Lianmin Zheng dc0705a504 Simplify prepare_extend_after_decode (#6987)

2025-06-09 16:39:21 -07:00

..

configs

[FEAT] Add transformers backend support (#5929)

2025-06-03 21:05:29 -07:00

connector

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

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

constrained

Correctly abort the failed grammar requests & Improve the handling of abort (#6803)

2025-06-01 19:00:07 -07:00

disaggregation

feat: add direct routing strategy to DP worker (#6884)

2025-06-09 11:44:05 -07:00

distributed

support 1 shot allreduce in 1-node and 2-node using mscclpp (#6277)

2025-06-04 22:11:24 -07:00

entrypoints

feat: add direct routing strategy to DP worker (#6884)

2025-06-09 11:44:05 -07:00

function_call

feat(tool call): Enhance Llama32Detector for improved JSON parsing in non-stream (#6784)

2025-05-30 17:05:17 -07:00

layers

Fallback to lower triton version for unfound fused moe configs (#7013)

2025-06-09 15:41:03 -07:00

lora

Eliminate stream sync to speed up LoRA batch init (#6960)

2025-06-09 00:22:45 -07:00

managers

Simplify prepare_extend_after_decode (#6987)

2025-06-09 16:39:21 -07:00

mem_cache

[VLM] Support chunk prefill for VLM (#6355)

2025-05-22 20:32:41 -07:00

metrics

Correctly abort the failed grammar requests & Improve the handling of abort (#6803)

2025-06-01 19:00:07 -07:00

model_executor

Simplify prepare_extend_after_decode (#6987)

2025-06-09 16:39:21 -07:00

model_loader

[FEAT] Add transformers backend support (#5929)

2025-06-03 21:05:29 -07:00

models

support qwen3 emebedding (#6990)

2025-06-09 01:32:49 -07:00

openai_api

bugfix(OAI): Fix image_data processing for jinja chat templates (#6877)

2025-06-05 13:37:01 -07:00

sampling

Slightly improve the sampler to skip unnecessary steps (#6956)

2025-06-08 03:18:54 -07:00

speculative

Simplify prepare_extend_after_decode (#6987)

2025-06-09 16:39:21 -07:00

_custom_ops.py

support 1 shot allreduce in 1-node and 2-node using mscclpp (#6277)

2025-06-04 22:11:24 -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

Add typo checker in pre-commit (#6179)

2025-05-11 12:55:00 +08:00

conversation.py

[Minor] Always append newline after image token when parsing chat message (#6797)

2025-05-31 20:50:33 -07:00

custom_op.py

[fix] recover auto-dispatch for rmsnorm and rope (#6745)

2025-06-03 21:44:20 -07:00

debug_utils.py

Add simple utility to dump tensors for debugging (#6815)

2025-06-02 11:15:31 -07:00

hf_transformers_utils.py

[New Model] Devstral support (#6547)

2025-05-26 19:27:48 -07:00

mm_utils.py

refactor: minor refactors regarding multimodal processing (#6187)

2025-05-17 22:53:20 -07:00

model_parallel.py

Make torch TP composable with torchao (#2436)

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

operations_strategy.py

Overlap two kernels in DeepSeek with communication (#6711)

2025-05-28 15:53:51 -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

reasoning_parser.py

Add typo checker in pre-commit (#6179)

2025-05-11 12:55:00 +08:00

server_args.py

Simplify prepare_extend_after_decode (#6987)

2025-06-09 16:39:21 -07:00

torch_memory_saver_adapter.py

[CI] Fix ci tests (#5769)

2025-04-27 07:18:10 -07:00

two_batch_overlap.py

Remove unnecessary kernels of num_token_non_padded (#6965)

2025-06-08 05:09:17 -07:00

utils.py

Slightly improve the sampler to skip unnecessary steps (#6956)

2025-06-08 03:18:54 -07:00

warmup.py

Support penalty in overlap mode; return logprob with chunked prefill; improve benchmark scripts (#3988)

2025-03-03 00:12:04 -08:00