Caize Wu

Exec export_onnx_cpu.py error

3

get error when use export_onnx_cpu.py to export wenet pretrain model (20220506_u2pp_conformer_exp) python3 wenet/bin/export_onnx_cpu.py \ --config wenet_mdl/20220506_u2pp_conformer_exp/train.yaml \ --checkpoint wenet_mdl/20220506_u2pp_conformer_exp/final.pt \ --output_dir wenet_mdl/20220506_u2pp_conformer_exp/ \ --chunk_size 16 \ --num_decoding_left_chunks 4 \ --reverse_weight...

always predict "classical" in real world record wav

1

Hi, I just test your model with wav that record by real world mic, and convert to 16k samplerate wav. I record the song in the dataset, and env is...

reverse NPU inviting

2

hi, your project is great! but all hack is in user space, we are now reversing NPU register and driver, to make an opensource AI toolchain, are you interested in...

invite join MaixPy project

1

hi, how are things going with you recently? we have port micropython/openMV IDE, Arduino IDE, and platformIO IDE to k210. I think your repo for k210 is clean and tidy,...

It use ckpt in train_softmax.py, but use pb in run_mobilenet_v1.sh. It case run_mobilenet_v1.sh error ``` if pretrained_model: print('Restoring pretrained model: %s' % pretrained_model) saver_restore.restore(sess, tf.train.latest_checkpoint(pretrained_model)) ```

tangprimer20k program slow and stucked

13

Hi, thank you for support tangprimer20k~ I just tested "tangprimer20k" options, and seems it have some bug, when program flash, it use 23s program 1%... when program sram, it stucked...

RISC-V (TH1520&D1) benchmark and hack for <1GB DDR device

7

Hi, Just test on RISC-V board: 4xC910 2.0G TH1520 LicheePi4A (https://sipeed.com/licheepi4a) with 16GB LPDDR4X. about 6s/token without any instruction acceleration, and it should be '' 15597 -> 'They' sampling parameters:...

use a st-node with this function to burn hairBoot to empty chip.