pub_soft/alphafast
3
0
Fork 0
mirror of https://github.com/RomeroLab/alphafast.git synced 2026-06-04 13:30:25 +08:00
Code Issues Packages Projects Releases Wiki Activity
Files
main
alphafast/scripts/setup_databases.sh

661 lines
22 KiB
Bash
Executable File
Raw Permalink Blame History

#!/bin/bash
# Copyright 2026 Romero Lab, Duke University
#
# Licensed under CC-BY-NC-SA 4.0. This file is part of AlphaFast,
# a derivative work of AlphaFold 3 by DeepMind Technologies Limited.
# https://creativecommons.org/licenses/by-nc-sa/4.0/
#
# Download AlphaFold 3 databases for AlphaFast.
#
# Default mode downloads protein and RNA databases from HuggingFace.
# Use --protein-only or --rna-only to narrow the scope.
# Use --from-source to download raw databases from Google Cloud Storage and
# build MMseqs databases locally.
#
# Usage:
# ./scripts/setup_databases.sh <target_dir> [OPTIONS]
#
# Arguments:
# target_dir: Directory where databases will be stored
#
# Options:
# --protein-only Download/build only protein databases and mmCIF structures
# --rna-only Download/build only RNA databases
# --include-nhmmer Download/build RNA FASTA fallback files
# --from-source Download raw databases from Google Cloud Storage and build locally
# --keep-fasta Keep raw FASTA files after conversion (--from-source only)
# --no-keep-fasta Remove raw FASTA files after conversion (--from-source only)
#
# Environment:
# HF_ENDPOINT HuggingFace-compatible endpoint for pre-built downloads,
# e.g. https://hf-mirror.com.
#
# Requirements (default / HuggingFace mode):
# - hf CLI (HuggingFace): curl -LsSf https://hf.co/cli/install.sh | bash -s
# - zstd, tar in PATH (for mmCIF extraction)
# - ~800 GB free disk space
#
# Requirements (--from-source mode):
# - wget, zstd, tar in PATH
# - mmseqs (GPU version) in PATH
# - ~800 GB free disk space (250 GB download + 540 GB MMseqs2 padded)
#
# Requirements (--from-prebuilt mode):
# - hf CLI in PATH (curl -LsSf https://hf.co/cli/install.sh | bash)
# - zstd, tar in PATH
# - ~569 GB free disk space
#
# Output directory structure:
# <target_dir>/
# mmcif_files/ # PDB structures for template retrieval
# mmseqs/
# uniref90_padded* # MMseqs2 GPU-ready databases (protein)
# mgnify_padded*
# small_bfd_padded*
# uniprot_padded*
# pdb_seqres_padded*
# mmseqs_rna/
# rfam* # MMseqs2 nucleotide databases (default RNA search)
# rnacentral*
# nt_rna*
# rnacentral_active_seq_id_90_cov_80_linclust.fasta # RNA FASTA (optional nhmmer fallback)
# rfam_14_9_clust_seq_id_90_cov_80_rep_seq.fasta # RNA FASTA (optional nhmmer fallback)
# nt_rna_2023_02_23_clust_seq_id_90_cov_80_rep_seq.fasta # RNA FASTA (optional nhmmer fallback)
set -euo pipefail
# ---------------------------------------------------------------------------
# Parse arguments
# ---------------------------------------------------------------------------
usage() {
local exit_code="${1:-1}"
echo "Usage: $0 <target_dir> [OPTIONS]"
echo ""
echo "Downloads AlphaFold 3 databases for AlphaFast."
echo ""
echo "Default mode downloads protein and RNA databases from HuggingFace."
echo "Use --protein-only or --rna-only to narrow the scope."
echo "Use --from-source to download raw databases from Google Cloud Storage"
echo "and build MMseqs databases locally."
echo ""
echo "Arguments:"
echo " target_dir Directory where databases will be stored"
echo ""
echo "Options:"
echo " -h, --help Show this help message and exit."
echo " --protein-only Download/build only protein databases and mmCIF structures."
echo " --rna-only Download/build only RNA databases."
echo " --include-nhmmer Download/build RNA FASTA fallback files."
echo " --from-source Build from FASTA files (Google Cloud Storage)."
echo " Requires wget, zstd, tar, mmseqs in PATH."
