Skip to content

Pull requests: torch-spyre/spyre-inference

New pull request New
61 Open 464 Closed
61 Open 464 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

E2e model quality upstream
#853 opened Sep 10, 2026 by yannicks1 Collaborator • Draft
5 tasks
2
perf(attn): widen the batched decode gather to the core count
#852 opened Sep 10, 2026 by tdoublep Collaborator • Draft
4 of 5 tasks
1
chore(deps): bump torch-spyre to 447d19e against latest Spyre RPMs
#851 opened Sep 10, 2026 by joerunde Collaborator Loading…
6 of 11 tasks
1
Add page-major KV layout to page attn microbench
#850 opened Sep 10, 2026 by cyang49 Member Loading…
3 of 5 tasks
8
ci: aggregate sharded test results into one report
#848 opened Sep 10, 2026 by joerunde Collaborator Loading…
5 tasks
2
fix(tests): tag every tier a test belongs to, not just the invoked one
#847 opened Sep 10, 2026 by ashokponkumar Contributor Loading…
1
tests: Enable a subset of GSM8k upstream vllm tests
#846 opened Sep 10, 2026 by joerunde Collaborator Loading…
5 tasks
7
[Feature] Enable FP32 RMSNorm
#845 opened Sep 10, 2026 by bohnstingl Collaborator Loading…
4 of 5 tasks
1
✨ Add Gemma4 multimodal vision
#843 opened Sep 10, 2026 by gkumbhat Collaborator • Draft
5 tasks
2
[Cleanup] Small edits for Gemma4-MoE
#842 opened Sep 10, 2026 by bohnstingl Collaborator Loading…
4 of 5 tasks
1
perf(fp8): Remove redundant clone calls from FP8 Linear Kernel
#841 opened Sep 10, 2026 by R3hankhan123 Contributor Loading…
3 of 5 tasks
1
1
feat(models): serve BharatGen Param checkpoints as native Llama
#839 opened Sep 10, 2026 by amaljoe Loading…
4 of 5 tasks
1
perf(attn): gate batched decode on KV utilisation ratio
#838 opened Sep 10, 2026 by jvlunteren Collaborator Loading…
4 of 5 tasks
1
Enable FP8 Parallel LM Head
#833 opened Sep 10, 2026 by R3hankhan123 Contributor Loading…
4 of 5 tasks
1
1
perf(attn): extend batched-decode fast path to mixed prefill+decode batches
#832 opened Sep 10, 2026 by jvlunteren Collaborator Loading…
4 of 5 tasks
1
[Feature] Grouping KV-cache pages
#829 opened Sep 9, 2026 by bohnstingl Collaborator • Draft
5 tasks done
1
[Cleanup] Small fixes to the multi-modal models
#827 opened Sep 9, 2026 by bohnstingl Collaborator Loading…
1
Add online benchmarking script
#826 opened Sep 9, 2026 by sducouedic Contributor Loading…
2
Remove temporary workarounds
#823 opened Sep 9, 2026 by nikheal2 Contributor Loading…
3 of 5 tasks
1
DO NOT MERGE: debugging compile cache
#822 opened Sep 9, 2026 by sducouedic Contributor • Draft
1
test: DO NOT MERGE — legacy-image control (fp8 baseline on main d97de1f)
#815 opened Sep 9, 2026 by HarikrishnanBalagopal Contributor Loading…
1
perf(custom_ops): compute FP8 per-token scale via quantscalepertokenfp8
#814 opened Sep 9, 2026 by prashantgupta24 Collaborator Loading…
3 of 5 tasks
1
ci(matrix): enable and persist SPYRE_KERNEL_CACHE across CI runs
#812 opened Sep 8, 2026 by prashantgupta24 Collaborator • Draft
2 of 3 tasks
5
refactor(attn): let the bucketer own the batched decode buckets
#810 opened Sep 8, 2026 by sducouedic Contributor Loading…
4 of 5 tasks
6
Adding different lx layout to micro-benchmarks
#807 opened Sep 8, 2026 by bringlein Collaborator Loading…
4 of 5 tasks
4
ProTip! Find all pull requests that aren't related to any open issues with -linked:issue.