-
Notifications
You must be signed in to change notification settings - Fork 118
Pull requests: ndif-team/nnsight
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
modeling: rebuild PEFT envoy paths on environment swap
#728
opened Sep 10, 2026 by
koriyoshi2041
Loading…
envoy: len, iteration and indexing answer for the same entries
#725
opened Sep 8, 2026 by
JadenFiotto-Kaufman
Member
Loading…
batching: scope token-flattened rows, and warn on writes that can't be scoped
#722
opened Sep 8, 2026 by
JadenFiotto-Kaufman
Member
Loading…
vllm: gather a fused projection into the layout tp=1 has
#719
opened Sep 8, 2026 by
JadenFiotto-Kaufman
Member
•
Draft
cache: resolve modules= against the tree instead of caching nothing
#716
opened Sep 8, 2026 by
JadenFiotto-Kaufman
Member
Loading…
test(backward): assert a batched invoke's gradient is its own rows
#715
opened Sep 7, 2026 by
Hotragn
Contributor
Loading…
TransformersModel.load_adapter: attach a PEFT adapter after construction
#714
opened Sep 5, 2026 by
JadenFiotto-Kaufman
Member
Loading…
feat(atp-star): add tested component-localization building blocks
#705
opened Aug 29, 2026 by
janmenjayap
Loading…
Fix remote execution for encoder models
#700
opened Aug 26, 2026 by
elliottower
Contributor
Loading…
Gather tensor-parallel sharded parameters on read in a trace
#677
opened Jun 20, 2026 by
khaiwang
Contributor
Loading…
feat(intervention): isolated GPU-worker execution backend for model.trace()
#676
opened Jun 20, 2026 by
khaiwang
Contributor
Loading…
feat: run:ai gpu_direct model loading with quantization-aware device maps
#635
opened Mar 20, 2026 by
khaiwang
Contributor
Loading…
6 tasks done
ProTip!
What’s not been updated in a month: updated:<2026-08-15.