echo " --keep-fasta Keep raw FASTA files after conversion (default, --from-source only)"
echo " --no-keep-fasta Remove raw FASTA files after conversion (--from-source only)"
echo ""
echo "Environment:"
echo " HF_ENDPOINT HuggingFace-compatible endpoint for pre-built downloads"
echo " (for example: https://hf-mirror.com)."
exit "$exit_code"
}
if [ "$#" -eq 0 ]; then
usage
fi
if [ "$1" = "-h" ] || [ "$1" = "--help" ]; then
usage 0
fi
TARGET_DIR="$1"
shift
KEEP_FASTA=true
FROM_PREBUILT=true # HuggingFace is the default
INSTALL_MODE="all" # "all", "protein-only", or "rna-only"
INCLUDE_NHMMER=false
DOWNLOAD_RNA_MMSEQS=true
DOWNLOAD_RNA_FASTA=false
while [ "$#" -gt 0 ]; do
case "$1" in
-h|--help)
usage 0
;;
--keep-fasta)
KEEP_FASTA=true
shift
;;
--no-keep-fasta)
KEEP_FASTA=false
shift
;;
--from-prebuilt)
FROM_PREBUILT=true
shift
;;
--from-source)
FROM_PREBUILT=false
shift
;;
--protein-only)
INSTALL_MODE="protein-only"
shift
;;
--rna-only)
INSTALL_MODE="rna-only"
shift
;;
--include-nhmmer)
INCLUDE_NHMMER=true
shift
;;
*)
echo "Unknown argument: $1"
usage
;;
esac
done
if $FROM_PREBUILT && [ "$KEEP_FASTA" = false ]; then
echo "ERROR: --no-keep-fasta can only be used with --from-source."
usage
fi
case "$INSTALL_MODE" in
all)
DOWNLOAD_PROTEIN=true
DOWNLOAD_RNA_MMSEQS=true
DOWNLOAD_RNA_FASTA="$INCLUDE_NHMMER"
;;
protein-only)
DOWNLOAD_PROTEIN=true
DOWNLOAD_RNA_MMSEQS=false
DOWNLOAD_RNA_FASTA=false
;;
rna-only)
DOWNLOAD_PROTEIN=false
DOWNLOAD_RNA_MMSEQS=true
DOWNLOAD_RNA_FASTA="$INCLUDE_NHMMER"
;;
esac
# ---------------------------------------------------------------------------
# Check prerequisites
# ---------------------------------------------------------------------------
MISSING=0
if $FROM_PREBUILT; then
for cmd in hf tar zstd; do
if ! command -v "$cmd" &>/dev/null; then
if [ "$cmd" = "hf" ]; then
echo "ERROR: $cmd is not installed. Install with: curl -LsSf https://hf.co/cli/install.sh | bash -s"
else
echo "ERROR: $cmd is not installed or not in PATH."
fi
MISSING=1
fi
done
else
for cmd in wget tar zstd mmseqs; do
if ! command -v "$cmd" &>/dev/null; then
echo "ERROR: $cmd is not installed or not in PATH."
MISSING=1
fi
done
fi
if [ "$MISSING" -ne 0 ]; then
echo ""
echo "Install missing dependencies before running this script."
if $FROM_PREBUILT; then
echo "For HuggingFace CLI:"
echo " curl -LsSf https://hf.co/cli/install.sh | bash -s"
else
echo "For mmseqs with GPU support:"
echo " wget https://mmseqs.com/latest/mmseqs-linux-gpu.tar.gz"
echo " tar xzf mmseqs-linux-gpu.tar.gz"
echo " sudo cp mmseqs/bin/mmseqs /usr/local/bin/"
fi
exit 1
fi
if ! $FROM_PREBUILT; then
echo "Using MMseqs2 version: $(mmseqs version)"
fi
# ---------------------------------------------------------------------------
# Configuration
# ---------------------------------------------------------------------------
readonly SOURCE="https://storage.googleapis.com/alphafold-databases/v3.0"
readonly HF_REPO="RomeroLab-Duke/af3-mmseqs-db"
MMSEQS_DIR="${TARGET_DIR}/mmseqs"
RNA_MMSEQS_DIR="${TARGET_DIR}/mmseqs_rna"
RNA_FASTAS=(
"rnacentral_active_seq_id_90_cov_80_linclust.fasta"
"nt_rna_2023_02_23_clust_seq_id_90_cov_80_rep_seq.fasta"
"rfam_14_9_clust_seq_id_90_cov_80_rep_seq.fasta"
)
mkdir -p "$TARGET_DIR"
if [ "$DOWNLOAD_PROTEIN" = true ]; then
mkdir -p "$MMSEQS_DIR"
fi
if [ "$DOWNLOAD_RNA_MMSEQS" = true ]; then
mkdir -p "$RNA_MMSEQS_DIR"
fi
reassemble_part_files() {
local target_dir="$1"
shopt -s nullglob
for part_prefix in "${target_dir}"/*.part00; do
if [ -f "$part_prefix" ]; then
local base="${part_prefix%.part00}"
echo "Reassembling $(basename "$base")..."
