Upload hf_upload_173.sh with huggingface_hub
Browse files- hf_upload_173.sh +149 -0
hf_upload_173.sh
ADDED
|
@@ -0,0 +1,149 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
#!/bin/bash
|
| 2 |
+
|
| 3 |
+
# Configuration
|
| 4 |
+
REPO_ID="benzlxs/objaverse_rendering_set"
|
| 5 |
+
REPO_TYPE="dataset"
|
| 6 |
+
LOCAL_DIR="zip_folder"
|
| 7 |
+
STABILITY_CHECK_INTERVAL=150 # Seconds between size checks
|
| 8 |
+
STABILITY_THRESHOLD=2 # Number of consecutive checks with same size before considering stable
|
| 9 |
+
LOG_FILE="upload_log_173.txt"
|
| 10 |
+
|
| 11 |
+
# Enable hf_transfer for faster uploads
|
| 12 |
+
export HF_HUB_ENABLE_HF_TRANSFER=1
|
| 13 |
+
|
| 14 |
+
# Colors for output
|
| 15 |
+
GREEN='\033[0;32m'
|
| 16 |
+
YELLOW='\033[1;33m'
|
| 17 |
+
RED='\033[0;31m'
|
| 18 |
+
NC='\033[0m' # No Color
|
| 19 |
+
|
| 20 |
+
# Function to log messages
|
| 21 |
+
log_message() {
|
| 22 |
+
echo -e "$1"
|
| 23 |
+
echo "[$(date '+%Y-%m-%d %H:%M:%S')] $1" >> "$LOG_FILE"
|
| 24 |
+
}
|
| 25 |
+
|
| 26 |
+
# Function to check if file size is stable
|
| 27 |
+
is_file_stable() {
|
| 28 |
+
local file=$1
|
| 29 |
+
local prev_size=0
|
| 30 |
+
local stable_count=0
|
| 31 |
+
|
| 32 |
+
log_message "${YELLOW}Monitoring $file for stability...${NC}"
|
| 33 |
+
|
| 34 |
+
while [ $stable_count -lt $STABILITY_THRESHOLD ]; do
|
| 35 |
+
if [ ! -f "$file" ]; then
|
| 36 |
+
log_message "${YELLOW}File $file not found yet, waiting...${NC}"
|
| 37 |
+
sleep $STABILITY_CHECK_INTERVAL
|
| 38 |
+
continue
|
| 39 |
+
fi
|
| 40 |
+
|
| 41 |
+
current_size=$(stat -f%z "$file" 2>/dev/null || stat -c%s "$file" 2>/dev/null)
|
| 42 |
+
|
| 43 |
+
if [ "$current_size" -eq "$prev_size" ] && [ "$current_size" -gt 0 ]; then
|
| 44 |
+
stable_count=$((stable_count + 1))
|
| 45 |
+
log_message "Size unchanged: $current_size bytes (check $stable_count/$STABILITY_THRESHOLD)"
|
| 46 |
+
else
|
| 47 |
+
stable_count=0
|
| 48 |
+
prev_size=$current_size
|
| 49 |
+
log_message "Size changed to: $current_size bytes, resetting counter"
|
| 50 |
+
fi
|
| 51 |
+
|
| 52 |
+
if [ $stable_count -lt $STABILITY_THRESHOLD ]; then
|
| 53 |
+
sleep $STABILITY_CHECK_INTERVAL
|
| 54 |
+
fi
|
| 55 |
+
done
|
| 56 |
+
|
| 57 |
+
log_message "${GREEN}File $file is stable at $current_size bytes${NC}"
|
| 58 |
+
return 0
|
| 59 |
+
}
|
| 60 |
+
|
| 61 |
+
# Function to wait for file to exist
|
| 62 |
+
wait_for_file() {
|
| 63 |
+
local file=$1
|
| 64 |
+
local max_wait=3600 # Maximum wait time in seconds (1 hour)
|
| 65 |
+
local waited=0
|
| 66 |
+
|
| 67 |
+
while [ ! -f "$file" ] && [ $waited -lt $max_wait ]; do
|
| 68 |
+
log_message "${YELLOW}Waiting for $file to be created...${NC}"
|
| 69 |
+
sleep 10
|
| 70 |
+
waited=$((waited + 10))
|
| 71 |
+
done
|
| 72 |
+
|
| 73 |
+
if [ ! -f "$file" ]; then
|
| 74 |
+
log_message "${RED}Timeout waiting for $file${NC}"
|
| 75 |
+
return 1
|
| 76 |
+
fi
|
| 77 |
+
|
| 78 |
+
return 0
|
| 79 |
+
}
|
| 80 |
+
|
| 81 |
+
# Main upload loop
|
| 82 |
+
log_message "${GREEN}=== Starting Hugging Face Upload Script ===${NC}"
|
| 83 |
+
log_message "Repository: $REPO_ID"
|
| 84 |
+
log_message "Local directory: $LOCAL_DIR"
|
| 85 |
+
|
| 86 |
+
|
| 87 |
+
# for i in 000-068 000-078 000-088 000-100 000-112 000-122 000-132 000-142 000-152 000-009 000-019 000-029 000-039 000-049 000-059 000-069 000-079 000-089 000-102 000-113 000-123 000-133 000-143 000-153
|
| 88 |
+
# for i in 000-068 000-078 000-088 000-100 000-112 000-122 000-132 000-142 000-152 000-009 000-019 000-029 000-039 000-059 000-069 000-079 000-089 000-102 000-113 000-123 000-133 000-143 000-153
|
| 89 |
+
# for i in 000-085 000-096 000-108 000-119 000-129 000-139 000-149 000-159 000-006 000-016 000-026 000-036 000-046 000-056 000-066 000-076 000-086 000-097 000-109 000-120 000-130 000-140 000-150 000-007 000-017 000-027 000-037 000-047 000-057 000-067 000-077 000-087 000-098 000-110 000-121 000-131 000-141 000-151 000-008 000-018 000-028 000-038 000-048 000-058
|
| 90 |
+
# for i in 000-010 000-020 000-030 000-040 000-050 000-060 000-070 000-080 000-090 000-103 000-114 000-124 000-134 000-144 000-154 000-001 000-011 000-021 000-031 000-041 000-051 000-061 000-071
|
| 91 |
+
for i in 000-126 000-136 000-146 000-156 000-003 000-013 000-023 000-033 000-043 000-053 000-063 000-073 000-083 000-094 000-106 000-117 000-127 000-137 000-147 000-157 000-004 000-014 000-024 000-034
|
| 92 |
+
do
|
| 93 |
+
tar_file="${LOCAL_DIR}/${i}.tar"
|
| 94 |
+
log_message "\n${GREEN}========================================${NC}"
|
| 95 |
+
log_message "${GREEN}Processing: ${i}.tar${NC}"
|
| 96 |
+
log_message "${GREEN}========================================${NC}"
|
| 97 |
+
|
| 98 |
+
# Wait for the tar file to appear (if it doesn't exist yet)
|
| 99 |
+
if [ ! -f "$tar_file" ]; then
|
| 100 |
+
log_message "${YELLOW}File ${i}.tar doesn't exist yet, waiting for creation...${NC}"
|
| 101 |
+
if ! wait_for_file "$tar_file"; then
|
| 102 |
+
log_message "${RED}Skipping ${i}.tar - file was not created within timeout${NC}"
|
| 103 |
+
continue
|
| 104 |
+
fi
|
| 105 |
+
fi
|
| 106 |
+
|
| 107 |
+
# Wait for file to be stable (packing complete)
|
| 108 |
+
if is_file_stable "$tar_file"; then
|
| 109 |
+
# Get final file size for logging
|
| 110 |
+
file_size=$(stat -f%z "$tar_file" 2>/dev/null || stat -c%s "$tar_file" 2>/dev/null)
|
| 111 |
+
file_size_mb=$((file_size / 1048576))
|
| 112 |
+
|
| 113 |
+
log_message "${GREEN}Uploading ${i}.tar (${file_size_mb} MB) to Hugging Face...${NC}"
|
| 114 |
+
|
| 115 |
+
# Upload with retry logic
|
| 116 |
+
max_retries=3
|
| 117 |
+
retry_count=0
|
| 118 |
+
upload_success=false
|
| 119 |
+
|
| 120 |
+
while [ $retry_count -lt $max_retries ] && [ "$upload_success" = false ]; do
|
| 121 |
+
if huggingface-cli upload "$REPO_ID" "$tar_file" "${i}.tar" --repo-type="$REPO_TYPE"; then
|
| 122 |
+
log_message "${GREEN}✓ Successfully uploaded ${i}.tar${NC}"
|
| 123 |
+
upload_success=true
|
| 124 |
+
|
| 125 |
+
# Optional: Delete local file after successful upload to save space
|
| 126 |
+
# rm "$tar_file"
|
| 127 |
+
# log_message "${YELLOW}Deleted local file ${i}.tar${NC}"
|
| 128 |
+
else
|
| 129 |
+
retry_count=$((retry_count + 1))
|
| 130 |
+
log_message "${RED}Upload failed for ${i}.tar (attempt $retry_count/$max_retries)${NC}"
|
| 131 |
+
if [ $retry_count -lt $max_retries ]; then
|
| 132 |
+
log_message "Retrying in 30 seconds..."
|
| 133 |
+
sleep 30
|
| 134 |
+
fi
|
| 135 |
+
fi
|
| 136 |
+
done
|
| 137 |
+
|
| 138 |
+
if [ "$upload_success" = false ]; then
|
| 139 |
+
log_message "${RED}✗ Failed to upload ${i}.tar after $max_retries attempts${NC}"
|
| 140 |
+
fi
|
| 141 |
+
else
|
| 142 |
+
log_message "${RED}File stability check failed for ${i}.tar${NC}"
|
| 143 |
+
fi
|
| 144 |
+
|
| 145 |
+
log_message "${GREEN}Completed processing ${i}.tar${NC}"
|
| 146 |
+
done
|
| 147 |
+
|
| 148 |
+
log_message "\n${GREEN}=== All files processed ===${NC}"
|
| 149 |
+
log_message "Check $LOG_FILE for detailed logs"
|