NeMo Framework Megatron Backend

NVIDIA

Container

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.

NVIDIA AI Enterprise Supported

Layer	Label		Created
sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4	LABEL	`com.nvidia.build.ref=`	01/31/2025 8:58 AM UTC
sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4	ARG	`NVIDIA_BUILD_REF`	01/31/2025 8:58 AM UTC
sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4	LABEL	`com.nvidia.build.id=134983853`	01/31/2025 8:58 AM UTC
sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4	ENV	`NVIDIA_BUILD_ID=134983853`	01/31/2025 8:58 AM UTC
sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4	ARG	`NVIDIA_BUILD_ID`	01/31/2025 8:58 AM UTC
sha256:4f4fb700ef54461cfa02571ae0db9a0dc1e0cdb5577484a6d75e68dc38e8acc1	RUN	NVIDIA_BIGNLP_VERSION= TRT_VER=10.8.0.24 CUDA_VER=12.8 CUDNN_VER=9.5.1.17-1 NCCL_VER= CUBLAS_VER=12.6.1.4-1 NVRTC_VER=12.6.85-1 REINSTALL_APEX=False APEX_COMMIT= MCORE_COMMIT=pyt_2501 SRC_DIR=/src/tensorrt_llm CPP_BUILD_DIR=/src/tensorrt_llm/cpp/build TARGET_ARCH=x86 INSTALL_VLLM=True ALIGNER_COMMIT=dev LAUNCHER_COMMIT= DATA_CURATOR_COMMIT=7ab04cefff64cb8914c3c871bc6626709fd3422a INSTALL_DATA_CURATOR=True MODELOPT_VERSION=0.21.0 NEMO_COMMIT= INSTALL_FAULT_TOLERANCE=False RESIL_COMMIT=98ebc4266a6b19203086c286f38b11b4fe38f4f1 NEMO_RUN_COMMIT= PRE_COMPILE_QUANT_KERNELS=False chmod -R a+w /workspace	01/31/2025 8:58 AM UTC
sha256:4f4fb700ef54461cfa02571ae0db9a0dc1e0cdb5577484a6d75e68dc38e8acc1	WORKDIR	`/workspace`	01/31/2025 8:58 AM UTC
sha256:4f4fb700ef54461cfa02571ae0db9a0dc1e0cdb5577484a6d75e68dc38e8acc1	RUN	NVIDIA_BIGNLP_VERSION= TRT_VER=10.8.0.24 CUDA_VER=12.8 CUDNN_VER=9.5.1.17-1 NCCL_VER= CUBLAS_VER=12.6.1.4-1 NVRTC_VER=12.6.85-1 REINSTALL_APEX=False APEX_COMMIT= MCORE_COMMIT=pyt_2501 SRC_DIR=/src/tensorrt_llm CPP_BUILD_DIR=/src/tensorrt_llm/cpp/build TARGET_ARCH=x86 INSTALL_VLLM=True ALIGNER_COMMIT=dev LAUNCHER_COMMIT= DATA_CURATOR_COMMIT=7ab04cefff64cb8914c3c871bc6626709fd3422a INSTALL_DATA_CURATOR=True MODELOPT_VERSION=0.21.0 NEMO_COMMIT= INSTALL_FAULT_TOLERANCE=False RESIL_COMMIT=98ebc4266a6b19203086c286f38b11b4fe38f4f1 NEMO_RUN_COMMIT= PRE_COMPILE_QUANT_KERNELS=False if [ $PRE_COMPILE_QUANT_KERNELS = "True" ]; then python -c "import modelopt.torch.quantization.extensions as ext; ext.precompile()"; fi	01/31/2025 8:58 AM UTC
sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4	ARG	`PRE_COMPILE_QUANT_KERNELS=False`	01/31/2025 8:58 AM UTC
sha256:83f07ce9b7fa1a624c1db18a71c6f35267953652f43795523035e19b707538f3	RUN	NVIDIA_BIGNLP_VERSION= TRT_VER=10.8.0.24 CUDA_VER=12.8 CUDNN_VER=9.5.1.17-1 NCCL_VER= CUBLAS_VER=12.6.1.4-1 NVRTC_VER=12.6.85-1 REINSTALL_APEX=False APEX_COMMIT= MCORE_COMMIT=pyt_2501 SRC_DIR=/src/tensorrt_llm CPP_BUILD_DIR=/src/tensorrt_llm/cpp/build TARGET_ARCH=x86 INSTALL_VLLM=True ALIGNER_COMMIT=dev LAUNCHER_COMMIT= DATA_CURATOR_COMMIT=7ab04cefff64cb8914c3c871bc6626709fd3422a INSTALL_DATA_CURATOR=True MODELOPT_VERSION=0.21.0 NEMO_COMMIT= INSTALL_FAULT_TOLERANCE=False RESIL_COMMIT=98ebc4266a6b19203086c286f38b11b4fe38f4f1 NEMO_RUN_COMMIT= sed -i '/^LayerId/s/^/#/' /usr/local/lib/python3.12/dist-packages/cv2/typing/__init__.py	01/31/2025 8:58 AM UTC