Spaces:
Running
Running
File size: 73,138 Bytes
ab4e093 cca1fa9 ab4e093 cca1fa9 ab4e093 cca1fa9 ab4e093 cca1fa9 ab4e093 cca1fa9 ab4e093 cca1fa9 ab4e093 cca1fa9 ab4e093 cca1fa9 ab4e093 cca1fa9 ab4e093 b0e7314 ab4e093 cca1fa9 ab4e093 cca1fa9 ab4e093 cca1fa9 ab4e093 cca1fa9 ab4e093 b0e7314 ab4e093 cca1fa9 ab4e093 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 |
"""
Multi-Modal Knowledge Distillation Web Application
A FastAPI-based web application for creating new AI models through knowledge distillation
from multiple pre-trained models across different modalities.
"""
import os
import asyncio
import logging
import uuid
from typing import List, Dict, Any, Optional, Union
from pathlib import Path
import json
import shutil
from datetime import datetime
from fastapi import FastAPI, File, UploadFile, Form, HTTPException, BackgroundTasks, WebSocket, WebSocketDisconnect, Request
from fastapi.staticfiles import StaticFiles
from fastapi.templating import Jinja2Templates
from fastapi.responses import HTMLResponse, FileResponse, JSONResponse
from fastapi.middleware.cors import CORSMiddleware
from pydantic import BaseModel, Field
import uvicorn
from src.model_loader import ModelLoader
from src.distillation import KnowledgeDistillationTrainer
from src.utils import setup_logging, validate_file, cleanup_temp_files, get_system_info
# Import new core components
from src.core.memory_manager import AdvancedMemoryManager
from src.core.chunk_loader import AdvancedChunkLoader
from src.core.cpu_optimizer import CPUOptimizer
from src.core.token_manager import TokenManager
# Import medical components
from src.medical.medical_datasets import MedicalDatasetManager
from src.medical.dicom_handler import DicomHandler
from src.medical.medical_preprocessing import MedicalPreprocessor
# Import database components
from database.database import DatabaseManager
from src.database_manager import DatabaseManager as PlatformDatabaseManager
from src.models_manager import ModelsManager
# Setup logging with error handling
try:
setup_logging()
logger = logging.getLogger(__name__)
except Exception as e:
# Fallback to basic logging if setup fails
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)
logger.warning(f"Failed to setup advanced logging: {e}")
# Custom JSON encoder for handling Path objects and other non-serializable types
class CustomJSONEncoder(json.JSONEncoder):
def default(self, obj):
if isinstance(obj, Path):
return str(obj)
elif hasattr(obj, '__dict__'):
return obj.__dict__
elif hasattr(obj, 'tolist'): # For numpy arrays
return obj.tolist()
elif hasattr(obj, 'detach'): # For PyTorch tensors
return obj.detach().cpu().numpy().tolist()
return super().default(obj)
def safe_json_serialize(data):
"""Safely serialize data to JSON, handling non-serializable objects"""
try:
return json.loads(json.dumps(data, cls=CustomJSONEncoder))
except Exception as e:
logger.warning(f"Failed to serialize data: {e}")
# Return a safe version
if isinstance(data, dict):
safe_data = {}
for k, v in data.items():
try:
json.dumps(v, cls=CustomJSONEncoder)
safe_data[k] = v
except:
safe_data[k] = str(v)
return safe_data
else:
return str(data)
def cleanup_training_session(session_id: str):
"""Clean up training session resources"""
try:
if session_id in training_sessions:
session = training_sessions[session_id]
# Clean up any temporary files
model_path = session.get("model_path")
if model_path and Path(model_path).exists():
try:
shutil.rmtree(model_path)
logger.info(f"Cleaned up model files for session {session_id}")
except Exception as e:
logger.warning(f"Failed to clean up model files: {e}")
# Remove from active sessions
del training_sessions[session_id]
# Remove WebSocket connection if exists
if session_id in active_connections:
del active_connections[session_id]
logger.info(f"Cleaned up training session: {session_id}")
except Exception as e:
logger.error(f"Error cleaning up session {session_id}: {e}")
def cleanup_old_sessions():
"""Clean up old completed or failed sessions"""
try:
current_time = datetime.now().timestamp()
sessions_to_remove = []
for session_id, session in training_sessions.items():
session_status = session.get("status", "unknown")
end_time = session.get("end_time")
# Remove sessions older than 1 hour if completed/failed
if session_status in ["completed", "failed", "cancelled"] and end_time:
if current_time - end_time > 3600: # 1 hour
sessions_to_remove.append(session_id)
for session_id in sessions_to_remove:
cleanup_training_session(session_id)
logger.info(f"Auto-cleaned old session: {session_id}")
except Exception as e:
logger.error(f"Error during automatic cleanup: {e}")
# Initialize FastAPI app
app = FastAPI(
title="Multi-Modal Knowledge Distillation",
description="Create new AI models through knowledge distillation from multiple pre-trained models",
version="2.1.0",
docs_url="/docs",
redoc_url="/redoc"
)
# Add CORS middleware
app.add_middleware(
CORSMiddleware,
allow_origins=["*"],
allow_credentials=True,
allow_methods=["*"],
allow_headers=["*"],
)
# Mount static files and templates
app.mount("/static", StaticFiles(directory="static"), name="static")
templates = Jinja2Templates(directory="templates")
# Global variables for tracking training sessions
training_sessions: Dict[str, Dict[str, Any]] = {}
active_connections: Dict[str, WebSocket] = {}
# Startup event to clean old sessions
@app.on_event("startup")
async def startup_event():
"""Initialize application and clean up old sessions"""
try:
logger.info("Starting Multi-Modal Knowledge Distillation Platform")
# Clean up any old sessions from previous runs
cleanup_old_sessions()
# Initialize core components
logger.info("Initializing core components...")
# Log system information
system_info = get_system_info()
logger.info(f"System Info: {system_info}")
logger.info("Application startup completed successfully")
except Exception as e:
logger.error(f"Error during startup: {e}")
# Shutdown event to clean up resources
@app.on_event("shutdown")
async def shutdown_event():
"""Clean up resources on shutdown"""
try:
logger.info("Shutting down application...")
