ahqzy

How can I get the intermediate output of a model?

6

Hi I'm debuging some onnx model, I want to compare the result between cpu and gpu., so I need get all intermediate output. I can get them when in cpu...

About mode of Pad

1

Hi, Timvx only supports constant-mode for operator Pad currently， do you have a plane to support reflect-mode and edge-mode?

Matmul cant't get expected values when input is 5d or above

1

Hi, I'm testing matmul in timvx(opencl). It works well when both input is 4d or below. But it works abnormally when input dimension is equal or greater than 5. For...

A bit of confusion about the function kdequant_mm_int32_fp16().

1

### System Info I am studying the function kdequant_mm_int32_fp16() and encountered a confusing issue. I constructed a matrix with a shape of [3, 96] and values ranging from 0 to...

Audio driven推理出来的视频，口型明显不对

3

命令：python -m scripts.audio2vid --config ./configs/prompts/animation_audio2.yaml -W 512 -H 512 -acc 视频： https://github.com/user-attachments/assets/feb9e3f1-42e9-4e97-9ee8-ce7d4426b33b

[BUG]onnx convert failed("KeyError: 'max_sequence_len'")

1

[hugectr2onnx_bst_test.zip](https://github.com/user-attachments/files/18269157/hugectr2onnx_bst_test.zip) when do onnx inference(HugeCTR/test/onnx_converter_test/hugectr2onnx_bst_test.py), it failed as below: ``` # python ./hugectr2onnx_bst_test.py [HUGECTR2ONNX][INFO]: Converting Data layer to ONNX [HUGECTR2ONNX][INFO]: Converting DistributedSlotSparseEmbeddingHash layer to ONNX [HUGECTR2ONNX][INFO]: Converting DistributedSlotSparseEmbeddingHash layer to...

[Question]May I use pytorch train + hps + onnxruntime inference?

2

Hi, If I want to train a model with pytorch, and inference by onnxruntime, may I use hps(redis、rocksdb) to pass parameters？

linux平台编译metaplayer7报错

3

拉取了最新代码，编译demo metaplayer7时报错，信息如下：  看了下编译出来的文件，确实没有libyangh264decoder.a，只有libmetartc7.a libmetartccore7.a

执行export_onnx.py报错

1

1 执行命令： python -m trt.export_onnx --model-root ./ckpts --infer-mode fa --no-enhance --load-key ema --model DiT-XL/2 --image-size 1024 1024 报错信息： (hydit) root@L20:~/HunyuanDiT# python -m trt.export_onnx --model-root ./ckpts --infer-mode fa --no-enhance --load-key ema...