-
Notifications
You must be signed in to change notification settings - Fork 325
Pull requests: sgl-project/SpecForge
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
docs: drop stale DTA references
#822
opened Sep 3, 2026 by
imitater-dou
Contributor
Loading…
2 tasks
feat(mtp): chunk the lm_head and CE objective over token positions
#821
opened Sep 3, 2026 by
curnane-lab
Contributor
Loading…
2 of 6 tasks
docs: clarify MTP needs merge_mtp_to_base (not shared export)
#820
opened Sep 3, 2026 by
imitater-dou
Contributor
Loading…
fix(domino): normalize training loss by the global token count
#818
opened Sep 3, 2026 by
curnane-lab
Contributor
Loading…
2 of 6 tasks
docs: add MTP to offline feature contract in data_preparation
#817
opened Sep 3, 2026 by
imitater-dou
Contributor
Loading…
1 task
docs: add MTP attention backends to customization table
#816
opened Sep 3, 2026 by
imitater-dou
Contributor
Loading…
2 tasks
docs: add MTP to Supported Methods in README
#814
opened Sep 3, 2026 by
imitater-dou
Contributor
Loading…
3 tasks
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
feat: add structured DFlash2 training metrics
#810
opened Sep 2, 2026 by
maocheng23
Collaborator
Loading…
docs: fix --is-performatted typo in prepare_hidden_states Usage
#809
opened Sep 2, 2026 by
imitater-dou
Contributor
Loading…
2 tasks done
docs: qualify configs README for unsupported VLM draft JSONs
#808
opened Sep 2, 2026 by
imitater-dou
Contributor
Loading…
3 tasks
[patch] support applying SGLang patch inside a git repo
#807
opened Sep 2, 2026 by
FrankLeeeee
Collaborator
Loading…
6 tasks
[doc] fixed installation command
#806
opened Sep 2, 2026 by
FrankLeeeee
Collaborator
Loading…
2 of 6 tasks
docs: fix stale --tp flag in regenerate_train_data SGLang example
#805
opened Sep 2, 2026 by
imitater-dou
Contributor
Loading…
docs: fix --tp to --tp-size in specbundle SGLang flags
#804
opened Sep 2, 2026 by
imitater-dou
Contributor
Loading…
2 of 3 tasks
docs: fix stale --tp flag in gates SGLang example
#803
opened Sep 2, 2026 by
imitater-dou
Contributor
Loading…
docs: fix stale --tp flag in bench SGLang examples
#802
opened Sep 2, 2026 by
imitater-dou
Contributor
Loading…
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
[feature] deepseek v4 flash dspark GQA drafter training
#798
opened Sep 1, 2026 by
sherlockwu
Collaborator
Loading…
[fix] small improvements for dataloader, metrics, optimizer robustness, and ops
#797
opened Sep 1, 2026 by
sherlockwu
Collaborator
Loading…
docs: add --no-deps to ROCm install in training.md
#796
opened Sep 1, 2026 by
imitater-dou
Contributor
Loading…
Previous Next
ProTip!
Add no:assignee to see everything that’s not assigned.