# Clean up all active sessions
for session_id in list(training_sessions.keys()):
cleanup_training_session(session_id)
# Clean up temporary files
cleanup_temp_files()
logger.info("Application shutdown completed")
except Exception as e:
logger.error(f"Error during shutdown: {e}")
# Pydantic models for API
class TrainingConfig(BaseModel):
session_id: str = Field(..., description="Unique session identifier")
teacher_models: List[Union[str, Dict[str, Any]]] = Field(..., description="List of teacher model paths/URLs or model configs")
student_config: Dict[str, Any] = Field(default_factory=dict, description="Student model configuration")
training_params: Dict[str, Any] = Field(default_factory=dict, description="Training parameters")
distillation_strategy: str = Field(default="ensemble", description="Distillation strategy")
hf_token: Optional[str] = Field(default=None, description="Hugging Face token")
trust_remote_code: bool = Field(default=False, description="Trust remote code execution")
existing_student_model: Optional[str] = Field(default=None, description="Path to existing trained student model for retraining")
incremental_training: bool = Field(default=False, description="Whether this is incremental training")
class TrainingStatus(BaseModel):
session_id: str
status: str
progress: float
current_step: int
total_steps: int
loss: Optional[float] = None
eta: Optional[str] = None
message: str = ""
class ModelInfo(BaseModel):
name: str
size: int
format: str
modality: str
architecture: Optional[str] = None
class DatabaseInfo(BaseModel):
name: str
name_ar: Optional[str] = ""
dataset_id: str
category: str = "general"
description: str = ""
description_ar: Optional[str] = ""
size: Optional[str] = "Unknown"
language: Optional[str] = "Unknown"
modality: str = "text"
license: Optional[str] = "Unknown"
class DatabaseSearchRequest(BaseModel):
query: str
limit: int = 20
category: Optional[str] = None
class DatabaseSelectionRequest(BaseModel):
database_ids: List[str]
class ModelSearchRequest(BaseModel):
query: str
limit: int = 20
model_type: Optional[str] = None
class ModelSelectionRequest(BaseModel):
teacher_models: List[str] = []
student_model: Optional[str] = None
# Initialize components
model_loader = ModelLoader()
distillation_trainer = KnowledgeDistillationTrainer()
# Initialize new advanced components
memory_manager = AdvancedMemoryManager(max_memory_gb=14.0) # 14GB for 16GB systems
chunk_loader = AdvancedChunkLoader(memory_manager)
cpu_optimizer = CPUOptimizer(memory_manager)
token_manager = TokenManager()
# Initialize database manager
platform_db_manager = PlatformDatabaseManager()
# Initialize models manager
models_manager = ModelsManager()
database_manager = DatabaseManager()
# Initialize medical components
medical_dataset_manager = MedicalDatasetManager(memory_manager)
dicom_handler = DicomHandler(memory_limit_mb=1000.0)
medical_preprocessor = MedicalPreprocessor()
@app.on_event("startup")
async def startup_event():
"""Initialize application on startup"""
logger.info("Starting Multi-Modal Knowledge Distillation application")
# Create necessary directories with error handling
for directory in ["uploads", "models", "temp", "logs"]:
try:
Path(directory).mkdir(exist_ok=True)
logger.info(f"Created/verified directory: {directory}")
except PermissionError:
logger.warning(f"Cannot create directory {directory}, using temp directory")
except Exception as e:
logger.warning(f"Error creating directory {directory}: {e}")
# Log system information
try:
system_info = get_system_info()
logger.info(f"System info: {system_info}")
except Exception as e:
logger.warning(f"Could not get system info: {e}")
@app.on_event("shutdown")
async def shutdown_event():
"""Cleanup on application shutdown"""
logger.info("Shutting down application")
cleanup_temp_files()
@app.get("/", response_class=HTMLResponse)
async def read_root():
"""Serve the main web interface"""
return templates.TemplateResponse("index.html", {"request": {}})
@app.get("/health")
async def health_check():
"""Health check endpoint for Docker and monitoring"""
try:
# Get system information
memory_info = memory_manager.get_memory_info()
# Check if default token is available
default_token = token_manager.get_token()
return {
"status": "healthy",
"version": "2.0.0",
"timestamp": datetime.now().isoformat(),
"memory": {
"usage_percent": memory_info.get("process_memory_percent", 0),
"available_gb": memory_info.get("system_memory_available_gb", 0),
"status": memory_manager.check_memory_status()
},
"tokens": {
"default_available": bool(default_token),
"total_tokens": len(token_manager.list_tokens())
},
"features": {
"memory_management": True,
"chunk_loading": True,
"cpu_optimization": True,
"medical_datasets": True,
"token_management": True
},
"system_info": get_system_info()
}
except Exception as e:
logger.error(f"Health check failed: {e}")
return {
"status": "unhealthy",
"error": str(e),
"timestamp": datetime.now().isoformat(),
"version": "2.0.0"
}
@app.get("/test-token")
async def test_token():
"""Test if HF token is working"""
hf_token = (
os.getenv('HF_TOKEN') or
os.getenv('HUGGINGFACE_TOKEN') or
os.getenv('HUGGINGFACE_HUB_TOKEN')
)
if not hf_token:
return {
"token_available": False,
"message": "No HF token found in environment variables"
}
try:
# Test token by trying to access a gated model's config
from transformers import AutoConfig
config = AutoConfig.from_pretrained("google/gemma-2b", token=hf_token)
return {
"token_available": True,
"token_valid": True,
"message": "Token is working correctly"
}
except Exception as e:
return {
"token_available": True,
"token_valid": False,
"message": f"Token validation failed: {str(e)}"
}
@app.post("/test-model")
async def test_model_loading(request: Dict[str, Any]):
"""Test loading a specific model"""
try:
model_path = request.get('model_path')
trust_remote_code = request.get('trust_remote_code', False)
if not model_path:
return {"success": False, "error": "model_path is required"}
# Get appropriate token based on access type
access_type = request.get('access_type', 'read')
hf_token = request.get('token')
if not hf_token or hf_token == 'auto':
# Get appropriate token for the access type
hf_token = token_manager.get_token_for_task(access_type)
if hf_token:
logger.info(f"Using {access_type} token for model testing")
else:
logger.warning(f"No suitable token found for {access_type} access")
# Fallback to environment variables
hf_token = (
os.getenv('HF_TOKEN') or
os.getenv('HUGGINGFACE_TOKEN') or
os.getenv('HUGGINGFACE_HUB_TOKEN')
)
# Test model loading
model_info = await model_loader.get_model_info(model_path)
return {
"success": True,
"model_info": model_info,
"message": f"Model {model_path} can be loaded"
}
except Exception as e:
error_msg = str(e)
suggestions = []
if 'trust_remote_code' in error_msg.lower():
suggestions.append("فعّل 'Trust Remote Code' للنماذج التي تتطلب كود مخصص")
elif 'gated' in error_msg.lower():
suggestions.append("النموذج يتطلب إذن وصول خاص - استخدم رمز مخصص")
elif 'siglip' in error_msg.lower():
suggestions.append("جرب تفعيل 'Trust Remote Code' لنماذج SigLIP")
elif '401' in error_msg or 'authentication' in error_msg.lower():
suggestions.append("تحقق من رمز Hugging Face الخاص بك")
suggestions.append("تأكد من أن الرمز له صلاحية الوصول لهذا النموذج")
elif '404' in error_msg or 'not found' in error_msg.lower():
suggestions.append("تحقق من اسم مستودع النموذج")
suggestions.append("تأكد من وجود النموذج على Hugging Face")
return {
"success": False,
"error": error_msg,
"suggestions": suggestions
}
@app.post("/upload", response_model=Dict[str, Any])
async def upload_model(
background_tasks: BackgroundTasks,
files: List[UploadFile] = File(...),
model_names: List[str] = Form(...)
