WuNein

cuda extension problem

5

I test to install in nvidia docker, the build ninja includes incorrent sm_id like `-gencode arch=compute_52,code=sm_52` ``` # Install kernels python setup_cuda.py install ``` ``` cuda_post_cflags = -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_CONVERSIONS__ -D__CUDA_NO_BFLOAT16_CONVERSIONS__...

Decode problem of 13B

2

Run in nvidia pytorch docker 23.05, with rtx a6000 48G ``` from transformers import LlamaForCausalLM, LlamaTokenizer import torch ckpt = '/root/hdd/WizardLM-13B-V1.0' device = torch.device('cuda') # model = LlamaForCausalLM.from_pretrained(ckpt, device_map='auto', load_in_8bit...

GoCV DockerHub image problem

1

docker image from dockerhub I see docker image from https://hub.docker.com/r/gocv/opencv/tags. I intended to use image [4.7.0-gpu-cuda-11.2.2](https://hub.docker.com/layers/gocv/opencv/4.7.0-gpu-cuda-11.2.2/images/sha256-9ea0535c2b4fcff1589f12123da1c0807691b22d1a127eefe01bd4203fee6c28?context=explore) to cross build my onnx image detection problem used in windows. ```bash sudo docker...

1. Fixed input from other than camera(rmtp, for example)``` cv2.imwrite(os.path.join('dst', args.character, args.output_dir, f'{frame_count:04d}.jpeg')) ``` has no `output_frame`. 2. add fp16 My RTX4090 made 50% decrease in computing. Require `pip install...

https://github.com/kohya-ss/sd-scripts/blob/71e2c91330a9d866ec05cdd10584bbb962896a99/networks/lora.py#L966 Here enables lora, i need to enable and disable lora. So I can train with dpo. Dpo required to compute with and without lora results.