Skip to content

Pull requests: quic/efficient-transformers

New pull request New
64 Open 1,177 Closed
64 Open 1,177 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(qwen3_5_moe): restore original chunk-attn solver to recover accuracy
#1260 opened Aug 10, 2026 by prabeesh-1 • Draft
1
feat(0810): Add continuous batching support to batch-folded decode attn implementation 1.23 Release 1.23 Features enhancement New feature or request
#1259 opened Aug 10, 2026 by vbaddi Contributor • Draft
@kdulla @vbaddi
1
tests: add dtype, golden output comparison and metrics to audio, embedding and diffuser nightly pipeline tests
#1258 opened Aug 10, 2026 by quic-vishali Contributor • Draft
Deepseek v4 oc
#1257 opened Aug 10, 2026 by ochougul Contributor • Draft
Consolidate text_generation examples into a single flag-driven basic_inference.py
#1256 opened Aug 10, 2026 by quic-rishinr Contributor • Draft
chore: update container setup and development environment
#1253 opened Aug 9, 2026 by talnufai Loading…
2
Gemma 4 blocking support
#1251 opened Aug 8, 2026 by tv-karthikeya Contributor Loading…
1
Gemma4 Moe and Prefill changes
#1249 opened Aug 7, 2026 by tchawada Contributor Loading…
feat(weight-free): weight-free ONNX export for causalLM ready for review
#1248 opened Aug 7, 2026 by quic-amitraj Contributor Loading…
Added subfunction fix for gemma 4 with CB
#1245 opened Aug 6, 2026 by abhishek-singh591 Contributor Loading…
1
Adding Multi Resolution Changes for Gemma4 Model
#1244 opened Aug 6, 2026 by tchawada Contributor Loading…
1
Fix: derive dummy_seq_len from the prefill_seq_len kwarg
#1242 opened Aug 5, 2026 by prabeesh-1 Loading…
2
feat(0805): Add semantic ONNX names for w_sub exports 1.23 Release 1.23 Features ci_ready enhancement New feature or request
#1241 opened Aug 5, 2026 by vbaddi Contributor Loading…
@vbaddi
2
Blocking in linear attention
#1240 opened Aug 5, 2026 by mohiso22 Contributor • Draft
Unit tests for disaggregated prefill/decode serving with efficient KV handoff (#1150)
#1239 opened Aug 4, 2026 by quic-abhamidi • Draft
Updating MOE in Mixtral
#1238 opened Aug 4, 2026 by mohiso22 Contributor Loading…
Enable weightfree export with dynamo for Kimi K2.5
#1237 opened Aug 4, 2026 by quic-mamta Contributor • Draft
@quic-mamta
1
Fix for subfn, rope issue in Grok1
#1236 opened Aug 4, 2026 by tv-karthikeya Contributor Loading…
2
Add cohere_asr support - validated-hw on 4 QAIC AI100 devices
#1233 opened Aug 3, 2026 by mabusaa • Draft
Add pwrite-based single-file ONNX export + fix GPT-OSS dynamo export
#1232 opened Aug 3, 2026 by smedhe Contributor • Draft
GPT-OSS: Hoist Layer-Invariant Attention Mask Preparation Outside Decoder Subfunctions
#1231 opened Aug 2, 2026 by abhishek-singh591 Contributor • Draft
2
feat: add w_sub ONNX semantic weight + node rename transform
#1230 opened Jul 31, 2026 by vishwasdivakar • Draft
7
Head parallel and batch fold blocking for Qwen3-VL MOE ci_ready
#1229 opened Jul 31, 2026 by kdulla Contributor Loading…
9
Fix for gemma4 sub-func and continous batching
#1224 opened Jul 30, 2026 by quic-hemagnih Contributor • Draft
feat(ci): add performance regression tracking stage
#1218 opened Jul 27, 2026 by quic-amitraj Contributor • Draft
ProTip! Exclude everything labeled bug with -label:bug.