AlexHT Hung

I solve it by replace loss functions. In losses.py, ``` # comment out following two lines #weights = (weights - tf.reduce_min(weights)) / (tf.reduce_max(weights) - tf.reduce_min(weights)) + 1. #mask = mask...

try version 5.3 `pip install PyYAML==5.3`

I follow the [wiki](https://github.com/openvinotoolkit/openvino_contrib/wiki/How-to-build-ARM-CPU-plugin#approach-1-build-opencv-openvino-and-the-plugin-using-pre-configured-dockerfile-cross-compiling-the-preferred-way) and it works: ```sh git clone --recurse-submodules --single-branch --branch=releases/2022/1 https://github.com/openvinotoolkit/openvino_contrib.git cd openvino_contrib/modules/arm_plugin docker image build -t arm-plugin -f dockerfiles/Dockerfile.RPi64_focal . mkdir build docker container run --rm...

> @brmarkus - Thanks for the response. I checked the link and looks like those plugins are for Raspberry Pi4 but I am using Raspberry pi 3. > > Do...

Paddle2onnx轉PPOCRv3，出現類似的問題，Reshape應該有支援才對 (使用rknntoolkit 1.7.1) ``` W Not match tensor Reshape_p2o.Reshape.23:out0 E Try match layer: Reshape_p2o.Reshape.23:out0 failed! The OP of this layer may not yet support, please check document ! E Catch...

Adding '-fPIC' to this line can solve the problem. https://github.com/microsoft/DeepSpeed/blob/3418b869f0ec784b90a35c1253320780b03645a9/op_builder/builder.py#L670

I've finished an initial version, but the results are not as promising as expected. The original combination of `accelerate` and `gradcache` could achieve nearly 250W GPU utilization, however, after switching...

I'm using DeepSpeed, and offload device is set to none. I modified `examples/causal_language_modeling/peft_lora_clm_accelerate_ds_zero3_offload.py` to load xP3 dataset, and load model into fp16: ```python # creating model model = AutoModelForCausalLM.from_pretrained(model_name_or_path, torch_dtype=torch.float16)...

Thanks for your information, @pacman100 ! We have tried megatron-deepspeed and network seems fast enough. I will check DataLoader later.

Should I worry about this warning? `[WARNING] [stage3.py:1939:step] 13 pytorch allocator cache flushes since last step. this happens when there is high memory pressure and is detrimental to performance. if...