-
Notifications
You must be signed in to change notification settings - Fork 3.2k
Pull requests: exo-explore/exo
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
Cluster reliability: master election stability + bootstrap peer dial-in + local model symlink fix
#2080
opened May 10, 2026 by
team-wcv
Contributor
Loading…
Drafter abstraction + Gemma 4 MTP + Qwen 3.5/3.6 DFlash + multi-device coupled-drafter speculative decoding
#2079
opened May 10, 2026 by
team-wcv
Contributor
Loading…
Add exo-diagnostics CLI + diagnostic snapshot loop
#2066
opened May 7, 2026 by
team-wcv
Contributor
Loading…
Asymmetric tensor parallelism for uneven-memory clusters
#2064
opened May 7, 2026 by
team-wcv
Contributor
Loading…
Thunderbolt 5 + nested-hub discovery; JACCL init retry; RDMA host selection
#2063
opened May 7, 2026 by
team-wcv
Contributor
Loading…
Stabilize GPT-OSS tool support; echo reasoning config in Responses API
#2062
opened May 7, 2026 by
team-wcv
Contributor
Loading…
fix: handle KeyError and missing text_config in vision loader
#2055
opened May 6, 2026 by
Drifter4242
Contributor
Loading…
fix: set processor_repo to vision weights repo in Kimi model cards
#2054
opened May 6, 2026 by
Drifter4242
Contributor
Loading…
fix(api): eliminate token-drop race in _send_text_generation_with_images
#2044
opened May 3, 2026 by
Drifter4242
Contributor
Loading…
Reconcile worker instance backoff from State
#2028
opened May 3, 2026 by
AlexCheema
Contributor
Loading…
Add WarnExtraModel: log warnings for unknown fields in API request models
#2015
opened May 2, 2026 by
imapotato123
•
Draft
4 tasks done
Unified model catalog across HF, LM Studio, Ollama, llama.cpp + content-based resolution
#2012
opened May 1, 2026 by
AlexCheema
Contributor
•
Draft
10 of 12 tasks
event-sourcing cleanup + snapshot-based catch-up
#2011
opened May 1, 2026 by
AlexCheema
Contributor
•
Draft
8 tasks done
feat(profilers): GPU + link profiling with directional measurements
#2010
opened Apr 30, 2026 by
AlexCheema
Contributor
Loading…
4 of 5 tasks
fix: avoid content leak when generation ends inside thinking block
#2001
opened Apr 28, 2026 by
mlpy0
Contributor
Loading…
perf(deepseek_v4): fuse switch_mlp gate_proj + up_proj into single gather_qmm
#1999
opened Apr 27, 2026 by
adurham
Contributor
Loading…
4 tasks done
fix(deepseek_v4): drop full-attention sharding for MoE-only strategy
#1996
opened Apr 27, 2026 by
adurham
Contributor
Loading…
4 tasks done
feat: peer-to-peer model distribution
#1992
opened Apr 26, 2026 by
adurham
Contributor
Loading…
4 tasks
fix: skip KV cache quantization in single-node BatchGenerator mode
#1990
opened Apr 26, 2026 by
adurham
Contributor
Loading…
Previous Next
ProTip!
no:milestone will show everything without a milestone.