[D-47?] mamba installation

1. 기존 mamba 환경 삭제 (선택사항)

conda deactivate conda env remove -n mamba

2. 새 환경 생성 (Python 3.10 권장)

conda create -n mamba python=3.10 -y conda activate mamba

3. GCC 9+ 설치 (conda로)

conda install -c conda-forge gxx_linux-64=11 gcc_linux-64=11 -y

4. CUDA 환경 설정

module load cuda/11.8 export CUDA_HOME=/nfs/cuda/cuda-11.8/ export PATH=$CUDA_HOME/bin:$PATH

5. PyTorch 설치 (CUDA 12.4 호환 버전)

pip install torch==2.4.0 –index-url https://download.pytorch.org/whl/cu118

6. 설치 확인

python -c “import torch; print(f’PyTorch: {torch.version}, CUDA: {torch.version.cuda}’)”

7. causal-conv1d 설치

pip install ninja packaging pip install numpy pip install causal-conv1d

8. mamba-ssm 설치

pip install mamba-ssm

srun --job-name=install      --gres=gpu:1      --constraint="gpu_mem:80gb"      --ntasks=1      --cpus-per-task=8      --mem=64GB      --partition=gpu-athene      --account=athene-student      --qos=gpu-small      --time=01:00:00      --pty bash

bash-4.4$ module load cuda/11.8
bash-4.4$ source /storage/athene/work/kim/miniconda3/etc/profile.d/conda.sh
bash-4.4$ conda activate mamba