Spaces:
Sleeping
Sleeping
File size: 37,529 Bytes
01db2f8 |
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 |
"""
Generative AI for Enhancing Programming Education
================================================
This project implements a fine-tuned CodeLlama-7B model to provide structured,
educational code feedback for programming students.
Problem Statement:
- High dropout rates in programming education
- Inefficient feedback loops
- Lack of personalized learning
- Limited instructor bandwidth
- Current AI tools prioritize productivity over learning
Solution:
- Fine-tuned CodeLlama-7B for educational feedback
- Structured, actionable code reviews
- Beginner-friendly explanations
- Personalized adaptation based on skill level
- Educational focus with ethical safeguards
Author: [Your Name]
Date: [Current Date]
"""
import re
from dataclasses import dataclass
from typing import Dict, List, Optional, Tuple
import logging
import json
from transformers import AutoTokenizer, AutoModelForCausalLM
import os
import gc
import torch
import warnings
warnings.filterwarnings("ignore", category=UserWarning)
# --- Critical Environment Setup (Must be before imports) ---
os.environ["TOKENIZERS_PARALLELISM"] = "false"
os.environ["DATASETS_DISABLE_MULTIPROCESSING"] = "1"
# Clear any existing CUDA cache (only if CUDA is available)
if torch.cuda.is_available():
os.environ["PYTORCH_CUDA_ALLOC_CONF"] = "max_split_size_mb:128,garbage_collection_threshold:0.6"
torch.cuda.empty_cache()
gc.collect()
# Configure logging
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)
def clear_cuda_cache():
"""Clear CUDA cache and run garbage collection"""
if torch.cuda.is_available():
torch.cuda.empty_cache()
torch.cuda.synchronize()
gc.collect()
def get_system_memory():
"""Get system memory information"""
try:
import psutil
memory = psutil.virtual_memory()
print(
f"System RAM: {memory.used / (1024**3):.1f}GB / {memory.total / (1024**3):.1f}GB used ({memory.percent:.1f}%)")
except Exception as e:
print(f"Could not get system memory info: {e}")
def get_gpu_memory():
"""Get GPU memory information (if available)"""
if torch.cuda.is_available():
try:
import subprocess
result = subprocess.run(['nvidia-smi', '--query-gpu=memory.used,memory.total', '--format=csv,nounits,noheader'],
capture_output=True, text=True)
lines = result.stdout.strip().split('\n')
for i, line in enumerate(lines):
used, total = map(int, line.split(', '))
print(
f"GPU {i}: {used}MB / {total}MB used ({used/total*100:.1f}%)")
except Exception as e:
print(f"Could not get GPU memory info: {e}")
else:
print("No GPU available - using CPU only")
@dataclass
class CodeFeedback:
"""Data structure for storing code feedback"""
code_snippet: str
feedback_type: str # 'syntax', 'logic', 'optimization', 'style', 'explanation'
feedback_message: str
suggested_improvement: Optional[str] = None
difficulty_level: str = 'beginner' # 'beginner', 'intermediate', 'advanced'
learning_objectives: List[str] = None
@dataclass
class ComprehensiveFeedback:
"""Comprehensive feedback structure with all educational components"""
code_snippet: str
student_level: str
# Analysis
strengths: List[str]
weaknesses: List[str]
issues: List[str]
# Educational content
step_by_step_improvement: List[str]
learning_points: List[str]
review_summary: str
# Interactive elements
comprehension_question: str
comprehension_answer: str
explanation: str
# Code fixes
improved_code: str
fix_explanation: str
# Metadata
difficulty_level: str
learning_objectives: List[str]
estimated_time_to_improve: str
class ProgrammingEducationAI:
"""
Main class for the fine-tuned CodeLlama model for programming education
"""
def __init__(self, model_path: str = "TomoriFarouk/codellama-7b-programming-education"):
"""
Initialize the fine-tuned model and tokenizer
Args:
model_path: Path to your fine-tuned CodeLlama-7B model
"""
self.model_path = model_path
self.tokenizer = None
self.model = None
self.feedback_templates = self._load_feedback_templates()
self.code_review_prompt_template = self._load_code_review_prompt()
self.code_feedback_prompt_template = self._load_code_feedback_prompt()
self.comprehensive_feedback_prompt = self._load_comprehensive_feedback_prompt()
self.comprehension_question_prompt = self._load_comprehension_question_prompt()
self.code_fix_prompt = self._load_code_fix_prompt()
def _load_code_review_prompt(self) -> str:
"""Load the code review prompt template used during fine-tuning"""
return """You are an expert programming tutor. Review the following student code and provide educational feedback.
