NVIDIA
NVIDIA
NeMo Framework Megatron Backend
Container
NVIDIA
NVIDIA
NeMo Framework Megatron Backend

NVIDIA NeMo™ framework Megatron backend supports pre-training, post-training, and reinforcement learning of LLMs and multi-modal generative AI models with state-of-the-art data processing, model training techniques, and flexible deployment options.

LayerLabelCreated
e72e211b1a25bb7893ab38911df6d78ca8bd8dea00f55080c956166cc0a7961cRUN
/bin/sh
12/06/2023 1:42 AM UTC
bd6c706a5f809595c34c1d43a129e465125237170dcfcaa166e590f80f0886a3ARG
NVIDIA_BUILD_REF
12/06/2023 1:42 AM UTC
e9e85cd385432cba9c366d40d4aaec9ed54a6acd0c049f3c6d884f237ca5288aLABEL
com.nvidia.build.id=71422337
12/06/2023 1:42 AM UTC
6255eba1fe6b3140db91a478dec47cfd6d5ea05939cc35a9c195dcfb53b523cbENV
NVIDIA_BUILD_ID=71422337
12/06/2023 1:42 AM UTC
77b9c025342c3095005547b0dcdbe9a2b85d8e55ed099e798c09e7ca5a873496ARG
NVIDIA_BUILD_ID
12/06/2023 1:42 AM UTC
d83f9809c4639067bcab84d08979628911df5cd866562a828f71dd3074c3e87eRUN
chmod -R a+w /workspace
12/06/2023 1:42 AM UTC
964fabe19dd1a5f1916d05fec522d4d50b292720a4a9430ea382ddd5b0a8da8fWORKDIR
/workspace
12/06/2023 1:42 AM UTC
9e75a99132d504746a19c969a251d8ada20836d4d80cc6670a9683922575c728ENV
GDRCOPY_VERSION=2.3 HPCX_VERSION=2.14 MOFED_VERSION=5.4-rdmacore36.0 OPENUCX_VERSION=1.14.0 OPENMPI_VERSION=4.1.4 RDMACORE_VERSION=36.0
12/06/2023 1:42 AM UTC
23786198cacdf7c1c9b559a9be29cf82daa4519c792c29c24bb62c272f6b2c02RUN
cd /opt/hpcx &&
  rm -rf * &&
  curl -f --proto '=https' https://content.mellanox.com/hpc/hpc-x/v2.14/hpcx-v2.14-gcc-inbox-ubuntu22.04-cuda12-gdrcopy2-nccl2.17-x86_64.tbz | tar jxf - --strip-components=1
12/06/2023 1:42 AM UTC
d90e7f3c29150f3970c671fcd186b398df549af9e07a3c4aa3ef296b64983ec6RUN
echo "    UserKnownHostsFile /dev/null" >> /etc/ssh/ssh_config &&
  sed -i 's/#\(StrictModes \).*/\1no/g' /etc/ssh/sshd_config &&
  sed -i 's/#   StrictHostKeyChecking ask/    StrictHostKeyChecking no/' /etc/ssh/ssh_config &&
  mkdir -p /var/run/sshd
12/06/2023 1:41 AM UTC
...