1459 Commits

Author SHA1 Message Date
sunyanguomt
24a2bb4f7b
[MUSA] mmcv support musa, split pr 3 (#3258)
* mmcv support musa, split pr 3

* fix focal_loss probelem

* fix when musa is not available, input have is_musa jugment
2025-03-17 22:02:49 +08:00
sunyanguomt
c7d3326d8a
[MUSA] Support MUSA PR2
* squash commit for support musa pr2

* fix conv2d_gradfix:Conv2d is_cuda_available judgment

* Revert "fix conv2d_gradfix:Conv2d is_cuda_available judgment"

This reverts commit 38faea16c126612bae0897f65cf722144b7cb24e.

* fix conv2d_gradfix:Conv2d comment
2025-03-11 12:28:07 +08:00
Epiphany
c4ea0e6d97
[Fix] fix ci (#3251)
* fix ci

* fix lint
2025-03-03 14:33:41 +08:00
sunyanguomt
99bae9e179
support moore threads MUSA gpu,split PR 1. (#3241)
* mmcv support musa, split pr 1

* bug fix : add spconv

* fix gpu struct and fix weird indent

* fix lint problem

* fix lint problem

* fix clang-format probelem

* fix clang-format

* fix clang-format
2025-02-27 11:51:30 +08:00
Epiphany
23ec934b8f
fix the ci (#3246)
* fix-ci

* fix lint

* remove circleci
2025-02-19 11:41:27 +08:00
huangyuan64
6ba6b7e679
change the npu code for roi align rotated (#3238) 2025-02-18 11:48:26 +08:00
Wang Yixuan
a4a884dd88
hccl fix (#3236) 2025-02-10 12:20:48 +08:00
huangyuan64
53e6b94ca7
fix the npu deform roi pool param name error (#3235) 2025-02-10 12:20:10 +08:00
Wang Yixuan
1c11e0c004
Revert "adapt boxes_overlap_bev & box_iou_rotated (#3162)" (#3232)
This reverts commit 44d871e7f36bb00afc3946ea36cc6dcf03e07e44.
2025-02-10 12:19:54 +08:00
Wang Yixuan
a5cfaefa2e
fix ops npu (#3231)
* fix roi_pool_backward npu

* fix lint in pixel_group
2025-01-21 18:03:06 +08:00
Jing-Yi Li
761d780a24
[NPU] modify internal calls of npu boxes_overlap_bev & box_iou_rotated (#3215) 2024-12-23 12:19:13 +08:00
Bosco-lab
5de993a6f7
添加pixel_group的npu适配 (#3206) 2024-12-23 12:18:41 +08:00
Wang Yixuan
fc85dbdd11
[NPU]a dd roiaware_pool3d_npu (#3213)
* fix lint

* fix lint 1

* fix
2024-12-11 12:39:48 +08:00
Wang Yixuan
edd672835b
[npu] add_npu_border_align (#3210) 2024-12-04 15:05:56 +08:00
Wang Yixuan
7893e66562
add npu adaptation for assign_score_withk op (#3209)
* test

* fix

* lint fix

* lint1
2024-12-04 15:05:23 +08:00
Zac
03ce9208d1
points_in_boxes_all和points_in_boxes_part的mmcv兼容npu判断 (#3189)
* points_in_boxes_all和points_in_boxes_part的mmcv兼容npu判断

* Update points_in_boxes.py

去除判断
2024-11-25 12:41:56 +08:00
huangyuan64
e1aab12f9b
更新knn和three_nn的NPU适配代码 (#3194) 2024-11-18 13:41:44 +08:00
Nikita Shulga
71437a361c
[Enhance] Do not use deprecated .type().scalarType() (#3195)
`at::Tensor::type()` method is deprecated and `tensor.scalar_type()` is a recommended way to replace `tensor.type().scalarType()` invocation
2024-11-04 14:45:49 +08:00
Mashiro
139325726b
[Fix] Fix ci (#3196)
* [Fix] Fix pre-commit-hook for python3.12

* [Fix] clang lint

* use ACTIONS_ALLOW_USE_UNSECURE_NODE_VERSION

* Add todo
2024-11-04 12:00:16 +08:00
Zac
c46684c30f
points_in_boxes_all的mmcv适配 (#3186) 2024-10-24 14:09:11 +08:00
Jing-Yi Li
c39e9ebbc4
adapt npu box_iou_quadri (#3179) 2024-10-21 12:47:44 +08:00
ason-rob
73fa88ad5e
RoiAlignRotatedV2算子调用适配层 (#3177) 2024-09-26 15:03:18 +08:00
Wang Yixuan
bd0c65ed67
npu version adaption (#3176) 2024-09-26 10:57:31 +08:00
Wang Yixuan
3c9d327cc7
This is a combination of 2 commits. (#3170)
version adaptation

lint fix
2024-09-02 16:12:34 +08:00
Jing-Yi Li
44d871e7f3
adapt boxes_overlap_bev & box_iou_rotated (#3162) 2024-09-02 16:12:11 +08:00
Mashiro
3ba02dfd35
Update version info (#3166)
* Update version info

* Update version info

* Update version info
2024-08-17 11:57:27 +08:00
Pr0Wh1teGivee
5916fbd878
fix npu_msda (#3153) 2024-07-24 15:06:14 +08:00
huhongsun
9f43b8c310
add npu roipoint_pool3d_forward (#3147)
* add npu roipoint_pool3d_forward

* nms_rotated npu add new attr is_angle
2024-07-19 01:33:34 +08:00
Pr0Wh1teGivee
44eab261b9
[NPU] npu msda supports aclnn (#3149) 2024-07-17 17:04:21 +08:00
Tsekai Lee
8f23a0b8f2
【Feature】knn/tnn npu added (#3125)
Co-authored-by: lizekai <lizekai3@hisilicon.com>
2024-06-11 17:05:24 +08:00
wujiadi1
b91cfded58
[Feature] add group_points support npu (#3120) 2024-06-04 14:35:39 +08:00
Annarine
210090075e
[Fix] Fix chamfer_distance cuda bug len should div to 4 (#3113) 2024-05-23 21:17:31 +08:00
Annarine
c0c81ce53e
[Fix] Fix ball_query precision bug fp16 to fp32 (#3110) 2024-05-22 11:58:51 +08:00
RRaoyzee
27df2b3b9c
Refactor the implementation of ms_deform_attn for Ascend (#3057) 2024-05-16 17:21:13 +08:00
Mashiro
7f9cf4e0da
[Docs] Update torch2.3 version (#3100)
* Update version

* Fix
2024-05-06 11:49:26 +08:00
Annarine
0d3cdc6ee0
fix fps csrc bug (#3094) 2024-04-25 15:54:29 +08:00
Mashiro
d7acd4b92d
Update version info (#3092) 2024-04-24 22:33:02 +08:00
sherie
6a1646f1a3
Support fps and points_in_box ops for Ascend device (#3085) 2024-04-23 17:31:51 +08:00
Mashiro
3a5fb4b96a
Bump version to 2.2.0 (#3087) v2.2.0 2024-04-16 17:00:23 +08:00
RRaoyzee
780ffed9f3
Add multi_scale_deform_attn_grad op adapter for NPU (#3042) 2024-03-13 15:15:46 +08:00
DaGaiBa
cd05d71254
Add multi_scale_deform_attn op adapter for NPU (#3032) 2024-03-01 15:02:23 +08:00
Chai
d9e10e1184
Fix ut of stack_grouping_points (#3026) 2024-01-29 19:18:01 +08:00
Chai
815f5d75a7
[Feature] Add the support for stack_group_points ops with Ascend (#3023) 2024-01-29 17:05:23 +08:00
sherie
c7c02a7c5b
[Feature] torch_npu support aclnn and add op (#2998) 2024-01-07 17:11:36 +08:00
huaweiZJX
2e44eaeba3
[Feature] Add NPU operator RotatedFeatureAlign (#2994) 2023-11-29 17:24:28 +08:00
shun001
504014847a
[Fix] Fix modulated_deform_conv.py for torch_npu v2.1 (#2983) 2023-11-13 17:17:52 +08:00
CokeDong
b366e923e4
DIOPI adapt codes for npu device (#2973) 2023-10-30 17:14:41 +08:00
shun001
00e92ab632
[Fix] Fix deform_conv.py for torch_npu v2.1 (#2967) 2023-10-24 11:43:19 +08:00
fanqiNO1
51a9943764
[Docs] Add api reference table (#2971) 2023-10-23 14:13:21 +08:00
Zaida Zhou
df2dadb78d
[Docs] Update version info (#2966) 2023-10-18 18:46:27 +08:00