Student Code:
{code}
Student Level: {level}
Please provide:
1. Syntax errors (if any)
2. Logic errors (if any)
3. Style improvements
4. Optimization suggestions
5. Educational explanations
Feedback:"""
def _load_code_feedback_prompt(self) -> str:
"""Load the code feedback prompt template used during fine-tuning"""
return """You are a helpful programming tutor. The student has written this code:
{code}
Student Level: {level}
Provide constructive, educational feedback that helps the student learn. Focus on:
- What they did well
- What can be improved
- Why the improvement matters
- How to implement the improvement
Feedback:"""
def _load_feedback_templates(self) -> Dict[str, str]:
"""Load predefined feedback templates for different scenarios"""
return {
"syntax_error": "I notice there's a syntax issue in your code. {error_description}. "
"Here's what's happening: {explanation}. "
"Try this correction: {suggestion}",
"logic_error": "Your code has a logical issue. {problem_description}. "
"The problem is: {explanation}. "
"Consider this approach: {suggestion}",
"optimization": "Your code works, but we can make it more efficient! "
"Current complexity: {current_complexity}. "
"Optimized version: {optimized_complexity}. "
"Here's how: {explanation}",
"style_improvement": "Great work! Here's a style tip: {tip}. "
"This makes your code more readable and maintainable.",
"concept_explanation": "Let me explain this concept: {concept}. "
"In simple terms: {simple_explanation}. "
"Example: {example}"
}
def load_model(self, load_in_8bit=False, device_map="auto"):
"""
Load the fine-tuned model and tokenizer optimized for HF Spaces
Args:
load_in_8bit: Whether to load model in 8-bit precision for memory savings
device_map: Device mapping strategy ("auto", "cpu", or specific device)
"""
try:
# Get HF token for private model access
hf_token = os.getenv("HF_TOKEN", None)
logger.info("Loading tokenizer...")
self.tokenizer = AutoTokenizer.from_pretrained(
self.model_path,
trust_remote_code=True,
token=hf_token # Use token for private models
)
# Set padding token
if self.tokenizer.pad_token is None:
self.tokenizer.pad_token = self.tokenizer.eos_token
self.tokenizer.pad_token_id = self.tokenizer.eos_token_id
logger.info(
f"Tokenizer loaded - Vocab size: {len(self.tokenizer)}")
# Load model with memory optimization options
print(
f"Loading model with 8-bit: {load_in_8bit}, device_map: {device_map}...")
# Prepare model loading arguments
model_kwargs = {
"torch_dtype": torch.float32,
"low_cpu_mem_usage": True,
"trust_remote_code": True,
"offload_folder": "offload", # Offload to disk if needed
"token": hf_token # Use token for private models
}
# Add 8-bit quantization if requested
if load_in_8bit:
model_kwargs["load_in_8bit"] = True
print("🔧 Using 8-bit quantization for memory optimization")
# Add device mapping
if device_map:
model_kwargs["device_map"] = device_map
print(f"🔧 Using device map: {device_map}")
self.model = AutoModelForCausalLM.from_pretrained(
self.model_path,
**model_kwargs
)
# Enable gradient checkpointing for memory savings
self.model.gradient_checkpointing_enable()
logger.info("Fine-tuned model loaded successfully")
logger.info(
f"Model loaded on devices: {getattr(self.model, 'hf_device_map', 'CPU')}")
except Exception as e:
logger.error(f"Error loading fine-tuned model: {e}")
raise
def generate_code_review(self, code: str, student_level: str = "beginner") -> str:
"""
Generate code review using the fine-tuned model
Args:
code: Student's code to review
student_level: Student's skill level
Returns:
Generated code review feedback
"""
if not self.model or not self.tokenizer:
raise ValueError("Model not loaded. Call load_model() first.")
