-
Notifications
You must be signed in to change notification settings - Fork 347
Pull requests: ModelTC/LightLLM
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
test/benchmark: turn bench_agents.py into an adaptive stress test
#1436
opened Aug 5, 2026 by
sufubao
Collaborator
Loading…
fix(pd): balance prefill nodes on decayed recent load to avoid cache-affinity starvation
#1433
opened Aug 5, 2026 by
sufubao
Collaborator
Loading…
fix(api): harden OpenAI streaming/non-streaming error handling
#1426
opened Aug 4, 2026 by
sufubao
Collaborator
Loading…
feat(pd): balance heterogeneous prefill load by input tokens
#1425
opened Aug 4, 2026 by
sufubao
Collaborator
Loading…
perf(pd): reduce qwen3next prefill linear-att state memory
#1421
opened Aug 4, 2026 by
sufubao
Collaborator
Loading…
fix(multinode): use JSON instead of pickle for node-IP exchange to prevent RCE
#1419
opened Aug 2, 2026 by
Anai-Guo
Loading…
fix(qwen3.5): restore BF16 attention gate semantics for RL consistency
#1415
opened Jul 30, 2026 by
shihaobai
Collaborator
Loading…
fix: clean up shared memory on process exit
#1414
opened Jul 30, 2026 by
sufubao
Collaborator
Loading…
fix: quarantine PD pages until transfer quiesces
#1406
opened Jul 27, 2026 by
sufubao
Collaborator
Loading…
feat: fuse post-attention all-reduce and RMSNorm
#1405
opened Jul 27, 2026 by
sufubao
Collaborator
Loading…
fix: fix bugs in the ep v2 and mega moe for sm100
#1392
opened Jul 15, 2026 by
blueswhen
Collaborator
Loading…
feat: gc.freeze hot processes and disable gc during cudagraph capture
#1382
opened Jul 5, 2026 by
sufubao
Collaborator
Loading…
Support GLM-5.2 (glm_moe_dsa, DeepSeek-V3.2-style DSA MoE)
#1370
opened Jun 25, 2026 by
sufubao
Collaborator
Loading…
fix(sampling_params): validate token-id input types via a shared helper
#1365
opened Jun 19, 2026 by
SuperMarioYL
Contributor
Loading…
fix(moe): avoid cross-warp stale read in ep_scatter prefix sum
#1362
opened Jun 16, 2026 by
Anai-Guo
Loading…
test(sampling_params): repair broken test collection and add verify() coverage
#1350
opened Jun 13, 2026 by
SuperMarioYL
Contributor
Loading…
feat: add multi-platform support with ascend and maca
#1335
opened Jun 8, 2026 by
zhangts20
Collaborator
Loading…
feat: update disk cache params and benchmark_multiturn.py
#1333
opened Jun 8, 2026 by
blueswhen
Collaborator
Loading…
Previous Next
ProTip!
Updated in the last three days: updated:>2026-08-03.