Skip to content

Pull requests: radixark/miles

New pull request New
487 Open 1,405 Closed
487 Open 1,405 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

fix: use paged MLA prefill on B200
#2047 opened Aug 2, 2026 by guapisolo Collaborator Loading…
1
Support DFlash rollouts without logprobs
#2046 opened Aug 2, 2026 by zianglih Contributor • Draft
1
Support native DeepSeek V4 MXFP4 checkpoint conversion
#2044 opened Aug 1, 2026 by zianglih Contributor • Draft
1
Compact NVFP4 BF16 MoE exclusion metadata
#2043 opened Aug 1, 2026 by zianglih Contributor Loading…
2
fix(megatron): honor args.use_distributed_optimizer on bridge LoRA DDP
#2042 opened Aug 1, 2026 by excepshenal Loading…
1
[AMD] Enable THD packed-sequence training for DeepSeek-V4-Flash, with context parallelism
#2038 opened Jul 31, 2026 by JessicaJiang-123 Contributor Loading…
1 task
1
Preserve custom loss metadata through Megatron training
#2037 opened Jul 31, 2026 by chinmay-simile • Draft
1
[fp8] Gate the ue8m0 weight quantizer on the backend scale format
#2034 opened Jul 31, 2026 by JessicaJiang-123 Contributor Loading…
1
scripts: let the GLM-5.2 744B LoRA launcher shard across nodes
#2033 opened Jul 31, 2026 by Shi-Dong Contributor Loading…
1
feat(fsdp): add hybrid sharding
#2031 opened Jul 31, 2026 by Rockdu Contributor Loading…
3 of 6 tasks
4
add staleness control in data buffer
#2030 opened Jul 31, 2026 by yueming-yuan Collaborator Loading…
1
test: run MXFP8 and NVFP4 colocate E2E on 8 GPUs
#2029 opened Jul 31, 2026 by zianglih Contributor • Draft
1
dashboard: advisory v2 — run-health alarms before config tuning
#2027 opened Jul 31, 2026 by yueming-yuan Collaborator Loading…
1
dashboard: scrape engines directly by default
#2026 opened Jul 31, 2026 by yueming-yuan Collaborator Loading…
1
dashboard: read open phase markers regardless of age
#2024 opened Jul 31, 2026 by yueming-yuan Collaborator Loading…
1
dashboard: dp-aware engine metrics
#2023 opened Jul 31, 2026 by yueming-yuan Collaborator Loading…
1
dashboard: cache full-window engine series queries
#2022 opened Jul 31, 2026 by yueming-yuan Collaborator Loading…
1
feat: expose Claude skills to agents
#2021 opened Jul 31, 2026 by guapisolo Collaborator Loading…
1
[WIP] Add experimental NVFP4 fake-QAT RL recipe
#2020 opened Jul 31, 2026 by zianglih Contributor • Draft
1
[lora] refactor - native LoRA
#2017 opened Jul 31, 2026 by yushengsu-thu Collaborator • 2/2 Loading…
10
ci(docker): move builds to docker-build nodes; resolve upstream refs up front
#2016 opened Jul 31, 2026 by guapisolo Collaborator Loading…
2
scripts, examples: stop forcing the deprecated Miles router in launchers
#2015 opened Jul 30, 2026 by Shi-Dong Contributor Loading…
3 of 4 tasks
1
[docs] Retrigger docs deploy
#2011 opened Jul 30, 2026 by Zhichenzzz Contributor Loading…
1
feat: add Tinker-compatible training API
#2007 opened Jul 30, 2026 by yushengsu-thu Collaborator Loading…
2
Launch inference engines through the inference engine worker spec
#2006 opened Jul 30, 2026 by fzyzcjy Collaborator Loading…
1
ProTip! Mix and match filters to narrow down what you’re looking for.