# Format the prompt using the template from fine-tuning
prompt = self.code_review_prompt_template.format(
code=code,
level=student_level
)
# Tokenize input
inputs = self.tokenizer(
prompt, return_tensors="pt", truncation=True, max_length=2048)
# Generate response
with torch.no_grad():
outputs = self.model.generate(
inputs.input_ids,
max_new_tokens=512,
temperature=0.7,
do_sample=True,
pad_token_id=self.tokenizer.eos_token_id
)
# Decode response
response = self.tokenizer.decode(outputs[0], skip_special_tokens=True)
# Extract only the generated part (after the prompt)
generated_text = response[len(prompt):].strip()
return generated_text
def generate_educational_feedback(self, code: str, student_level: str = "beginner") -> str:
"""
Generate educational feedback using the fine-tuned model
Args:
code: Student's code to provide feedback on
student_level: Student's skill level
Returns:
Generated educational feedback
"""
if not self.model or not self.tokenizer:
raise ValueError("Model not loaded. Call load_model() first.")
# Format the prompt using the template from fine-tuning
prompt = self.code_feedback_prompt_template.format(
code=code,
level=student_level
)
# Tokenize input
inputs = self.tokenizer(
prompt, return_tensors="pt", truncation=True, max_length=2048)
# Generate response
with torch.no_grad():
outputs = self.model.generate(
inputs.input_ids,
max_new_tokens=512,
temperature=0.7,
do_sample=True,
pad_token_id=self.tokenizer.eos_token_id
)
# Decode response
response = self.tokenizer.decode(outputs[0], skip_special_tokens=True)
# Extract only the generated part (after the prompt)
generated_text = response[len(prompt):].strip()
return generated_text
def analyze_student_code(self, code: str, student_level: str = "beginner") -> List[CodeFeedback]:
"""
Analyze student code and provide educational feedback using the fine-tuned model
Args:
code: The student's code to analyze
student_level: Student's skill level ('beginner', 'intermediate', 'advanced')
Returns:
List of CodeFeedback objects
"""
feedback_list = []
# Use fine-tuned model for comprehensive code review
try:
code_review = self.generate_code_review(code, student_level)
educational_feedback = self.generate_educational_feedback(
code, student_level)
# Create structured feedback from model output
feedback_list.append(CodeFeedback(
code_snippet=code,
feedback_type="comprehensive_review",
feedback_message=code_review,
difficulty_level=student_level,
learning_objectives=["code_analysis", "best_practices"]
))
feedback_list.append(CodeFeedback(
code_snippet=code,
feedback_type="educational_guidance",
feedback_message=educational_feedback,
difficulty_level=student_level,
learning_objectives=["learning", "improvement"]
))
except Exception as e:
logger.warning(
f"Fine-tuned model failed, falling back to rule-based analysis: {e}")
# Fallback to rule-based analysis if model fails
feedback_list = self._fallback_analysis(code, student_level)
return feedback_list
def _fallback_analysis(self, code: str, student_level: str) -> List[CodeFeedback]:
"""Fallback analysis using rule-based methods if fine-tuned model fails"""
feedback_list = []
# Analyze syntax
syntax_feedback = self._check_syntax(code, student_level)
if syntax_feedback:
feedback_list.append(syntax_feedback)
# Analyze logic and structure
logic_feedback = self._check_logic(code, student_level)
if logic_feedback:
feedback_list.extend(logic_feedback)
# Check for optimization opportunities
optimization_feedback = self._check_optimization(code, student_level)
if optimization_feedback:
feedback_list.append(optimization_feedback)
# Provide style suggestions
style_feedback = self._check_style(code, student_level)
if style_feedback:
feedback_list.append(style_feedback)
return feedback_list
def _check_syntax(self, code: str, student_level: str) -> Optional[CodeFeedback]:
"""Check for syntax errors and provide educational feedback"""
# This would integrate with the fine-tuned model
# For now, using basic pattern matching as placeholder
common_syntax_errors = {
r"print\s*\([^)]*\)\s*$": "Remember to add a colon after print statements in some contexts",
r"if\s+[^:]+$": "Don't forget the colon after your if condition",
r"for\s+[^:]+$": "Don't forget the colon after your for loop",
}
for pattern, message in common_syntax_errors.items():
if re.search(pattern, code):
return CodeFeedback(
code_snippet=code,
feedback_type="syntax",
feedback_message=message,
difficulty_level=student_level,
learning_objectives=["syntax", "basic_python"]
)
return None
def _check_logic(self, code: str, student_level: str) -> List[CodeFeedback]:
"""Check for logical errors and provide educational feedback"""