):
"""Upload model files"""
try:
uploaded_models = []
for file, name in zip(files, model_names):
# Validate file
validation_result = validate_file(file)
if not validation_result["valid"]:
raise HTTPException(status_code=400, detail=validation_result["error"])
# Generate unique filename
file_id = str(uuid.uuid4())
file_extension = Path(file.filename).suffix
safe_filename = f"{file_id}{file_extension}"
file_path = Path("uploads") / safe_filename
# Save file
with open(file_path, "wb") as buffer:
content = await file.read()
buffer.write(content)
# Get model info
model_info = await model_loader.get_model_info(str(file_path))
uploaded_models.append({
"id": file_id,
"name": name,
"filename": file.filename,
"path": str(file_path),
"size": len(content),
"info": model_info
})
logger.info(f"Uploaded model: {name} ({file.filename})")
# Schedule cleanup of old files
background_tasks.add_task(cleanup_temp_files, max_age_hours=24)
return {
"success": True,
"models": uploaded_models,
"message": f"Successfully uploaded {len(uploaded_models)} model(s)"
}
except Exception as e:
logger.error(f"Error uploading models: {str(e)}")
raise HTTPException(status_code=500, detail=str(e))
@app.post("/start-training", response_model=Dict[str, Any])
async def start_training(
background_tasks: BackgroundTasks,
config: TrainingConfig
):
"""Start knowledge distillation training"""
try:
session_id = config.session_id
# Handle existing sessions
if session_id in training_sessions:
existing_session = training_sessions[session_id]
existing_status = existing_session.get("status", "unknown")
# Allow restarting failed or completed sessions
if existing_status in ["failed", "completed", "cancelled"]:
logger.info(f"Restarting session {session_id} (previous status: {existing_status})")
# Clean up old session
cleanup_training_session(session_id)
elif existing_status in ["running", "initializing"]:
raise HTTPException(
status_code=400,
detail=f"Training session already running (status: {existing_status})"
)
else:
# Unknown status, clean up and restart
logger.warning(f"Unknown session status {existing_status}, cleaning up")
cleanup_training_session(session_id)
# Set HF token from environment if available
hf_token = os.getenv('HF_TOKEN') or os.getenv('HUGGINGFACE_TOKEN')
if hf_token:
os.environ['HF_TOKEN'] = hf_token
logger.info("Using Hugging Face token from environment")
# Check for large models and warn
large_models = []
for model_info in config.teacher_models:
model_path = model_info if isinstance(model_info, str) else model_info.get('path', '')
if any(size_indicator in model_path.lower() for size_indicator in ['27b', '70b', '13b']):
large_models.append(model_path)
# Initialize training session with safe config serialization
safe_config = safe_json_serialize(config.dict())
training_sessions[session_id] = {
"status": "initializing",
"progress": 0.0,
"current_step": 0,
"total_steps": config.training_params.get("max_steps", 1000),
"config": safe_config,
"start_time": None,
"end_time": None,
"model_path": None,
"logs": [],
"large_models": large_models,
"message": "Initializing training session..." + (
f" (Large models detected: {', '.join(large_models)})" if large_models else ""
)
}
# Start training in background
background_tasks.add_task(run_training, session_id, config)
logger.info(f"Started training session: {session_id}")
return {
"success": True,
"session_id": session_id,
"message": "Training started successfully"
}
except Exception as e:
logger.error(f"Error starting training: {str(e)}")
raise HTTPException(status_code=500, detail=str(e))
async def run_training(session_id: str, config: TrainingConfig):
"""Run knowledge distillation training in background"""
try:
session = training_sessions[session_id]
session["status"] = "running"
session["start_time"] = asyncio.get_event_loop().time()
# Set timeout for the entire operation (30 minutes)
timeout_seconds = 30 * 60
# Set HF token for this session - prioritize config token
config_token = getattr(config, 'hf_token', None)
env_token = (
os.getenv('HF_TOKEN') or
os.getenv('HUGGINGFACE_TOKEN') or
os.getenv('HUGGINGFACE_HUB_TOKEN')
)
hf_token = config_token or env_token
if hf_token:
logger.info(f"Using Hugging Face token from {'config' if config_token else 'environment'}")
# Set token in environment for this session
os.environ['HF_TOKEN'] = hf_token
else:
logger.warning("No Hugging Face token found - private models may fail")
# Handle existing student model for incremental training
existing_student = None
if config.existing_student_model and config.incremental_training:
try:
await update_training_status(session_id, "loading_student", 0.05, "Loading existing student model...")
# Determine student source and load accordingly
student_source = getattr(config, 'student_source', 'local')
student_path = config.existing_student_model
if student_source == 'huggingface' or ('/' in student_path and not Path(student_path).exists()):
logger.info(f"Loading student model from Hugging Face: {student_path}")
existing_student = await model_loader.load_trained_student(student_path)
elif student_source == 'space':
logger.info(f"Loading student model from Hugging Face Space: {student_path}")
# For spaces, we'll try to load from the space's models directory
space_model_path = f"spaces/{student_path}/models"
existing_student = await model_loader.load_trained_student_from_space(student_path)
else:
logger.info(f"Loading student model from local path: {student_path}")
existing_student = await model_loader.load_trained_student(student_path)
logger.info(f"Successfully loaded existing student model: {existing_student.get('type', 'unknown')}")
# Merge original teachers with new teachers
original_teachers = existing_student.get('original_teachers', [])
new_teachers = [
model_info if isinstance(model_info, str) else model_info.get('path', '')
for model_info in config.teacher_models
]
# Combine teachers (avoid duplicates)
all_teachers = original_teachers.copy()
for teacher in new_teachers:
if teacher not in all_teachers:
all_teachers.append(teacher)
logger.info(f"Incremental training: Original teachers: {original_teachers}")
logger.info(f"Incremental training: New teachers: {new_teachers}")
logger.info(f"Incremental training: All teachers: {all_teachers}")
# Update config with all teachers
config.teacher_models = all_teachers
except Exception as e:
logger.error(f"Error loading existing student model: {e}")
await update_training_status(session_id, "failed", session.get("progress", 0), f"Failed to load existing student: {str(e)}")
return
# Load teacher models
await update_training_status(session_id, "loading_models", 0.1, "Loading teacher models...")
teacher_models = []
trust_remote_code = config.training_params.get('trust_remote_code', False)
total_models = len(config.teacher_models)
for i, model_info in enumerate(config.teacher_models):
try:
# Handle both old format (string) and new format (dict)
if isinstance(model_info, str):
model_path = model_info
model_token = hf_token
model_trust_code = trust_remote_code
else:
model_path = model_info.get('path', model_info)
model_token = model_info.get('token') or hf_token
model_trust_code = model_info.get('trust_remote_code', trust_remote_code)
# Update progress
progress = 0.1 + (i * 0.3 / total_models) # 0.1 to 0.4
await update_training_status(
session_id,
"loading_models",
progress,
f"Loading model {i+1}/{total_models}: {model_path}..."
