Skip to content

Pull requests: sgl-project/SpecForge

New pull request New
96 Open 591 Closed
96 Open 591 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(export): honor explicit vocabulary mappings in HF exports
#872 opened Sep 9, 2026 by StevenWang-CY Contributor Loading…
1
[refactor] dedupe attention-mask helpers in llama3_eagle
#871 opened Sep 9, 2026 by FrankLeeeee Collaborator Loading…
6 tasks
feat: support Ling-3.0 (Bailing V3) draft training
#870 opened Sep 9, 2026 by bighead-liat Loading…
3 of 4 tasks
4
[refactor] add hardware-specific installation instruction
#861 opened Sep 8, 2026 by FrankLeeeee Collaborator • Draft
2 of 6 tasks
5
perf(disagg): default to pinned receives and automatic RDMA GPU publication
#858 opened Sep 8, 2026 by maocheng23 Collaborator Loading…
Normalise dflash-variant config
#852 opened Sep 8, 2026 by cih9088 Contributor Loading…
1 of 6 tasks
[Bugfix] Fix Issue 828
#848 opened Sep 7, 2026 by jiapingW Collaborator Loading…
6 tasks
1
3
feat(dflash): clarify W&B metrics and diagnose prefix acceptance failures
#844 opened Sep 7, 2026 by maocheng23 Collaborator Loading…
3 of 4 tasks
feat(musa): add MUSA device support
#843 opened Sep 7, 2026 by wenqf11 Contributor Loading…
6 tasks
2
dflash: read selector schedule attributes through the DDP wrapper (NO_SHARD)
#842 opened Sep 6, 2026 by cpuchip Loading…
2 of 6 tasks
1
feat(spec-capture): publish captured tensors from device memory (GPU put)
#841 opened Sep 6, 2026 by maocheng23 Collaborator Loading…
feat(mooncake): pooled pinned or device receive buffers for disaggregated feature reads
#840 opened Sep 6, 2026 by maocheng23 Collaborator Loading…
1
feat(core): add label-ids CE variant(stacked on #826)
#837 opened Sep 5, 2026 by curnane-lab Collaborator Loading…
4 of 6 tasks
feat(dflash): KDA draft attention with context scan (block-local policy, scan policy, batched scan)
#836 opened Sep 4, 2026 by maocheng23 Collaborator • Draft
9 tasks done
feat(dflash): report sampled serving-walk accepted length
#835 opened Sep 4, 2026 by lvhua898-cpu Loading…
1
feat(dspark): add fused TV loss (stacked on #826)
#829 opened Sep 3, 2026 by curnane-lab Collaborator Loading…
6 tasks done
feat(core): add fused loss kernels
#826 opened Sep 3, 2026 by curnane-lab Collaborator Loading…
4 of 6 tasks
feat(dflash): add full-vocabulary AngelSpec LK loss
#823 opened Sep 3, 2026 by maocheng23 Collaborator • Draft
1
DeepSeek-V4 MoE FFN for DFlash-family drafters + DSV4-Flash ablation arm
#812 opened Sep 2, 2026 by sherlockwu Collaborator • Draft
MoE FFN skeleton for DFlash-family drafters (contracts, config, seams)
#811 opened Sep 2, 2026 by sherlockwu Collaborator • Draft
Drafter eval tooling: AIME25 benchmark, jsonl accept probe, template override
#801 opened Sep 1, 2026 by sherlockwu Collaborator • Draft
DSparkV4DraftModel: official DeepSeek-V4 DSpark drafter architecture
#800 opened Sep 1, 2026 by sherlockwu Collaborator • Draft
Sparse MoE FFN for DFlash-family drafters
#799 opened Sep 1, 2026 by sherlockwu Collaborator • Draft
add FastMTP-style per-step weighted loss
#786 opened Aug 28, 2026 by curnane-lab Collaborator • Draft
4 of 5 tasks
feat(dspark): add multimodal (image+text) training support
#785 opened Aug 28, 2026 by curnane-lab Collaborator Loading…
3 of 6 tasks
ProTip! Find all pull requests that aren't related to any open issues with -linked:issue.