feedback_list = []
# Check for infinite loops
if "while True:" in code and "break" not in code:
feedback_list.append(CodeFeedback(
code_snippet=code,
feedback_type="logic",
feedback_message="This while loop will run forever! Make sure to include a break statement or condition to exit the loop.",
difficulty_level=student_level,
learning_objectives=["control_flow", "loops"]
))
# Check for unused variables
# This is a simplified check - the actual model would be more sophisticated
if "x = " in code and "x" not in code.replace("x = ", ""):
feedback_list.append(CodeFeedback(
code_snippet=code,
feedback_type="logic",
feedback_message="You created variable 'x' but didn't use it. Consider removing unused variables to keep your code clean.",
difficulty_level=student_level,
learning_objectives=["variables", "code_cleanliness"]
))
return feedback_list
def _check_optimization(self, code: str, student_level: str) -> Optional[CodeFeedback]:
"""Check for optimization opportunities"""
# Check for nested loops that could be optimized
if code.count("for") > 1 and code.count("in range") > 1:
return CodeFeedback(
code_snippet=code,
feedback_type="optimization",
feedback_message="You have nested loops here. Consider if you can optimize this to O(n) instead of O(n²).",
suggested_improvement="Use a hashmap or set to reduce complexity",
difficulty_level=student_level,
learning_objectives=["algorithms",
"complexity", "optimization"]
)
return None
def _check_style(self, code: str, student_level: str) -> Optional[CodeFeedback]:
"""Check for style improvements"""
# Check for meaningful variable names
if "x" in code or "y" in code or "z" in code:
return CodeFeedback(
code_snippet=code,
feedback_type="style",
feedback_message="Consider using more descriptive variable names instead of x, y, z. This makes your code easier to understand.",
difficulty_level=student_level,
learning_objectives=["naming_conventions", "readability"]
)
return None
def generate_explanation(self, concept: str, student_level: str) -> str:
"""
Generate explanations for programming concepts based on student level
Args:
concept: The concept to explain
student_level: Student's skill level
Returns:
Explanation tailored to the student's level
"""
explanations = {
"variables": {
"beginner": "Variables are like labeled boxes where you store information. Think of 'name = \"John\"' as putting \"John\" in a box labeled 'name'.",
"intermediate": "Variables are memory locations that store data. They have a name, type, and value. Python is dynamically typed, so the type is inferred.",
"advanced": "Variables in Python are references to objects in memory. They're dynamically typed and use reference counting for memory management."
},
"loops": {
"beginner": "Loops repeat code multiple times. 'for' loops are great when you know how many times to repeat, 'while' loops when you don't.",
"intermediate": "Loops control program flow. 'for' iterates over sequences, 'while' continues until a condition is False. Consider time complexity.",
"advanced": "Loops are fundamental control structures. Python's 'for' is actually a foreach loop. Consider iterator patterns and generator expressions."
}
}
return explanations.get(concept, {}).get(student_level, f"Explanation for {concept} at {student_level} level")
def _load_comprehensive_feedback_prompt(self) -> str:
"""Load the comprehensive feedback prompt template"""
return """You are an expert programming tutor. Provide comprehensive educational feedback for the following student code.
Student Code:
{code}
Student Level: {level}
Please provide a detailed analysis in the following JSON format:
{{
"strengths": ["strength1", "strength2", "strength3"],
"weaknesses": ["weakness1", "weakness2", "weakness3"],
"issues": ["issue1", "issue2", "issue3"],
"step_by_step_improvement": [
"Step 1: Description of first improvement",
"Step 2: Description of second improvement",
"Step 3: Description of third improvement"
],
"learning_points": [
"Learning point 1: What the student should understand",
"Learning point 2: Key concept to grasp",
"Learning point 3: Best practice to follow"
],
"review_summary": "A comprehensive review of the code highlighting key areas for improvement",
"learning_objectives": ["objective1", "objective2", "objective3"],
"estimated_time_to_improve": "5-10 minutes"
}}
Focus on educational value and constructive feedback that helps the student learn and improve."""
def _load_comprehension_question_prompt(self) -> str:
"""Load the comprehension question generation prompt"""
return """Based on the learning points and improvements discussed, generate a comprehension question to test the student's understanding.
