sglang

configs

[bug] fix errors related to context length in SD (#9388 )

2025-08-21 10:32:34 +08:00

connector

fix: fix MLA for ShardedModelLoader/RemoteModelLoader (#6287 )

2025-08-28 16:10:09 -07:00

constrained

Update grok.py and tiktoken tokenizer (#9532 )

2025-08-23 05:40:18 -07:00

debug_utils

Minor tool for comparison of benchmark results (#7974 )

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

disaggregation

Fix memory leak when aborting decode request in PD-Disagg (#9817 )

2025-08-30 14:36:03 +08:00

distributed

Overlapped weight offload (#8034 )

2025-08-23 02:06:46 -07:00

entrypoints

chore: upgrade flashinfer 0.3.0rc1 (#9793 )

2025-08-29 16:24:17 -07:00

eplb

feature(eplb): add min-rebalancing-utilization-threshold for eplb (#8345 )

2025-08-30 11:24:29 +08:00

function_call

Support DeepSeek-V3.1 tool call (#9446 )

2025-08-26 20:22:19 -07:00

layers

Fix TRTLLM MLA Cuda KV Blocks Causing accuracy drop (#9675 )

2025-08-29 17:16:10 -07:00

lora

Add target module validation for init adapters (#9429 )

2025-08-24 20:24:50 -07:00

managers

Fix memory leak when aborting decode request in PD-Disagg (#9817 )

2025-08-30 14:36:03 +08:00

mem_cache

feat(hicache-3fs): 3FS-Store Backup Optimizations For MLA Model. (#9692 )

2025-08-29 10:48:51 -07:00

metrics

[PD] Improve disaggregation metrics output: update the metrics to keep reflecting real stats (#7317 )

2025-08-24 23:16:43 -07:00

model_executor

Support speculative decoding in hybrid attention backend (#9573 )

2025-08-28 01:11:42 -07:00

model_loader

fix: fix MLA for ShardedModelLoader/RemoteModelLoader (#6287 )

2025-08-28 16:10:09 -07:00

models

fix: dsv3 lite q_lora_rank none (#9815 )

2025-08-29 23:24:14 -07:00

multimodal

fix: InternS1 don't recognize image, updates image token for InternVL processor (#9381 )

2025-08-20 16:43:03 -07:00

sampling

[feat] Reduce GPU memory overhead by using weakref (#9673 )

2025-08-28 01:09:06 -07:00

speculative

feat: add original logprobs to response (#8375 )

2025-08-29 11:43:57 -07:00

tokenizer

Add reasoning_effort param in TiktokenTokenizer.apply_chat_template (#9630 )

2025-08-26 00:44:20 -07:00

weight_sync

Support Flatten Tensor Update Weights to speed up MOE Update Weights by 20% (#8079 )

2025-08-10 16:08:59 -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

bench_utils.py

Fuse writing KV buffer into rope kernel (part 1: sgl-kernel) (#9077 )

2025-08-12 01:46:40 -07:00

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: (chat-template matching) enhance multimodal model detection with config.json (#9597 )

2025-08-26 17:55:40 -07:00

custom_op.py

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

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

harmony_parser.py

Fix Harmony reasoning parser for and auto-separation for gpt-oss models (#9190 )

2025-08-25 15:26:26 -07:00

hf_transformers_utils.py

feat(draft_model): support draft_model for RemoteModelLoader (#6407 )

2025-08-28 16:09:52 -07:00

host_shared_memory.py

Overlapped weight offload (#8034 )

2025-08-23 02:06:46 -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

offloader.py

Offload tensors by sharding on GPU (#9536 )

2025-08-25 00:02:49 -07:00

operations_strategy.py

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

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

operations.py

Add fp4 quantize before all-gather for Flashinfer cutlass MoE DP (max throughput) (#7667 )

2025-08-15 22:08:11 -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

Fix Harmony reasoning parser for and auto-separation for gpt-oss models (#9190 )

2025-08-25 15:26:26 -07:00

server_args.py

feature(eplb): add min-rebalancing-utilization-threshold for eplb (#8345 )

2025-08-30 11:24:29 +08:00

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

[6/N] MoE Refactor: Cleanup MoE-related configs (#8849 )

2025-08-14 21:14:53 -07:00

utils.py

Fix get_ip when no external network (#9700 )

2025-08-27 10:28:52 -07:00

warmup.py

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

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