cat "${base}.part"* > "$base"
rm -f "${base}.part"*
fi
done
shopt -u nullglob
}
decompress_zst_files() {
local target_dir="$1"
shopt -s nullglob
for compressed in "${target_dir}"/*.zst; do
if [ -f "$compressed" ]; then
local output="${compressed%.zst}"
if [ -f "$output" ]; then
echo "SKIP: $(basename "$output") already decompressed"
rm -f "$compressed"
continue
fi
echo "Decompressing $(basename "$compressed")..."
zstd --decompress --force --rm -o "$output" "$compressed"
fi
done
shopt -u nullglob
}
echo "=========================================="
echo "AlphaFast Database Setup"
echo "=========================================="
echo "Target directory: $TARGET_DIR"
echo "MMseqs2 directory: $MMSEQS_DIR"
echo "Mode: $(if $FROM_PREBUILT; then echo 'pre-built (HuggingFace)'; else echo 'download raw + build locally'; fi)"
echo "Install mode: $INSTALL_MODE"
echo "RNA MMseqs2: $DOWNLOAD_RNA_MMSEQS"
echo "RNA FASTA: $DOWNLOAD_RNA_FASTA"
echo "Keep FASTA files: $KEEP_FASTA"
echo "Start time: $(date)"
echo "=========================================="
echo ""
# ===========================================================================
# Pre-built mode: download from HuggingFace
# ===========================================================================
if $FROM_PREBUILT; then
echo "=== Downloading pre-built databases from HuggingFace ==="
echo "Repository: $HF_REPO"
echo ""
if [ "$DOWNLOAD_PROTEIN" = true ]; then
# Download mmCIF structures
MMCIF_DIR="${TARGET_DIR}/mmcif_files"
if [ -d "$MMCIF_DIR" ] && [ "$(ls -A "$MMCIF_DIR" 2>/dev/null)" ]; then
echo "SKIP: mmcif_files already exists"
else
echo "Downloading mmCIF structures..."
hf download "$HF_REPO" --repo-type dataset --include "mmcif_files.tar.zst.*" --local-dir "$TARGET_DIR"
# Reassemble and extract
cat "${TARGET_DIR}/mmcif_files.tar.zst.part"* | tar --use-compress-program=zstd -xf - --directory="$TARGET_DIR"
rm -f "${TARGET_DIR}/mmcif_files.tar.zst.part"*
echo "Done: mmCIF structures"
fi
echo ""
else
echo "SKIP: Protein databases (--rna-only mode)"
echo ""
fi
# Download protein MMseqs2 padded databases
if [ "$DOWNLOAD_PROTEIN" = true ]; then
if [ -f "${MMSEQS_DIR}/uniref90_padded.dbtype" ]; then
echo "SKIP: Protein MMseqs2 databases already exist"
else
echo "Downloading protein MMseqs2 padded databases..."
hf download "$HF_REPO" --repo-type dataset \
--include "mmseqs/*.zst" --include "mmseqs/*.zst.*" \
--local-dir "$TARGET_DIR"
# Reassemble any split .partNN files and decompress .zst payloads.
reassemble_part_files "$MMSEQS_DIR"
decompress_zst_files "$MMSEQS_DIR"
echo "Done: Protein MMseqs2 databases"
fi
echo ""
fi
if [ "$DOWNLOAD_RNA_MMSEQS" = true ]; then
RNA_MMSEQS_DIR="${TARGET_DIR}/mmseqs_rna"
if [ -f "${RNA_MMSEQS_DIR}/rfam.dbtype" ]; then
echo "SKIP: RNA MMseqs2 databases already exist"
else
# Download everything including pre-built indices (default)
echo "Downloading RNA MMseqs2 nucleotide databases (with pre-built indices)..."