)
logger.info(f"Loading model {model_path} with trust_remote_code={model_trust_code}")
# Special handling for known problematic models
if model_path == 'Wan-AI/Wan2.2-TI2V-5B':
logger.info(f"Detected ti2v model {model_path}, forcing trust_remote_code=True")
model_trust_code = True
elif model_path == 'deepseek-ai/DeepSeek-V3.1-Base':
logger.warning(f"Skipping {model_path}: Requires GPU with FP8 quantization support")
await update_training_status(
session_id,
"loading_models",
progress,
f"Skipping {model_path}: Requires GPU with FP8 quantization"
)
continue
model = await model_loader.load_model(
model_path,
token=model_token,
trust_remote_code=model_trust_code
)
teacher_models.append(model)
logger.info(f"Successfully loaded model: {model_path}")
# Update progress after successful load
progress = 0.1 + ((i + 1) * 0.3 / total_models)
await update_training_status(
session_id,
"loading_models",
progress,
f"Loaded {i+1}/{total_models} models successfully"
)
except Exception as e:
error_msg = f"Failed to load model {model_path}: {str(e)}"
logger.error(error_msg)
# Provide helpful suggestions based on the error
suggestions = []
error_str = str(e).lower()
# Check if we should retry with trust_remote_code=True
if not model_trust_code and ('ti2v' in error_str or 'does not recognize this architecture' in error_str):
try:
logger.info(f"Retrying {model_path} with trust_remote_code=True")
await update_training_status(
session_id,
"loading_models",
progress,
f"Retrying {model_path} with trust_remote_code=True..."
)
model = await model_loader.load_model(
model_path,
token=model_token,
trust_remote_code=True
)
teacher_models.append(model)
logger.info(f"Successfully loaded model on retry: {model_path}")
# Update progress after successful retry
progress = 0.1 + ((i + 1) * 0.3 / total_models)
await update_training_status(
session_id,
"loading_models",
progress,
f"Loaded {i+1}/{total_models} models successfully (retry)"
)
continue
except Exception as retry_e:
logger.error(f"Retry also failed for {model_path}: {str(retry_e)}")
error_msg = f"Failed even with trust_remote_code=True: {str(retry_e)}"
if 'trust_remote_code' in error_str:
suggestions.append("Try enabling 'Trust Remote Code' option")
elif 'gated' in error_str or 'access' in error_str:
suggestions.append("This model requires access permission and a valid HF token")
elif 'siglip' in error_str or 'unknown' in error_str:
suggestions.append("This model may require special loading. Try enabling 'Trust Remote Code'")
elif 'connection' in error_str or 'network' in error_str:
suggestions.append("Check your internet connection")
elif 'ti2v' in error_str:
suggestions.append("This ti2v model requires trust_remote_code=True")
if suggestions:
error_msg += f". Suggestions: {'; '.join(suggestions)}"
await update_training_status(session_id, "failed", session.get("progress", 0), error_msg)
return
# Initialize student model
await update_training_status(session_id, "initializing_student", 0.2, "Initializing student model...")
student_model = await distillation_trainer.create_student_model(
teacher_models, config.student_config
)
# Run distillation training
await update_training_status(session_id, "training", 0.3, "Starting knowledge distillation...")
async def progress_callback(step: int, total_steps: int, loss: float, metrics: Dict[str, Any]):
progress = 0.3 + (step / total_steps) * 0.6 # 30% to 90%
await update_training_status(
session_id, "training", progress,
f"Training step {step}/{total_steps}, Loss: {loss:.4f}",
current_step=step, loss=loss
)
trained_model = await distillation_trainer.train(
student_model, teacher_models, config.training_params, progress_callback
)
# Save trained model with metadata
await update_training_status(session_id, "saving", 0.9, "Saving trained model...")
# Create model directory with proper structure
model_dir = Path("models") / f"distilled_model_{session_id}"
model_dir.mkdir(parents=True, exist_ok=True)
model_path = model_dir / "pytorch_model.safetensors"
# Prepare training metadata for saving
training_metadata = {
'session_id': session_id,
'teacher_models': [
model_info if isinstance(model_info, str) else model_info.get('path', '')
for model_info in config.teacher_models
],
'strategy': config.distillation_strategy,
'training_params': config.training_params,
'incremental_training': config.incremental_training,
'existing_student_model': config.existing_student_model
}
await distillation_trainer.save_model(trained_model, str(model_path), training_metadata)
# Complete training
session["status"] = "completed"
session["progress"] = 1.0
session["end_time"] = asyncio.get_event_loop().time()
session["model_path"] = model_path
session["training_metadata"] = training_metadata
await update_training_status(session_id, "completed", 1.0, "Training completed successfully!")
logger.info(f"Training session {session_id} completed successfully")
except Exception as e:
logger.error(f"Training session {session_id} failed: {str(e)}")
session = training_sessions.get(session_id, {})
session["status"] = "failed"
session["error"] = str(e)
await update_training_status(session_id, "failed", session.get("progress", 0), f"Training failed: {str(e)}")
async def update_training_status(
session_id: str,
status: str,
progress: float,
message: str,
current_step: int = None,
loss: float = None
):
"""Update training status and notify connected clients"""
if session_id in training_sessions:
session = training_sessions[session_id]
session["status"] = status
session["progress"] = progress
session["message"] = message
if current_step is not None:
session["current_step"] = current_step
if loss is not None:
session["loss"] = loss
# Calculate ETA
if session.get("start_time") and progress > 0:
elapsed = asyncio.get_event_loop().time() - session["start_time"]
if progress < 1.0:
eta_seconds = (elapsed / progress) * (1.0 - progress)
eta = f"{int(eta_seconds // 60)}m {int(eta_seconds % 60)}s"
session["eta"] = eta
# Notify WebSocket clients
if session_id in active_connections:
try:
# Safely serialize session data
safe_session_data = safe_json_serialize(session)
await active_connections[session_id].send_json({
"type": "training_update",
"data": safe_session_data
})
except Exception as e:
logger.warning(f"Failed to send WebSocket update: {e}")
# Remove disconnected client
if session_id in active_connections:
del active_connections[session_id]
@app.get("/progress/{session_id}", response_model=TrainingStatus)
async def get_training_progress(session_id: str):
"""Get training progress for a session"""
if session_id not in training_sessions:
raise HTTPException(status_code=404, detail="Training session not found")
session = training_sessions[session_id]
return TrainingStatus(
session_id=session_id,
status=session["status"],
progress=session["progress"],
current_step=session["current_step"],
total_steps=session["total_steps"],
loss=session.get("loss"),
eta=session.get("eta"),
message=session.get("message", "")
)
@app.get("/download/{session_id}")
async def download_model(session_id: str):
"""Download trained model"""
try:
if session_id not in training_sessions:
raise HTTPException(status_code=404, detail="Training session not found")
session = training_sessions[session_id]
if session["status"] != "completed":
raise HTTPException(status_code=400, detail="Training not completed")
model_path = session.get("model_path")
if not model_path:
# Try to find model in models directory
models_dir = Path("models")
possible_paths = [
models_dir / f"distilled_model_{session_id}",
models_dir / f"distilled_model_{session_id}.safetensors",
models_dir / f"model_{session_id}",
models_dir / f"student_model_{session_id}"
]
for path in possible_paths:
if path.exists():
model_path = str(path)
break
if not model_path or not Path(model_path).exists():
raise HTTPException(status_code=404, detail="Model file not found. The model may not have been saved properly.")
