ViperMCP

#download-models.sh #!/usr/bin/env bash SCRIPT_DIR="$(cd -- "$(dirname -- "${BASH_SOURCE[0]}")" &> /dev/null && pwd)" PRETRAINED_MODELS_PATH="$SCRIPT_DIR/viper/pretrained_models" # ------------ Create directories for pretrained models ---------- for d in midas blip xvlm bert llama codex dino sam; do mkdir -p "$PRETRAINED_MODELS_PATH/$d" done mkdir -p "$PRETRAINED_MODELS_PATH/glip/checkpoints" "$PRETRAINED_MODELS_PATH/glip/configs" # ---------- Depth-estimation (DPT-Hybrid MiDaS) ---------- huggingface-cli download Intel/dpt-hybrid-midas \ --local-dir "$PRETRAINED_MODELS_PATH/midas" \ --local-dir-use-symlinks False # ---------- Simple VQA (BLIP-2 Flan T5 xl) ---------- # huggingface-cli download Salesforce/blip2-flan-t5-xl \ # --local-dir "$PRETRAINED_MODELS_PATH/blip" \ # --local-dir-use-symlinks False \ # --include "*.safetensors" \ # ---------- BERT tokenizer ---------- huggingface-cli download google-bert/bert-base-uncased \ --local-dir "$PRETRAINED_MODELS_PATH/bert" \ --local-dir-use-symlinks False \ --include "*.json" "*.txt" "*.safetensors" # ---------- Similarity (XVLM) ---------- if [ ! -f "$PRETRAINED_MODELS_PATH/xvlm/retrieval_mscoco_checkpoint_9.pth" ]; then echo "Downloading XVLM model..." gdown "https://drive.google.com/uc?id=1bv6_pZOsXW53EhlwU0ZgSk03uzFI61pN" \ -O "$PRETRAINED_MODELS_PATH/xvlm/retrieval_mscoco_checkpoint_9.pth" else echo "XVLM model already exists, skipping download." fi ## ---------- Object detection (GLIP) ---------- #wget -nc -P "$PRETRAINED_MODELS_PATH/glip/checkpoints" \ # https://huggingface.co/GLIPModel/GLIP/resolve/main/glip_large_model.pth #wget -nc -P "$PRETRAINED_MODELS_PATH/glip/configs" \ # https://raw.githubusercontent.com/microsoft/GLIP/main/configs/pretrain/glip_Swin_L.yaml # ---------- (Optional) Llama-2 7B ---------- # ~13 GB download; uncomment if/when you really need it # huggingface-cli download NousResearch/Llama-2-7b-hf \ # --local-dir "$PRETRAINED_MODELS_PATH/llama" \ # --local-dir-use-symlinks False \ # --exclude "*.bin" \ # ---------- Object Detection (Grounding DINO) ---------- huggingface-cli download IDEA-Research/grounding-dino-base \ --local-dir "$PRETRAINED_MODELS_PATH/dino" \ --local-dir-use-symlinks False \ --exclude "*.bin" \ # ---------- Segment Anything (SAM) ---------- huggingface-cli download facebook/sam-vit-base \ --local-dir "$PRETRAINED_MODELS_PATH/sam" \ --local-dir-use-symlinks False \ --exclude "*.bin" \ # ---------- Code Llama 7B Instruct ---------- # huggingface-cli download codellama/CodeLlama-7b-Instruct-hf \ # --local-dir "$PRETRAINED_MODELS_PATH/codex" \ # --local-dir-use-symlinks False \ # --exclude "*.bin" \