YiXR 0418b9d4ea [Optimization] Update estimated_num_new_pages logic in TokenToKVPoolAllocator (#8794) ...

Signed-off-by: Xingrui Yi <yixingrui@linux.alibaba.com>
Co-authored-by: Xingrui Yi <yixingrui@linux.alibaba.com>

2025-08-10 16:01:51 -07:00

..

configs

Support glm4.1v and glm4.5v (#8798)

2025-08-09 00:59:13 -07:00

connector

feat(remote_model): support variable remote backend for model loader (#3964)

2025-03-14 00:40:44 -07:00

constrained

bugfix: Fix XGrammar backend to use model's EOS tokens for constrained generation (#8422)

2025-07-28 10:01:02 +08:00

debug_utils

Minor tool for comparison of benchmark results (#7974)

2025-07-27 00:27:50 -07:00

disaggregation

[Feat] QWen-1M context support[2/2]: Update block sparse attention backend (#5949)

2025-08-06 23:49:36 -07:00

distributed

[Feature] Optimize DeepSeek's DeepEP on Ascend NPU (#8355)

2025-08-09 01:35:00 -07:00

entrypoints

Fix CI (#9012)

2025-08-09 13:33:42 -07:00

eplb

[2/3] Optimize Slime Update Weights: Avoid GPU-to-CPU Device Sync when update expert weights (#8753)

2025-08-05 22:09:52 -07:00

function_call

Support glm4.1v and glm4.5v (#8798)

2025-08-09 00:59:13 -07:00

layers

[Fix] Fix flashinfer cpu <-> gpu synchronization (#8340)

2025-08-10 03:11:40 +00:00

lora

Improve LoRA Perf by Deprecating FlashInfer and Eliminating Redundant Tensor Ops (#8940)

2025-08-10 01:04:45 -07:00

managers

[Fix] Fix flashinfer cpu <-> gpu synchronization (#8340)

2025-08-10 03:11:40 +00:00

mem_cache

[Optimization] Update estimated_num_new_pages logic in TokenToKVPoolAllocator (#8794)

2025-08-10 16:01:51 -07:00

metrics

feat: add production metric for retracted requests due to insufficient kvcache (#7030)

2025-07-17 11:59:05 -07:00

model_executor

[Optimization] Update estimated_num_new_pages logic in TokenToKVPoolAllocator (#8794)

2025-08-10 16:01:51 -07:00

model_loader

Support OCP MXFP4 quantization on AMD GPUs (#8255)

2025-08-04 18:14:52 -07:00

models

Improve LoRA Perf by Deprecating FlashInfer and Eliminating Redundant Tensor Ops (#8940)

2025-08-10 01:04:45 -07:00

multimodal

fix: fix obsolete qwen-audio processor arg (#9003)

2025-08-09 13:18:36 -07:00

sampling

fix: prevent crashes due to logit bias dimension mismatch (#7685)

2025-07-23 15:30:55 -07:00

speculative

refine aiter_backend for mtp (#7279)

2025-08-08 11:06:02 -07:00

weight_sync

Add tensor.detach() back to update weight util (#8691)

2025-08-02 00:41:05 -07:00

_custom_ops.py

[Feature] Integrate quick allreduce and select the best allreduce implementation (#6619)

2025-07-24 20:48:42 -07:00

aio_rwlock.py

A better aio rwlock that guarantees the order (#2547)

2024-12-22 15:44:32 -08:00

code_completion_parser.py

feat(oai refactor): Replace openai_api with entrypoints/openai (#7351)

2025-06-21 13:21:06 -07:00

constants.py

Multi-Stage Awake: Support Resume and Pause KV Cache and Weights separately (#7099)

2025-06-19 00:56:37 -07:00

conversation.py

[feat] Add detail in image_data (#8596)

2025-08-05 14:01:38 +08:00

custom_op.py

[1/N] MoE Refactor: refactor select_experts (#7966)

2025-07-19 00:51:15 -07:00

hf_transformers_utils.py

[Feat] QWen-1M context support[2/2]: Update block sparse attention backend (#5949)

2025-08-06 23:49:36 -07:00

jinja_template_utils.py

Support glm4.1v and glm4.5v (#8798)

2025-08-09 00:59:13 -07:00

model_parallel.py

Make torch TP composable with torchao (#2436)

2024-12-11 04:21:42 -08:00

operations_strategy.py

[5/N] MoE Refactor: Update MoE parallelism arguments (#8658)

2025-08-01 01:20:03 -07:00

operations.py

Support overlapping two batches (#4068)

2025-05-24 17:39:07 -07:00

patch_torch.py

Fix torch.compile cacheing (#5259)

2025-04-10 18:08:45 -07:00

poll_based_barrier.py

Support colocating requests (#7973)

2025-07-28 22:51:49 -07:00

reasoning_parser.py

Enables force reasoning based on chat template for Qwen3-Thinking (#8369)

2025-08-06 20:02:47 -07:00

server_args.py

Clean up server_args.py to have a dedicated function for model specific adjustments (#8983)

2025-08-08 19:56:50 -07:00

torch_memory_saver_adapter.py

Multi-Stage Awake: Support Resume and Pause KV Cache and Weights separately (#7099)

2025-06-19 00:56:37 -07:00

two_batch_overlap.py

Enable TBO on ROCm (#8329)

2025-08-09 01:59:55 -07:00

utils.py

Support glm4.1v and glm4.5v (#8798)

2025-08-09 00:59:13 -07:00

warmup.py

Improve streaming, log_level, memory report, weight loading, and benchmark script (#7632)

2025-06-29 23:16:19 -07:00