internvl and minicpmv (vllm-project#8250)

[Hardware][CPU] Enable mrope and support Qwen2-VL on CPU backend (vllm-project#8770)

[Bugfix] load fc bias from config for eagle (vllm-project#8790)

[Frontend] OpenAI server: propagate usage accounting to FastAPI middleware layer (vllm-project#8672)

[Bugfix] Ray 2.9.x doesn't expose available_resources_per_node (vllm-project#8767)

Signed-off-by: darthhexx <darthhexx@gmail.com>

[Misc] Fix minor typo in scheduler (vllm-project#8765)

[CI/Build][Bugfix][Doc][ROCm] CI fix and doc update after ROCm 6.2 upgrade (vllm-project#8777)

[Kernel] Fullgraph and opcheck tests (vllm-project#8479)

[[Misc]] Add extra deps for openai server image (vllm-project#8792)

[VLM][Bugfix] internvl with num_scheduler_steps > 1 (vllm-project#8614)

rename PromptInputs and inputs with backward compatibility (vllm-project#8760)

[Frontend] MQLLMEngine supports profiling. (vllm-project#8761)

[Misc] Support FP8 MoE for compressed-tensors (vllm-project#8588)

Revert "rename PromptInputs and inputs with backward compatibility (vllm-project#8760) (vllm-project#8810)

[Model] Add support for the multi-modal Llama 3.2 model (vllm-project#8811)

Co-authored-by: simon-mo <xmo@berkeley.edu>
Co-authored-by: Chang Su <chang.s.su@oracle.com>
Co-authored-by: Simon Mo <simon.mo@hey.com>
Co-authored-by: Roger Wang <136131678+ywang96@users.noreply.github.com>
Co-authored-by: Roger Wang <ywang@roblox.com>

[Doc] Update doc for Transformers 4.45 (vllm-project#8817)

[Misc] Support quantization of MllamaForCausalLM (vllm-project#8822)

[Misc] Update config loading for Qwen2-VL and remove Granite (vllm-project#8837)

[Build/CI] Upgrade to gcc 10 in the base build Docker image (vllm-project#8814)

[Docs] Add README to the build docker image (vllm-project#8825)

[CI/Build] Fix missing ci dependencies (vllm-project#8834)

[misc][installation] build from source without compilation (vllm-project#8818)

[ci] Soft fail Entrypoints, Samplers, LoRA, Decoder-only VLM (vllm-project#8872)

Signed-off-by: kevin <kevin@anyscale.com>

[Bugfix] Include encoder prompts len to non-stream api usage response (vllm-project#8861)

[Misc] Change dummy profiling and BOS fallback warns to log once (vllm-project#8820)

[Bugfix] Fix print_warning_once's line info (vllm-project#8867)

fix validation: Only set tool_choice `auto` if at least one tool is provided (vllm-project#8568)

[Bugfix] Fixup advance_step.cu warning (vllm-project#8815)

[BugFix] Fix test breakages from transformers 4.45 upgrade (vllm-project#8829)

[Installation] Allow lower versions of FastAPI to maintain Ray 2.9 compatibility (vllm-project#8764)

[Feature] Add support for Llama 3.1 and 3.2 tool use (vllm-project#8343)

Signed-off-by: Max de Bayser <mbayser@br.ibm.com>

[Core] rename`PromptInputs` and `inputs` (vllm-project#8876)

[misc] fix collect env (vllm-project#8894)

[MISC] Fix invalid escape sequence '\' (vllm-project#8830)

Signed-off-by: Peter Pan <Peter.Pan@daocloud.io>

[Bugfix][VLM] Fix Fuyu batching inference with `max_num_seqs>1` (vllm-project#8892)

[TPU] Update pallas.py to support trillium (vllm-project#8871)

[torch.compile] use empty tensor instead of None for profiling (vllm-project#8875)

[Kernel] AQ AZP 4/4: Integrate asymmetric quantization to linear method (vllm-project#7271)

[Bugfix] fix for deepseek w4a16 (vllm-project#8906)

Co-authored-by: mgoin <michael@neuralmagic.com>

[Core] Multi-Step + Single Step Prefills via Chunked Prefill code path (vllm-project#8378)

Co-authored-by: Varun Sundar Rabindranath <varun@neuralmagic.com>

[misc][distributed] add VLLM_SKIP_P2P_CHECK flag (vllm-project#8911)

[Core] Priority-based scheduling in async engine (vllm-project#8850)

[misc] fix wheel name (vllm-project#8919)

[Bugfix][Intel] Fix XPU Dockerfile Build (vllm-project#7824)

Signed-off-by: tylertitsworth <tyler.titsworth@intel.com>
Co-authored-by: youkaichao <youkaichao@126.com>

[Misc] Remove vLLM patch of `BaichuanTokenizer` (vllm-project#8921)

[Bugfix] Fix code for downloading models from modelscope (vllm-project#8443)

[Bugfix] Fix PP for Multi-Step (vllm-project#8887)

[CI/Build] Update models tests & examples (vllm-project#8874)

Co-authored-by: Roger Wang <ywang@roblox.com>

[Frontend] Make beam search emulator temperature modifiable (vllm-project#8928)

Co-authored-by: Eduard Balzin <nfunctor@yahoo.fr>

[Bugfix] Support testing prefill throughput with benchmark_serving.py --hf-output-len 1 (vllm-project#8891)

[doc] organize installation doc and expose per-commit docker (vllm-project#8931)

[Core] Improve choice of Python multiprocessing method (vllm-project#8823)

Signed-off-by: Russell Bryant <rbryant@redhat.com>
Co-authored-by: youkaichao <youkaichao@126.com>

[Bugfix] Block manager v2 with preemption and lookahead slots (vllm-project#8824)

[Bugfix] Fix Marlin MoE act order when is_k_full == False (vllm-project#8741)

Co-authored-by: Tyler Michael Smith <tyler@neuralmagic.com>

[CI/Build] Add test decorator for minimum GPU memory (vllm-project#8925)

[Build/CI] Set FETCHCONTENT_BASE_DIR to one location for better caching (vllm-project#8930)

[Model] Support Qwen2.5-Math-RM-72B (vllm-project#8896)

[Model][LoRA]LoRA support added for MiniCPMV2.5 (vllm-project#7199)

[BugFix] Fix seeded random sampling with encoder-decoder models (vllm-project#8870)

Co-authored-by: Roger Wang <ywang@roblox.com>

[Misc] Fix typo in BlockSpaceManagerV1 (vllm-project#8944)

[Frontend] Added support for HF's new `continue_final_message` parameter (vllm-project#8942)

[Kernel][Model] Varlen prefill + Prefill chunking support for mamba kernels and Jamba model (vllm-project#8533)
Alvant pushed a commit to compressa-ai/vllm that referenced this pull request Oct 26, 2024
KuntaiDu pushed a commit to KuntaiDu/vllm that referenced this pull request Nov 20, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants