bdf

MaskedCol2imForward exist the bug of array out of bounds

3

Thanks for reporting the unexpected results and we appreciate it a lot. **Checklist** 1. I have searched related issues but cannot get the expected help. 2. I have read the...

ERROR: Could not find a version that satisfies the requirement mdformat-openmmlab (from versions: none)

2

The version of python noted in README is 3.6+. My python version is python3.6. After executing the following commands:pip install -U pre-commit and pre-commit install, I excute git commit, some...

Thanks for your contribution and we appreciate it a lot. The following instructions would make your pull request more healthy and more easily get feedback. If you do not understand...

terminate called after throwing an instance of 'std::runtime_error' what(): Duration is out of dual 32-bit range

10

Hello,can anyone help me,when i run the code:rosrun lsd_slam_core dataset_slam _files:=/home/lab/rosbuild_ws/LSD_room/images _hz:=0 _calib:=/home/lab/rosbuild_ws/LSD_room/cameraCalibration.cfg rosrun lsd_slam_core dataset_slam _files:=/home/lab/rosbuild_ws/LSD_room/images _hz:=0 _calib:=/home/lab/rosbuild_ws/LSD_room/cameraCalibration.cfg Reading Calibration from file /home/lab/rosbuild_ws/LSD_room/cameraCalibration.cfg ... found! found ATAN camera model,...

[Refactor] Adapt mlu code to cntoolkit3.0.1

1

Thanks for your contribution and we appreciate it a lot. The following instructions would make your pull request more healthy and more easily get feedback. If you do not understand...

@foralliance Hi！I think we should speak English because your questions may help the people in other countries.

3

@foralliance Hi！I think we should speak English because your questions may help the people in other countries. 1. No. When you train the model with large input image size (e.g.,...

FA-V3 deterministic usage

1

Hi, @tridao I'm glad you used the semaphore solution to solve the problem of backward deterministic computing (which used for dq or dk and dv while GQA). I found that...

NCCL timeout while different ranks execute DeepEP and NCCL communications in different order

5

Reproduced code ```python if rank % 2 == 0: with torch.cuda.stream(test_alt_stream): dist.all_gather_into_tensor(all_topk_idx, topk_idx, group=group) buffer.low_latency_dispatch(current_x, topk_idx, num_tokens, num_experts, use_fp8=dispatch_use_fp8, round_scale=round_scale, use_ue8m0=use_ue8m0, cumulative_local_expert_recv_stats=cumulative_local_expert_recv_stats, async_finish=not return_recv_hook, return_recv_hook=return_recv_hook) if rank % 2 !=...