# Create a zip file with all model files
import zipfile
import tempfile
model_dir = Path(model_path)
if model_dir.is_file():
# Single file
return FileResponse(
model_path,
media_type="application/octet-stream",
filename=f"distilled_model_{session_id}.safetensors"
)
else:
# Directory with multiple files
temp_zip = tempfile.NamedTemporaryFile(delete=False, suffix='.zip')
with zipfile.ZipFile(temp_zip.name, 'w') as zipf:
for file_path in model_dir.rglob('*'):
if file_path.is_file():
zipf.write(file_path, file_path.relative_to(model_dir))
return FileResponse(
temp_zip.name,
media_type="application/zip",
filename=f"distilled_model_{session_id}.zip"
)
except Exception as e:
logger.error(f"Error downloading model: {e}")
raise HTTPException(status_code=500, detail=f"Download failed: {str(e)}")
@app.post("/upload-to-hf/{session_id}")
async def upload_to_huggingface(
session_id: str,
repo_name: str = Form(...),
description: str = Form(""),
private: bool = Form(False),
hf_token: str = Form(...)
):
"""Upload trained model to Hugging Face Hub"""
try:
if session_id not in training_sessions:
raise HTTPException(status_code=404, detail="Training session not found")
session = training_sessions[session_id]
if session["status"] != "completed":
raise HTTPException(status_code=400, detail="Training not completed")
model_path = session.get("model_path")
if not model_path or not Path(model_path).exists():
raise HTTPException(status_code=404, detail="Model file not found")
# Import huggingface_hub
try:
from huggingface_hub import HfApi, create_repo
except ImportError:
raise HTTPException(status_code=500, detail="huggingface_hub not installed")
# Initialize HF API
api = HfApi(token=hf_token)
# Validate repository name format
if '/' not in repo_name:
raise HTTPException(status_code=400, detail="Repository name must be in format 'username/model-name'")
username, model_name = repo_name.split('/', 1)
# Create repository with better error handling
try:
repo_url = create_repo(
repo_id=repo_name,
token=hf_token,
private=private,
exist_ok=True
)
logger.info(f"Created/accessed repository: {repo_url}")
except Exception as e:
error_msg = str(e)
if "403" in error_msg or "Forbidden" in error_msg:
raise HTTPException(
status_code=403,
detail=f"Permission denied. Please check: 1) Your token has 'Write' permissions, 2) You own the namespace '{username}', 3) The repository name is correct. Error: {error_msg}"
)
elif "401" in error_msg or "Unauthorized" in error_msg:
raise HTTPException(
status_code=401,
detail=f"Invalid token. Please check your Hugging Face token. Error: {error_msg}"
)
else:
raise HTTPException(status_code=400, detail=f"Failed to create repository: {error_msg}")
# Upload model files
model_path_obj = Path(model_path)
uploaded_files = []
# Determine the model directory
if model_path_obj.is_file():
model_dir = model_path_obj.parent
else:
model_dir = model_path_obj
# Upload all files in the model directory
essential_files = [
'pytorch_model.safetensors', 'config.json', 'model.py',
'training_history.json', 'README.md'
]
# Upload essential files first
for file_name in essential_files:
file_path = model_dir / file_name
if file_path.exists():
try:
api.upload_file(
path_or_fileobj=str(file_path),
path_in_repo=file_name,
repo_id=repo_name,
token=hf_token
)
uploaded_files.append(file_name)
logger.info(f"Uploaded {file_name}")
except Exception as e:
logger.warning(f"Failed to upload {file_name}: {e}")
# Upload any additional files
for file_path in model_dir.rglob('*'):
if file_path.is_file() and file_path.name not in essential_files:
try:
relative_path = file_path.relative_to(model_dir)
api.upload_file(
path_or_fileobj=str(file_path),
path_in_repo=str(relative_path),
repo_id=repo_name,
token=hf_token
)
uploaded_files.append(str(relative_path))
logger.info(f"Uploaded additional file: {relative_path}")
except Exception as e:
logger.warning(f"Failed to upload {relative_path}: {e}")
# Create README.md
config_info = session.get("config", {})
teacher_models_raw = config_info.get("teacher_models", [])
# Extract model paths from teacher_models (handle both string and dict formats)
teacher_models = []
for model in teacher_models_raw:
if isinstance(model, str):
teacher_models.append(model)
elif isinstance(model, dict):
teacher_models.append(model.get('path', str(model)))
else:
teacher_models.append(str(model))
readme_content = f"""---
license: apache-2.0
tags:
- knowledge-distillation
- pytorch
- transformers
base_model: {teacher_models[0] if teacher_models else 'unknown'}
---
# {repo_name}
This model was created using knowledge distillation from the following teacher model(s):
{chr(10).join([f"- {model}" for model in teacher_models])}
## Model Description
{description if description else 'A distilled model created using multi-modal knowledge distillation.'}
## Training Details
- **Teacher Models**: {', '.join(teacher_models)}
- **Distillation Strategy**: {config_info.get('distillation_strategy', 'ensemble')}
- **Training Steps**: {config_info.get('training_params', {}).get('max_steps', 'unknown')}
- **Learning Rate**: {config_info.get('training_params', {}).get('learning_rate', 'unknown')}
## Usage
```python
from transformers import AutoModel, AutoTokenizer
model = AutoModel.from_pretrained("{repo_name}")
tokenizer = AutoTokenizer.from_pretrained("{teacher_models[0] if teacher_models else 'bert-base-uncased'}")
```
## Created with
This model was created using the Multi-Modal Knowledge Distillation platform.