hf download "$HF_REPO" --repo-type dataset \
--include "mmseqs_rna/*.zst" --include "mmseqs_rna/*.zst.*" \
--local-dir "$TARGET_DIR"
reassemble_part_files "$RNA_MMSEQS_DIR"
decompress_zst_files "$RNA_MMSEQS_DIR"
echo "Done: RNA MMseqs2 databases"
fi
echo ""
else
echo "SKIP: RNA MMseqs2 databases (--protein-only mode)"
echo ""
fi
if [ "$DOWNLOAD_RNA_FASTA" = true ]; then
# Download RNA FASTA databases (for nhmmer fallback)
echo "Downloading RNA FASTA databases (for nhmmer fallback)..."
hf download "$HF_REPO" --repo-type dataset --include "*.fasta" --include "*.fasta.*" --local-dir "$TARGET_DIR"
# Reassemble any split files
for part_prefix in "${TARGET_DIR}"/*.fasta.part00; do
if [ -f "$part_prefix" ]; then
base="${part_prefix%.part00}"
echo "Reassembling $(basename "$base")..."
cat "${base}.part"* >"$base"
rm -f "${base}.part"*
fi
done
echo "Done: RNA FASTA databases"
echo ""
elif [ "$INSTALL_MODE" = "rna-only" ] || [ "$INSTALL_MODE" = "all" ]; then
echo "SKIP: RNA FASTA databases (use --include-nhmmer to download them)"
echo ""
else
echo "SKIP: RNA databases (--protein-only mode)"
echo ""
fi
echo "=== Pre-built download complete ==="
echo ""
# ===========================================================================
# Build mode: download FASTA and convert
# ===========================================================================
else
# ---------------------------------------------------------------------------
# Step 1: Download databases
# ---------------------------------------------------------------------------
echo "=== Step 1: Download databases ==="
echo ""
# PDB mmCIF structures (for template retrieval)
MMCIF_DIR="${TARGET_DIR}/mmcif_files"
if [ -d "$MMCIF_DIR" ] && [ "$(ls -A "$MMCIF_DIR" 2>/dev/null)" ]; then
echo "SKIP: mmcif_files already exists at $MMCIF_DIR"
else
echo "Downloading PDB mmCIF structures..."
wget --progress=bar:force:noscroll -O - \
"${SOURCE}/pdb_2022_09_28_mmcif_files.tar.zst" |
tar --no-same-owner --no-same-permissions \
--use-compress-program=zstd -xf - --directory="$TARGET_DIR"
echo "Done: mmCIF structures"
fi
echo ""
# Protein FASTA databases (for MSA search)
declare -A PROTEIN_FASTAS=(
["uniref90_2022_05.fa"]="UniRef90"
["mgy_clusters_2022_05.fa"]="MGnify"
["bfd-first_non_consensus_sequences.fasta"]="Small BFD"
["uniprot_all_2021_04.fa"]="UniProt"
["pdb_seqres_2022_09_28.fasta"]="PDB SeqRes"
)
if [ "$DOWNLOAD_PROTEIN" = true ]; then
for fasta_file in "${!PROTEIN_FASTAS[@]}"; do
db_label="${PROTEIN_FASTAS[$fasta_file]}"
target_path="${TARGET_DIR}/${fasta_file}"
if [ -f "$target_path" ]; then
echo "SKIP: $db_label already exists at $target_path"
else
echo "Downloading $db_label ($fasta_file)..."
wget --progress=bar:force:noscroll -O - \
"${SOURCE}/${fasta_file}.zst" |
zstd --decompress >"$target_path"
echo "Done: $db_label"
fi
done
else
echo "SKIP: Protein FASTA downloads (--rna-only mode)"
fi
echo ""
# RNA databases (for RNA MSA search via nhmmer).
RNA_FASTAS=(
"rnacentral_active_seq_id_90_cov_80_linclust.fasta"
"nt_rna_2023_02_23_clust_seq_id_90_cov_80_rep_seq.fasta"
"rfam_14_9_clust_seq_id_90_cov_80_rep_seq.fasta"
)
if [ "$DOWNLOAD_RNA_MMSEQS" = true ] || [ "$DOWNLOAD_RNA_FASTA" = true ]; then
for fasta_file in "${RNA_FASTAS[@]}"; do
target_path="${TARGET_DIR}/${fasta_file}"
if [ -f "$target_path" ]; then
echo "SKIP: RNA database already exists at $target_path"
else
echo "Downloading RNA database ($fasta_file)..."
