# AutoMoT + Bench2Drive requirements # Python 3.10, CUDA 12.8 # # PyTorch (install separately before this file): # pip install torch==2.7.1+cu128 torchvision==0.22.1+cu128 torchaudio==2.7.1+cu128 \ # --index-url https://download.pytorch.org/whl/cu128 # # flash-attn (install after torch): # pip install flash-attn==2.8.3 --no-build-isolation # --- Model inference --- transformers==4.57.3 safetensors==0.4.5 accelerate==1.10.1 huggingface-hub==0.36.0 tokenizers==0.22.1 peft==0.18.1 timm==1.0.22 einops==0.8.1 qwen-vl-utils==0.0.14 # --- Agent utilities --- filterpy==1.4.5 jsonpickle==4.1.1 scipy==1.15.3 numpy==1.26.4 pillow==11.3.0 imageio==2.37.2 opencv-python==4.7.0.72 laspy==2.6.1 shapely==2.0.7 transforms3d==0.4.2 networkx==3.4.2 matplotlib==3.7.0 psutil==7.2.1 # --- CARLA / leaderboard --- pygame==2.6.1 dictor==0.1.12 pexpect==4.9.0 tabulate==0.9.0 py-trees==0.8.3 ephem==4.2 xmlschema==4.2.0 six==1.17.0 simple-watchdog-timer==0.1.1 pytorch-lightning==2.6.1 torchmetrics==1.9.0 # --- Training --- wandb==0.25.1