"""
# Upload README
api.upload_file(
path_or_fileobj=readme_content.encode(),
path_in_repo="README.md",
repo_id=repo_name,
token=hf_token
)
uploaded_files.append("README.md")
return {
"success": True,
"repo_url": f"https://huggingface.co/{repo_name}",
"uploaded_files": uploaded_files,
"message": f"Model successfully uploaded to {repo_name}"
}
except Exception as e:
logger.error(f"Error uploading to Hugging Face: {e}")
raise HTTPException(status_code=500, detail=f"Upload failed: {str(e)}")
@app.post("/validate-repo-name")
async def validate_repo_name(request: Dict[str, Any]):
"""Validate repository name and check permissions"""
try:
repo_name = request.get('repo_name', '').strip()
hf_token = request.get('hf_token', '').strip()
if not repo_name or not hf_token:
return {"valid": False, "error": "Repository name and token are required"}
if '/' not in repo_name:
return {"valid": False, "error": "Repository name must be in format 'username/model-name'"}
username, model_name = repo_name.split('/', 1)
# Check if username matches token owner
try:
from huggingface_hub import HfApi
api = HfApi(token=hf_token)
# Try to get user info
user_info = api.whoami()
token_username = user_info.get('name', '')
if username != token_username:
return {
"valid": False,
"error": f"Username mismatch. Token belongs to '{token_username}' but trying to create repo under '{username}'. Use '{token_username}/{model_name}' instead.",
"suggested_name": f"{token_username}/{model_name}"
}
return {
"valid": True,
"message": f"Repository name '{repo_name}' is valid for your account",
"username": token_username
}
except Exception as e:
return {"valid": False, "error": f"Token validation failed: {str(e)}"}
except Exception as e:
return {"valid": False, "error": f"Validation error: {str(e)}"}
@app.post("/test-space")
async def test_space(request: Dict[str, Any]):
"""Test if a Hugging Face Space exists and has trained models"""
try:
space_name = request.get('space_name', '').strip()
hf_token = request.get('hf_token', '').strip()
if not space_name:
return {"success": False, "error": "Space name is required"}
if '/' not in space_name:
return {"success": False, "error": "Space name must be in format 'username/space-name'"}
try:
from huggingface_hub import HfApi
api = HfApi(token=hf_token if hf_token else None)
# Check if the Space exists
try:
space_info = api.space_info(space_name)
logger.info(f"Found Space: {space_name}")
except Exception as e:
return {"success": False, "error": f"Space not found or not accessible: {str(e)}"}
# Try to list files in the Space to see if it has models
try:
files = api.list_repo_files(space_name, repo_type="space")
model_files = [f for f in files if f.endswith(('.safetensors', '.bin', '.pt'))]
# Check for models directory
models_dir_files = [f for f in files if f.startswith('models/')]
return {
"success": True,
"space_info": {
"name": space_name,
"model_files": model_files,
"models_directory": len(models_dir_files) > 0,
"total_files": len(files)
},
"models": model_files,
"message": f"Space {space_name} is accessible"
}
except Exception as e:
# Space exists but we can't list files (might be private or no access)
return {
"success": True,
"space_info": {"name": space_name},
"models": [],
"message": f"Space {space_name} exists but file listing not available (might be private)"
}
except Exception as e:
return {"success": False, "error": f"Error accessing Hugging Face: {str(e)}"}
except Exception as e:
logger.error(f"Error testing Space: {e}")
return {"success": False, "error": f"Test failed: {str(e)}"}
@app.get("/trained-students")
async def list_trained_students():
"""List available trained student models for retraining"""
try:
models_dir = Path("models")
trained_students = []
if models_dir.exists():
for model_dir in models_dir.iterdir():
if model_dir.is_dir():
try:
# Check if it's a trained student model
config_files = list(model_dir.glob("*config.json"))
history_files = list(model_dir.glob("*training_history.json"))
if config_files:
with open(config_files[0], 'r') as f:
config = json.load(f)
if config.get('is_student_model', False):
history = {}
if history_files:
with open(history_files[0], 'r') as f:
history = json.load(f)
model_info = {
"id": model_dir.name,
"name": model_dir.name,
"path": str(model_dir),
"type": "trained_student",
"created_at": config.get('created_at', 'unknown'),
"architecture": config.get('architecture', 'unknown'),
"modalities": config.get('modalities', ['text']),
"can_be_retrained": config.get('can_be_retrained', True),
"original_teachers": history.get('retraining_info', {}).get('original_teachers', []),
"training_sessions": len(history.get('training_sessions', [])),
"last_training": history.get('training_sessions', [{}])[-1].get('timestamp', 'unknown') if history.get('training_sessions') else 'unknown'
}
trained_students.append(model_info)
except Exception as e:
logger.warning(f"Error reading model {model_dir}: {e}")
continue
return {"trained_students": trained_students}
except Exception as e:
logger.error(f"Error listing trained students: {e}")
raise HTTPException(status_code=500, detail=str(e))
@app.get("/models", response_model=List[ModelInfo])
async def list_models():
"""List available models"""
models = []
# List uploaded models
uploads_dir = Path("uploads")
if uploads_dir.exists():
for file_path in uploads_dir.iterdir():
if file_path.is_file():
try:
info = await model_loader.get_model_info(str(file_path))
models.append(ModelInfo(
name=file_path.stem,
size=file_path.stat().st_size,
format=file_path.suffix[1:],
modality=info.get("modality", "unknown"),
architecture=info.get("architecture")
))
except Exception as e:
logger.warning(f"Error getting info for {file_path}: {e}")
return models
@app.websocket("/ws/{session_id}")
async def websocket_endpoint(websocket: WebSocket, session_id: str):
"""WebSocket endpoint for real-time training updates"""
await websocket.accept()
active_connections[session_id] = websocket
try:
# Send current status if session exists
if session_id in training_sessions:
await websocket.send_json({
"type": "training_update",
"data": training_sessions[session_id]
})
# Keep connection alive
while True:
await websocket.receive_text()
except WebSocketDisconnect:
if session_id in active_connections:
del active_connections[session_id]
except Exception as e:
logger.error(f"WebSocket error for session {session_id}: {e}")
if session_id in active_connections:
del active_connections[session_id]
# ==================== NEW ADVANCED ENDPOINTS ====================
# Token Management Endpoints
@app.get("/tokens")
async def token_management_page(request: Request):
"""Token management page"""
return templates.TemplateResponse("token-management.html", {"request": request})
@app.post("/api/tokens")
async def save_token(
name: str = Form(...),
token: str = Form(...),
token_type: str = Form("read"),
description: str = Form(""),
is_default: bool = Form(False)
):
"""Save HF token"""
try:
success = token_manager.save_token(name, token, token_type, description, is_default)
if success:
return {"success": True, "message": f"Token '{name}' saved successfully"}
else:
raise HTTPException(status_code=400, detail="Failed to save token")
except Exception as e:
logger.error(f"Error saving token: {e}")
raise HTTPException(status_code=500, detail=str(e))
@app.get("/api/tokens")
async def list_tokens():
"""List all saved tokens"""
try:
tokens = token_manager.list_tokens()
return {"tokens": tokens}
except Exception as e:
logger.error(f"Error listing tokens: {e}")
raise HTTPException(status_code=500, detail=str(e))
@app.delete("/api/tokens/{token_name}")
async def delete_token(token_name: str):
"""Delete a token"""
try:
success = token_manager.delete_token(token_name)
if success:
return {"success": True, "message": f"Token '{token_name}' deleted"}
else:
raise HTTPException(status_code=404, detail="Token not found")
except Exception as e:
logger.error(f"Error deleting token: {e}")
raise HTTPException(status_code=500, detail=str(e))
@app.post("/api/tokens/{token_name}/set-default")
async def set_default_token(token_name: str):
"""Set token as default"""
try:
success = token_manager.set_default_token(token_name)
if success:
return {"success": True, "message": f"Token '{token_name}' set as default"}
else:
raise HTTPException(status_code=404, detail="Token not found")
except Exception as e:
logger.error(f"Error setting default token: {e}")
raise HTTPException(status_code=500, detail=str(e))
@app.post("/api/tokens/validate")
async def validate_token(token: str = Form(...)):
"""Validate HF token"""
try:
result = token_manager.validate_token(token)
return result
except Exception as e:
logger.error(f"Error validating token: {e}")
raise HTTPException(status_code=500, detail=str(e))
@app.get("/api/tokens/for-task/{task_type}")
async def get_token_for_task(task_type: str):
"""Get appropriate token for specific task"""
try:
# Get token for task
token = token_manager.get_token_for_task(task_type)
if not token:
raise HTTPException(status_code=404, detail=f"No suitable token found for task: {task_type}")
# Get token information
tokens = token_manager.list_tokens()
token_info = None
# Find which token was selected
for t in tokens:
test_token = token_manager.get_token(t['name'])
if test_token == token:
token_info = t
break
if not token_info:
# Token from environment variable
token_info = {
'name': f'{task_type}_token',
'type': task_type,
'description': f'رمز من متغيرات البيئة للمهمة: {task_type}',
'last_used': None,
'usage_count': 0
}
# Get token type information
type_info = token_manager.token_types.get(token_info['type'], {})
return {
"success": True,
"task_type": task_type,
"token_info": {
"token_name": token_info['name'],
"type": token_info['type'],
"type_name": type_info.get('name', token_info['type']),
"description": token_info['description'],
"security_level": type_info.get('security_level', 'medium'),
"recommended_for": type_info.get('recommended_for', 'general'),
"last_used": token_info.get('last_used'),
"usage_count": token_info.get('usage_count', 0)
}
}
except HTTPException:
raise
except Exception as e:
logger.error(f"Error getting token for task {task_type}: {e}")
raise HTTPException(status_code=500, detail=str(e))
# Medical Dataset Endpoints
@app.get("/medical-datasets")
async def medical_datasets_page(request: Request):
"""Medical datasets management page"""
return templates.TemplateResponse("medical-datasets.html", {"request": request})
@app.get("/api/medical-datasets")
async def list_medical_datasets():
"""List supported medical datasets"""
try:
datasets = medical_dataset_manager.list_supported_datasets()
return {"datasets": datasets}
except Exception as e:
logger.error(f"Error listing medical datasets: {e}")
raise HTTPException(status_code=500, detail=str(e))
@app.post("/api/medical-datasets/load")
async def load_medical_dataset(
dataset_name: str = Form(...),
streaming: bool = Form(True),
split: str = Form("train")
):
"""Load medical dataset"""
try:
# Get appropriate token for medical datasets (fine-grained preferred)
hf_token = token_manager.get_token_for_task('medical')
if not hf_token:
logger.warning("No suitable token found for medical datasets, trying default")
hf_token = token_manager.get_token()
dataset_info = await medical_dataset_manager.load_dataset(
dataset_name=dataset_name,
streaming=streaming,
split=split,
token=hf_token
)
return {
"success": True,
"dataset_info": {
"name": dataset_info['config']['name'],
"size_gb": dataset_info['config']['size_gb'],
"num_samples": dataset_info['config']['num_samples'],
"streaming": dataset_info['streaming']
}
}
except Exception as e:
logger.error(f"Error loading medical dataset: {e}")
raise HTTPException(status_code=500, detail=str(e))
# Memory and Performance Endpoints
@app.get("/api/system/memory")
async def get_memory_info():
"""Get current memory information"""
try:
memory_info = memory_manager.get_memory_info()
return memory_info
except Exception as e:
logger.error(f"Error getting memory info: {e}")
raise HTTPException(status_code=500, detail=str(e))
@app.get("/api/system/performance")
async def get_performance_info():
"""Get system performance information"""
try:
memory_info = memory_manager.get_memory_info()
recommendations = memory_manager.get_memory_recommendations()
return {
"memory": memory_info,
"recommendations": recommendations,
"cpu_cores": cpu_optimizer.cpu_count,
"optimizations_applied": cpu_optimizer.optimizations_applied
}
except Exception as e:
logger.error(f"Error getting performance info: {e}")
raise HTTPException(status_code=500, detail=str(e))
@app.post("/api/system/cleanup")
async def force_memory_cleanup():
"""Force memory cleanup"""
try:
memory_manager.force_cleanup()
return {"success": True, "message": "Memory cleanup completed"}
except Exception as e:
logger.error(f"Error during memory cleanup: {e}")
raise HTTPException(status_code=500, detail=str(e))
# Google Models Support
@app.get("/api/models/google")
async def list_google_models():
"""List available Google models"""
try:
google_models = [
{
"name": "google/medsiglip-448",
"description": "Medical SigLIP model for medical image-text understanding",
"type": "vision-language",
"size_gb": 1.1,
"modality": "multimodal",
"medical_specialized": True
},
{
"name": "google/gemma-3n-E4B-it",
"description": "Gemma 3 model for instruction following",
"type": "language",
"size_gb": 8.5,
"modality": "text",
"medical_specialized": False
}
]
return {"models": google_models}
except Exception as e:
logger.error(f"Error listing Google models: {e}")
raise HTTPException(status_code=500, detail=str(e))
# Database Management API Endpoints
@app.get("/api/databases")
async def get_all_databases():
"""Get all configured databases"""
try:
databases = platform_db_manager.get_all_databases()
selected = platform_db_manager.get_selected_databases()
return {
"success": True,
"databases": databases,
"selected": selected,
"total": len(databases)
}
except Exception as e:
logger.error(f"Error getting databases: {e}")
raise HTTPException(status_code=500, detail=str(e))
@app.post("/api/databases/search")
async def search_databases(request: DatabaseSearchRequest):
"""Search for databases on Hugging Face"""
try:
results = await platform_db_manager.search_huggingface_datasets(
query=request.query,
limit=request.limit
)
return {
"success": True,
"results": results,
"count": len(results),
"query": request.query
}
except Exception as e:
logger.error(f"Error searching databases: {e}")
raise HTTPException(status_code=500, detail=str(e))
@app.post("/api/databases/add")
async def add_database(database_info: DatabaseInfo):
"""Add a new database to the configuration"""
try:
success = await platform_db_manager.add_database(database_info.dict())
if success:
return {
"success": True,
"message": f"Database {database_info.dataset_id} added successfully"
}
else:
raise HTTPException(status_code=400, detail="Failed to add database")
except Exception as e:
logger.error(f"Error adding database: {e}")