wget --progress=bar:force:noscroll -O - \
"${SOURCE}/${fasta_file}.zst" |
zstd --decompress >"$target_path"
echo "Done: $fasta_file"
fi
done
elif [ "$INSTALL_MODE" = "rna-only" ] || [ "$INSTALL_MODE" = "all" ]; then
echo "SKIP: RNA downloads (RNA mode disabled)"
else
echo "SKIP: RNA FASTA downloads (--protein-only mode)"
fi
echo ""
# ---------------------------------------------------------------------------
# Step 1b: Build MMseqs2 databases from RNA FASTA (for nucleotide search)
# ---------------------------------------------------------------------------
if [ "$DOWNLOAD_RNA_MMSEQS" = true ]; then
echo "=== Step 1b: Build MMseqs2 RNA databases (optional) ==="
echo ""
RNA_MMSEQS_DIR="${TARGET_DIR}/mmseqs_rna"
mkdir -p "$RNA_MMSEQS_DIR"
declare -A RNA_MMSEQS_DATABASES=(
["rfam"]="rfam_14_9_clust_seq_id_90_cov_80_rep_seq.fasta"
["rnacentral"]="rnacentral_active_seq_id_90_cov_80_linclust.fasta"
["nt_rna"]="nt_rna_2023_02_23_clust_seq_id_90_cov_80_rep_seq.fasta"
)
for db_name in "${!RNA_MMSEQS_DATABASES[@]}"; do
source_fasta="${TARGET_DIR}/${RNA_MMSEQS_DATABASES[$db_name]}"
target_db="${RNA_MMSEQS_DIR}/${db_name}"
if [ -f "${target_db}.dbtype" ]; then
echo "SKIP: MMseqs2 RNA database ${db_name} already exists"
continue
fi
if [ ! -f "$source_fasta" ]; then
echo "SKIP: Source FASTA not found: $source_fasta"
continue
fi
echo "Creating MMseqs2 nucleotide database: $db_name"
time mmseqs createdb "$source_fasta" "$target_db"
# Build k-mer index for fast search.
if [ ! -f "${target_db}.idx" ]; then
echo "Creating search index for $db_name..."
idx_tmp=$(mktemp -d)
source_size_gb=$(du -B1G "${TARGET_DIR}/${RNA_MMSEQS_DATABASES[$db_name]}" 2>/dev/null | cut -f1)
split_flag=""
if [ "${source_size_gb:-0}" -gt 10 ]; then
split_flag="--split 4"
echo " Using --split 4 for large database (${source_size_gb}G)"
fi
time mmseqs createindex "$target_db" "$idx_tmp" --search-type 3 $split_flag
rm -rf "$idx_tmp"
fi
echo "Done: $db_name"
done
echo ""
else
echo "SKIP: RNA MMseqs2 database build (--protein-only mode)"
echo ""
RNA_MMSEQS_DIR="${TARGET_DIR}/mmseqs_rna"
fi
# ---------------------------------------------------------------------------
# Step 2: Convert protein FASTA to MMseqs2 padded format
# ---------------------------------------------------------------------------
if [ "$DOWNLOAD_PROTEIN" = true ]; then
echo "=== Step 2: Convert to MMseqs2 GPU format ==="
echo ""
# Database mapping: mmseqs_name -> source_fasta_filename
declare -A MMSEQS_DATABASES=(
["uniref90"]="uniref90_2022_05.fa"
["mgnify"]="mgy_clusters_2022_05.fa"
["small_bfd"]="bfd-first_non_consensus_sequences.fasta"
["uniprot"]="uniprot_all_2021_04.fa"
["pdb_seqres"]="pdb_seqres_2022_09_28.fasta"
)
total_dbs=${#MMSEQS_DATABASES[@]}
current_db=0
for db_name in "${!MMSEQS_DATABASES[@]}"; do
current_db=$((current_db + 1))
source_fasta="${TARGET_DIR}/${MMSEQS_DATABASES[$db_name]}"
target_base="${MMSEQS_DIR}/${db_name}"
target_padded="${target_base}_padded"
echo "[$current_db/$total_dbs] Converting: $db_name"
# Check if padded database already exists
if [ -f "${target_padded}.dbtype" ]; then
echo " SKIP: Padded database already exists"
echo ""
continue
fi
if [ ! -f "$source_fasta" ]; then
echo " WARNING: Source FASTA not found: $source_fasta"
echo " Skipping."