Learning Points: {learning_points}
Code Issues: {issues}
Student Level: {level}
Generate a question that tests understanding of the key concepts discussed. The question should be appropriate for the student's level.
Format your response as JSON:
{{
"question": "Your comprehension question here",
"answer": "The correct answer",
"explanation": "Detailed explanation of why this answer is correct"
}}
Make the question challenging but fair for the student's level."""
def _load_code_fix_prompt(self) -> str:
"""Load the code fix generation prompt"""
return """You are an expert programming tutor. Based on the analysis and learning points, provide an improved version of the student's code.
Original Code:
{code}
Issues Identified: {issues}
Learning Points: {learning_points}
Student Level: {level}
Provide an improved version of the code that addresses the issues while maintaining educational value. Include comments to explain the improvements.
Format your response as JSON:
{{
"improved_code": "The improved code with comments",
"fix_explanation": "Detailed explanation of what was changed and why"
}}
Focus on educational improvements that help the student understand better practices."""
def adapt_feedback_complexity(self, feedback: CodeFeedback, student_level: str) -> CodeFeedback:
"""
Adapt feedback complexity based on student level
Args:
feedback: Original feedback
student_level: Student's skill level
Returns:
Adapted feedback
"""
if student_level == "beginner":
# Simplify language and add more examples
feedback.feedback_message = feedback.feedback_message.replace(
"O(n²)", "quadratic time (slower)"
).replace(
"O(n)", "linear time (faster)"
)
elif student_level == "advanced":
# Add more technical details
if "optimization" in feedback.feedback_type:
feedback.feedback_message += " Consider the space-time tradeoff and cache locality."
return feedback
def generate_comprehensive_feedback(self, code: str, student_level: str = "beginner") -> ComprehensiveFeedback:
"""
Generate comprehensive educational feedback with all components
Args:
code: Student's code to analyze
student_level: Student's skill level
Returns:
ComprehensiveFeedback object with all educational components
"""
if not self.model or not self.tokenizer:
raise ValueError("Model not loaded. Call load_model() first.")
try:
# Step 1: Generate comprehensive analysis
comprehensive_analysis = self._generate_comprehensive_analysis(
code, student_level)
# Step 2: Generate comprehension question
comprehension_data = self._generate_comprehension_question(
comprehensive_analysis["learning_points"],
comprehensive_analysis["issues"],
student_level
)
# Step 3: Generate improved code
code_fix_data = self._generate_code_fix(
code,
comprehensive_analysis["issues"],
comprehensive_analysis["learning_points"],
student_level
)
# Create comprehensive feedback object
return ComprehensiveFeedback(
code_snippet=code,
student_level=student_level,
strengths=comprehensive_analysis["strengths"],
weaknesses=comprehensive_analysis["weaknesses"],
issues=comprehensive_analysis["issues"],
step_by_step_improvement=comprehensive_analysis["step_by_step_improvement"],
learning_points=comprehensive_analysis["learning_points"],
review_summary=comprehensive_analysis["review_summary"],
comprehension_question=comprehension_data["question"],
comprehension_answer=comprehension_data["answer"],
explanation=comprehension_data["explanation"],
improved_code=code_fix_data["improved_code"],
fix_explanation=code_fix_data["fix_explanation"],
difficulty_level=student_level,
learning_objectives=comprehensive_analysis["learning_objectives"],
estimated_time_to_improve=comprehensive_analysis["estimated_time_to_improve"]
)
except Exception as e:
logger.error(f"Error generating comprehensive feedback: {e}")
# Return a basic comprehensive feedback if model fails
return self._create_fallback_comprehensive_feedback(code, student_level)
def _generate_comprehensive_analysis(self, code: str, student_level: str) -> Dict:
"""Generate comprehensive analysis using the fine-tuned model"""
prompt = self.comprehensive_feedback_prompt.format(
code=code,
level=student_level
)
response = self._generate_model_response(prompt)
try:
# Try to parse JSON response
import json
return json.loads(response)
except json.JSONDecodeError:
logger.warning("Failed to parse JSON response, using fallback")
return self._create_fallback_analysis(code, student_level)
def _generate_comprehension_question(self, learning_points: List[str], issues: List[str], student_level: str) -> Dict:
"""Generate comprehension question using the fine-tuned model"""
prompt = self.comprehension_question_prompt.format(
learning_points=", ".join(learning_points),
issues=", ".join(issues),
level=student_level
)
response = self._generate_model_response(prompt)
try:
import json
return json.loads(response)
except json.JSONDecodeError:
logger.warning(
"Failed to parse comprehension question JSON, using fallback")
return {
"question": "What is the main concept you learned from this code review?",
"answer": "The main concept is understanding code structure and best practices.",
"explanation": "This question tests your understanding of the key learning points discussed."