raise HTTPException(status_code=500, detail=str(e))
@app.post("/api/databases/validate/{dataset_id:path}")
async def validate_database(dataset_id: str):
"""Validate a dataset"""
try:
validation_result = await platform_db_manager.validate_dataset(dataset_id)
return {
"success": True,
"validation": validation_result,
"dataset_id": dataset_id
}
except Exception as e:
logger.error(f"Error validating database: {e}")
raise HTTPException(status_code=500, detail=str(e))
@app.post("/api/databases/select")
async def select_databases(request: DatabaseSelectionRequest):
"""Select databases for use"""
try:
results = []
for database_id in request.database_ids:
success = platform_db_manager.select_database(database_id)
results.append({
"database_id": database_id,
"success": success
})
return {
"success": True,
"results": results,
"selected": platform_db_manager.get_selected_databases()
}
except Exception as e:
logger.error(f"Error selecting databases: {e}")
raise HTTPException(status_code=500, detail=str(e))
@app.delete("/api/databases/{database_id:path}")
async def remove_database(database_id: str):
"""Remove a database from configuration"""
try:
success = platform_db_manager.remove_database(database_id)
if success:
return {
"success": True,
"message": f"Database {database_id} removed successfully"
}
else:
raise HTTPException(status_code=400, detail="Failed to remove database")
except Exception as e:
logger.error(f"Error removing database: {e}")
raise HTTPException(status_code=500, detail=str(e))
@app.get("/api/databases/{database_id:path}")
async def get_database_info(database_id: str):
"""Get detailed information about a specific database"""
try:
database_info = platform_db_manager.get_database_info(database_id)
if database_info:
return {
"success": True,
"database": database_info
}
else:
raise HTTPException(status_code=404, detail="Database not found")
except Exception as e:
logger.error(f"Error getting database info: {e}")
raise HTTPException(status_code=500, detail=str(e))
@app.get("/api/databases/category/{category}")
async def get_databases_by_category(category: str):
"""Get databases filtered by category"""
try:
databases = platform_db_manager.get_databases_by_category(category)
return {
"success": True,
"databases": databases,
"category": category,
"count": len(databases)
}
except Exception as e:
logger.error(f"Error getting databases by category: {e}")
raise HTTPException(status_code=500, detail=str(e))
@app.post("/api/databases/load-selected")
async def load_selected_databases(max_samples: int = 1000):
"""Load data from selected databases"""
try:
loaded_data = await platform_db_manager.load_selected_datasets(max_samples)
return {
"success": True,
"loaded_datasets": loaded_data,
"total_datasets": len(loaded_data)
}
except Exception as e:
logger.error(f"Error loading selected databases: {e}")
raise HTTPException(status_code=500, detail=str(e))
# Models Management API Endpoints
@app.get("/api/models")
async def get_all_models():
"""Get all configured models"""
try:
models = models_manager.get_all_models()
teachers = models_manager.get_selected_teachers()
student = models_manager.get_selected_student()
return {
"success": True,
"models": models,
"selected_teachers": teachers,
"selected_student": student,
"total": len(models)
}
except Exception as e:
logger.error(f"Error getting models: {e}")
raise HTTPException(status_code=500, detail=str(e))
@app.get("/api/models/teachers")
async def get_teacher_models():
"""Get all teacher models"""
try:
teachers = models_manager.get_teacher_models()
selected = models_manager.get_selected_teachers()
return {
"success": True,
"teachers": teachers,
"selected": selected,
"total": len(teachers)
}
except Exception as e:
logger.error(f"Error getting teacher models: {e}")
raise HTTPException(status_code=500, detail=str(e))
@app.get("/api/models/students")
async def get_student_models():
"""Get all student models"""
try:
students = models_manager.get_student_models()
selected = models_manager.get_selected_student()
return {
"success": True,
"students": students,
"selected": selected,
"total": len(students)
}
except Exception as e:
logger.error(f"Error getting student models: {e}")
raise HTTPException(status_code=500, detail=str(e))
@app.post("/api/models/search")
async def search_models(request: ModelSearchRequest):
"""Search for models on Hugging Face"""
try:
results = await models_manager.search_huggingface_models(
query=request.query,
limit=request.limit,
model_type=request.model_type
)
return {
"success": True,
"results": results,
"count": len(results),
"query": request.query
}
except Exception as e:
logger.error(f"Error searching models: {e}")
raise HTTPException(status_code=500, detail=str(e))
@app.post("/api/models/add")
async def add_model(model_info: Dict[str, Any]):
"""Add a new model to the configuration"""
try:
success = await models_manager.add_model(model_info)
if success:
return {
"success": True,
"message": f"Model {model_info.get('model_id')} added successfully"
}
else:
raise HTTPException(status_code=400, detail="Failed to add model")
except Exception as e:
logger.error(f"Error adding model: {e}")
raise HTTPException(status_code=500, detail=str(e))
@app.post("/api/models/validate/{model_id:path}")
async def validate_model(model_id: str):
"""Validate a model"""
try:
validation_result = await models_manager.validate_model(model_id)
return {
"success": True,
"validation": validation_result,
"model_id": model_id
}
except Exception as e:
logger.error(f"Error validating model: {e}")
raise HTTPException(status_code=500, detail=str(e))
@app.post("/api/models/select")
async def select_models(request: ModelSelectionRequest):
"""Select teacher and student models"""
try:
results = []
# Select teacher models
for teacher_id in request.teacher_models:
success = models_manager.select_teacher(teacher_id)
results.append({
"model_id": teacher_id,
"type": "teacher",
"success": success
})
# Select student model
if request.student_model is not None:
success = models_manager.select_student(request.student_model)
results.append({
"model_id": request.student_model,
"type": "student",
"success": success
})
return {
"success": True,
"results": results,
"selected_teachers": models_manager.get_selected_teachers(),
"selected_student": models_manager.get_selected_student()
}
except Exception as e:
logger.error(f"Error selecting models: {e}")
raise HTTPException(status_code=500, detail=str(e))
@app.delete("/api/models/{model_id:path}")
async def remove_model(model_id: str):
"""Remove a model from configuration"""
try:
success = models_manager.remove_model(model_id)
if success:
return {
"success": True,
"message": f"Model {model_id} removed successfully"
}
else:
raise HTTPException(status_code=400, detail="Failed to remove model")
except Exception as e:
logger.error(f"Error removing model: {e}")
raise HTTPException(status_code=500, detail=str(e))
@app.get("/api/models/{model_id:path}")
async def get_model_info(model_id: str):
"""Get detailed information about a specific model"""
try:
model_info = models_manager.get_model_info(model_id)
if model_info:
return {
"success": True,
"model": model_info
}
else:
raise HTTPException(status_code=404, detail="Model not found")
except Exception as e:
logger.error(f"Error getting model info: {e}")
raise HTTPException(status_code=500, detail=str(e))
if __name__ == "__main__":
uvicorn.run(
"app:app",
host="0.0.0.0",
port=int(os.getenv("PORT", 7860)),
reload=False,
log_level="info"
)
|