Skip to content

Pull requests: ServiceNow/Fast-LLM

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
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort

Pull requests list

Fix yarn rope factor and mask_token_id in HF conversion
#529 opened Jun 1, 2026 by jlamypoirier Collaborator Loading…
Add fp32_lm_head flag for vLLM precision parity
#526 opened May 27, 2026 by jlamypoirier Collaborator Draft
1 task done
Tool: evaluate layer-wise numerical-error propagation
#525 opened May 26, 2026 by jlamypoirier Collaborator Loading…
1 of 2 tasks
Add docs_per_step for dynamic microbatch accumulation
#520 opened May 19, 2026 by jlamypoirier Collaborator Loading…
1 task done
Canonicalize varlen cu_seqlens_k; share K/V buffer across micro-sequences
#514 opened May 14, 2026 by jlamypoirier Collaborator Loading…
3 tasks done
Allow no bos for Qwen
#473 opened Mar 7, 2026 by shruthan Collaborator Loading…
1 of 25 tasks
[EXTERNAL] Add vLLM Apriel2 model with plugin-based registration
#447 opened Jan 12, 2026 by tscholak Collaborator Loading…
5 of 6 tasks
[WIP] Changes for generate and lm_eval after code refactoring
#438 opened Jan 6, 2026 by bigximik Collaborator Draft
25 tasks
Add IS evaluator
#432 opened Dec 21, 2025 by tscholak Collaborator Draft
[Prototype] Concatenated weights and linear layers
#366 opened Sep 22, 2025 by jlamypoirier Collaborator Draft
ProTip! no:milestone will show everything without a milestone.