echo ""
continue
fi
# Step 1: createdb
if [ -f "${target_base}.dbtype" ]; then
echo " Found intermediate database, skipping createdb..."
else
echo " Creating MMseqs2 database..."
time mmseqs createdb "$source_fasta" "$target_base"
fi
# Step 2: makepaddedseqdb
echo " Creating padded database for GPU..."
time mmseqs makepaddedseqdb "$target_base" "$target_padded"
# Clean up intermediate (non-padded) database files
echo " Cleaning up intermediate database files..."
rm -f "${target_base}" "${target_base}".dbtype "${target_base}".index \
"${target_base}".lookup "${target_base}".source \
"${target_base}_h" "${target_base}_h".dbtype "${target_base}_h".index
if [ -f "${target_padded}.dbtype" ]; then
echo " SUCCESS: Created ${target_padded}"
else
echo " ERROR: Failed to create padded database"
fi
echo ""
done
else
echo "=== Step 2: Convert to MMseqs2 GPU format ==="
echo "SKIP: Protein MMseqs2 conversion (--rna-only mode)"
echo ""
fi
# ---------------------------------------------------------------------------
# Step 3: Optionally clean up raw FASTA files
# ---------------------------------------------------------------------------
if [ "$KEEP_FASTA" = false ]; then
echo "=== Step 3: Removing raw FASTA files ==="
for fasta_file in "${!PROTEIN_FASTAS[@]}"; do
target_path="${TARGET_DIR}/${fasta_file}"
if [ -f "$target_path" ]; then
echo "Removing: $target_path"
rm -f "$target_path"
fi
done
if [ "$DOWNLOAD_RNA_FASTA" = false ]; then
for fasta_file in "${RNA_FASTAS[@]}"; do
target_path="${TARGET_DIR}/${fasta_file}"
if [ -f "$target_path" ]; then
echo "Removing: $target_path"
rm -f "$target_path"
fi
done
fi
echo ""
fi
fi # end of: if $FROM_PREBUILT; then ... else (build mode)
# ---------------------------------------------------------------------------
# Summary
# ---------------------------------------------------------------------------
echo "=========================================="
echo "Setup Complete!"
echo "=========================================="
echo "End time: $(date)"
echo ""
echo "Database directory: $TARGET_DIR"
echo "MMseqs2 directory: $MMSEQS_DIR"
echo "mmCIF directory: ${TARGET_DIR}/mmcif_files"
echo ""
echo "RNA MMseqs2 databases (default RNA search):"
if [ "$DOWNLOAD_RNA_MMSEQS" = true ]; then
for rna_db in rfam rnacentral nt_rna; do
if [ -f "${RNA_MMSEQS_DIR}/${rna_db}.dbtype" ]; then
echo " OK: ${rna_db}"
else
echo " NOT BUILT: ${rna_db}"
fi
done
else
echo " SKIPPED: --protein-only mode"
fi
echo ""
echo "RNA FASTA databases (nhmmer fallback, --use_nhmmer):"
if [ "$DOWNLOAD_RNA_FASTA" = true ]; then
for rna_f in "${RNA_FASTAS[@]}"; do
if [ -f "${TARGET_DIR}/${rna_f}" ]; then
echo " OK: ${rna_f}"
else
echo " MISSING: ${rna_f}"
fi
done
elif [ "$INSTALL_MODE" = "rna-only" ] || [ "$INSTALL_MODE" = "all" ]; then
echo " SKIPPED: use --include-nhmmer to download them"
else
echo " SKIPPED: --protein-only mode"
fi
echo ""
echo "Use these paths with run_alphafast.sh:"
echo " ./scripts/run_alphafast.sh \\"
echo " --db_dir $TARGET_DIR \\"
echo " --weights_dir /path/to/weights \\"
echo " --input_dir /path/to/inputs \\"
echo " --output_dir /path/to/outputs"
echo "=========================================="
Reference in New Issue View Git Blame Copy Permalink