TAO 5.0 google colab

garazraz · August 25, 2023, 1:42pm

When are TAO 5 models (such as DINO) expected to be supported via the google colab notebooks? I cant find any documentation explaining how to run it in colab. Could it be because the pip package nvidia-tao-tf1 is still at 4.0.0.657.dev0? Or does everything have to be reworked to make it possible?

Morganh · August 25, 2023, 5:33pm

Hi,
We did not support all the notebooks in Colab. We just keep few notebooks for Colab.
The available notebooks are mentioned in

garazraz · August 28, 2023, 7:04am

Since the new models are not supported in colab notebooks then as you said, is it possible to get information on what docker image is actually running underneath the “tao” command? is it any of these three?

github.com

NVIDIA/tao_tensorflow2_backend/blob/main/docker/Dockerfile

ARG FROM_IMAGE_NAME=nvcr.io/nvidia/tensorflow:23.02-tf2-py3
FROM ${FROM_IMAGE_NAME}

RUN apt-get update
RUN apt-get install ffmpeg libsm6 libxext6  -y
# upgrade pip
RUN pip install --upgrade pip

# Copy requirement and install
WORKDIR /workspace/
COPY docker/requirements.txt /workspace/
RUN pip install nvidia-pyindex Cython==0.29.36
RUN pip install -r requirements.txt && \
    rm -rf requirements.txt

# install public version of Nvidia QAT Toolkit
RUN git clone https://github.com/NVIDIA/TensorRT.git
COPY docker/tensorflow_quantization/setup.py /workspace/TensorRT/tools/tensorflow-quantization
WORKDIR /workspace/TensorRT/tools/tensorflow-quantization
RUN python -m pip install .

This file has been truncated. show original

github.com

NVIDIA/tao_pytorch_backend/blob/main/docker/Dockerfile

ARG BASE_IMAGE=nvcr.io/nvidia/pytorch:23.02-py3
FROM ${BASE_IMAGE}

# Ensure apt-get won't prompt for selecting options
ENV DEBIAN_FRONTEND=noninteractive
COPY docker/requirements-apt.txt requirements-apt.txt
RUN apt-get upgrade && apt-get update && \
  xargs apt-get install -y < requirements-apt.txt && \
  rm requirements-apt.txt && \
  rm -rf /var/lib/apt/lists/*

# uninstall stuff from base container
RUN pip uninstall -y sacrebleu torchtext
RUN pip install parametrized ninja
# Installing custom packages in /opt.
WORKDIR /opt

# Clone and checkout TensorRT OSS
# Moving TensorRT to 8.5 branch.
ENV TRT_TAG "release/8.5"

This file has been truncated. show original

github.com

NVIDIA/tao_tensorflow1_backend/blob/main/docker/Dockerfile

ARG FROM_IMAGE_NAME=nvcr.io/nvidia/tensorflow:23.02-tf1-py3
FROM ${FROM_IMAGE_NAME}

RUN apt-get update
COPY docker/requirements-apt.txt requirements-apt.txt
RUN apt-get update \
    && DEBIAN_FRONTEND=noninteractive apt-get install -y curl apt-transport-https software-properties-common \
    && (echo "deb [arch=amd64] https://download.docker.com/linux/ubuntu bionic stable" | tee /etc/apt/sources.list.d/docker.list) \
    && (curl https://download.docker.com/linux/ubuntu/gpg | apt-key add -) \
    && apt-get update \
    && DEBIAN_FRONTEND=noninteractive xargs apt-get install -y < requirements-apt.txt \
    && rm requirements-apt.txt
# upgrade pip
RUN pip install --upgrade pip

# Copy requirement and install
WORKDIR /workspace/

ENV TF_XLA_FLAGS="--tf_xla_enable_lazy_compilation=false tf_xla_async_io_level=0"

This file has been truncated. show original

The reason is, if i know what environment its using and what the ways are to build it, then i could create my own script from scratch that would enable me to train on colab or colab like notebook environments.

Morganh · August 28, 2023, 7:10am

There is no update from you for a period, assuming this is not an issue anymore. Hence we are closing this topic. If need further support, please open a new one. Thanks

You can find the info in Working With the Containers - NVIDIA Docs.

For example,
Dino network locates at docker nvcr.io/nvidia/tao/tao-toolkit:5.0.0-pyt

More info can be found in TAO Toolkit | NVIDIA NGC as well.

system · September 11, 2023, 7:55am

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.

Topic		Replies	Views
TAO Toolkit Google colab Error TAO Toolkit	7	812	February 22, 2024
Running tao toolkit in google colab TAO Toolkit tao	14	1927	July 9, 2023
A few questions regarding TAO 5.0 TAO Toolkit	9	624	August 3, 2023
Running TAO Toolkit on Google Colab TAO Toolkit	8	640	February 22, 2024
Can't run the provided TAO toolkit sample code TAO Toolkit	20	1994	July 31, 2023
TAO 5.0 Documentation and getting started TAO Toolkit	7	632	August 14, 2023
TAO on Google Colab TAO Toolkit	7	925	February 22, 2024
Updating tao toolkit TAO Toolkit	12	1004	July 11, 2022
Failed to import TensorRT package, exporting TLT to a TensorRT engine will not be available TAO Toolkit	3	468	August 31, 2023
Nvidia TAO tookit on AWS vm TAO Toolkit	18	283	July 9, 2024

TAO 5.0 google colab

Related topics