Skip to content

Pull requests: XPU-Forces/mojo_opset

New pull request New
49 Open 353 Closed
49 Open 353 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

lightning_indexer acc fix
#449 opened Aug 7, 2026 by Liwansi Contributor Loading…
[NPU] flex_attention: enable preload + lazy-load for large tensors
#438 opened Jul 30, 2026 by OuYangxin12 Loading…
1
feat(swa_train): enable preload by default
#422 opened Jul 23, 2026 by ZS-YANG Loading…
1
[KMCompiler] [ttx/npu] add DeepSeek-V4 operator moe_gating_topk_hash
#418 opened Jul 21, 2026 by YangLong114514 Contributor Loading…
5
[KMCompiler] [ttx/npu] add DeepSeek-V4 operator rms_norm_dynamic_quant
#416 opened Jul 20, 2026 by YangLong114514 Contributor Loading…
10
[XCS][ttx/npu] optimize LightningIndexer operater performance
#406 opened Jul 15, 2026 by Quamaly Loading…
5
lazy-load: _sdpa_single_block_bwd_dkdv & _sdpa_single_block_bwd_dq
#405 opened Jul 15, 2026 by LXW-github Loading…
[XCS][ttx] optimize ResidualAddLayerNorm operater performance
#404 opened Jul 15, 2026 by Quamaly Loading…
3
optimize triton ops to improve performance
#398 opened Jul 13, 2026 by MichelleWu351 Contributor Loading…
9
[ttx/npu] Optimize paged attention decode GQA with cube and split-KV
#395 opened Jul 13, 2026 by lyujheng Contributor Loading…
7
causal_conv1d_update_kernel_bdt_fwd optimized version
#388 opened Jul 7, 2026 by Liwansi Contributor Loading…
4
[npu/ttx] feat: add DeepSeekV4 hc_pre operator on Triton
#383 opened Jul 2, 2026 by lyujheng Contributor Loading…
10
[npu/ttx] feat: add DeepSeekV4 hc_post operator on Triton
#382 opened Jul 2, 2026 by lyujheng Contributor Loading…
2
feat: add DeepSeek-V4 inference on Ascend with TP/EP/DP/CP parallel support
#375 opened Jun 26, 2026 by Song-begin Collaborator Loading…
8
[ilu/ttx] opt swa_paged_prefill: remove boundary_check in nomask path…
#373 opened Jun 22, 2026 by sky-fun Collaborator Loading…
3
[KMCompiler][ttx]opt for rope
#368 opened Jun 15, 2026 by YangLong114514 Contributor Loading…
5
[ilu/ttx] optimize int8-KV paged prefill: dequant + reuse bf16 FA2
#361 opened Jun 12, 2026 by AbeFei Collaborator Loading…
6
Profiler tensor shape to MojoOperator
#359 opened Jun 12, 2026 by NASA1473 Collaborator Loading…
5
MojoFusedNormRoPESageQuantStore: fused RoPE + KV-Quant with Key per-token Quant operator
#358 opened Jun 12, 2026 by NASA1473 Collaborator Loading…
8
[ilu/ttx] opt swa: decode add fast path when windows cover the sequence;quant prefill use dequant + bf16 swa paged prefill
#354 opened Jun 11, 2026 by sky-fun Collaborator Loading…
6
Support router replay
#353 opened Jun 11, 2026 by zhanzy178 Collaborator Loading…
5
add deepep and test, modify test_moe_quant and test_attention_quant
#349 opened Jun 5, 2026 by qiushi13 Collaborator Loading…
5
fix layernorm_fwd_kernel
#347 opened Jun 4, 2026 by Todobe Loading…
WIP [ilu/ttx] optimize quant_moe_export
#343 opened Jun 3, 2026 by songliucsu Collaborator • Draft
@madengfei
8
[uc] add activation, norm, pos emb, quant, sdpa operators for uc backend.
#342 opened Jun 3, 2026 by shengw-bd Collaborator Loading…
12
ProTip! Exclude everything labeled bug with -label:bug.