CUDA内存不足

[gaizi123]

Loaded cached embeddings from file. Checking if the server is listening on port 8890... Server not ready, waiting 4 seconds... Traceback (most recent call last): File "D:\LivePortrait-Windows-v20240806\app_animals.py", line 47, in gradio_pipeline_animal: GradioPipelineAnimal = GradioPipelineAnimal( File "D:\LivePortrait-Windows-v20240806\src\gradio_pipeline.py", line 508, in init super().init(inference_cfg, crop_cfg) File "D:\LivePortrait-Windows-v20240806\src\live_portrait_pipeline_animal.py", line 42, in init self.cropper: Cropper = Cropper(crop_cfg=crop_cfg, image_type='animal_face', flag_use_half_precision=inference_cfg.flag_use_half_precision) File "D:\LivePortrait-Windows-v20240806\src\utils\cropper.py", line 86, in init self.animal_landmark_runner.warmup() File "D:\LivePortrait-Windows-v20240806\src\utils\animal_landmark_runner.py", line 135, in warmup self.run(img_rgb, 'face', 'face', box_threshold=0.0, IoU_threshold=0.0) File "D:\LivePortrait-Windows-v20240806\src\utils\animal_landmark_runner.py", line 118, in run boxes_filt, keypoints_filt = self.get_unipose_output(image, instance_text_prompt, keypoint_text_prompt, box_threshold, IoU_threshold) File "D:\LivePortrait-Windows-v20240806\src\utils\animal_landmark_runner.py", line 75, in get_unipose_output "object_embeddings_text": ins_text_embeddings.float(), RuntimeError: CUDA error: out of memory CUDA kernel errors might be asynchronously reported at some other API call,so the stacktrace below might be incorrect.

[medalawi]

what is your GPU VRAM? what command are you use to run animal mode ? -type nvidia-smi and paste result here give us a screenshot contain settings and a simple of photo

[zzzweakman]

@gaizi123 Thanks for your attention. You'd better make sure your GPU has at least 8GB of VRAM.