}
def _generate_code_fix(self, code: str, issues: List[str], learning_points: List[str], student_level: str) -> Dict:
"""Generate improved code using the fine-tuned model"""
prompt = self.code_fix_prompt.format(
code=code,
issues=", ".join(issues),
learning_points=", ".join(learning_points),
level=student_level
)
response = self._generate_model_response(prompt)
try:
import json
return json.loads(response)
except json.JSONDecodeError:
logger.warning("Failed to parse code fix JSON, using fallback")
return {
"improved_code": "# Improved version of your code\n# Add comments and improvements here",
"fix_explanation": "This is a fallback improved version. The model should provide specific improvements."
}
def _generate_model_response(self, prompt: str) -> str:
"""Generate response from the fine-tuned model"""
inputs = self.tokenizer(
prompt, return_tensors="pt", truncation=True, max_length=2048)
# Move to CPU if no GPU available
if not torch.cuda.is_available():
inputs = {k: v.cpu() for k, v in inputs.items()}
with torch.no_grad():
outputs = self.model.generate(
inputs.input_ids,
max_new_tokens=512,
temperature=0.7,
do_sample=True,
pad_token_id=self.tokenizer.eos_token_id
)
response = self.tokenizer.decode(outputs[0], skip_special_tokens=True)
return response[len(prompt):].strip()
def _create_fallback_analysis(self, code: str, student_level: str) -> Dict:
"""Create fallback analysis when model fails"""
return {
"strengths": ["Your code has a clear structure", "You're using appropriate data types"],
"weaknesses": ["Could improve variable naming", "Consider adding comments"],
"issues": ["Basic syntax and style issues"],
"step_by_step_improvement": [
"Step 1: Add descriptive variable names",
"Step 2: Include comments explaining your logic",
"Step 3: Consider code optimization"
],
"learning_points": [
"Good variable naming improves code readability",
"Comments help others understand your code",
"Always consider efficiency in your solutions"
],
"review_summary": "Your code works but could be improved with better practices.",
"learning_objectives": ["code_quality", "best_practices", "readability"],
"estimated_time_to_improve": "10-15 minutes"
}
def _create_fallback_comprehensive_feedback(self, code: str, student_level: str) -> ComprehensiveFeedback:
"""Create fallback comprehensive feedback when model fails"""
fallback_analysis = self._create_fallback_analysis(code, student_level)
return ComprehensiveFeedback(
code_snippet=code,
student_level=student_level,
strengths=fallback_analysis["strengths"],
weaknesses=fallback_analysis["weaknesses"],
issues=fallback_analysis["issues"],
step_by_step_improvement=fallback_analysis["step_by_step_improvement"],
learning_points=fallback_analysis["learning_points"],
review_summary=fallback_analysis["review_summary"],
comprehension_question="What is the importance of good variable naming in programming?",
comprehension_answer="Good variable naming makes code more readable and maintainable.",
explanation="Descriptive variable names help other developers (and yourself) understand what the code does.",
improved_code="# Improved version\n# Add your improvements here",
fix_explanation="This is a fallback version. The model should provide specific improvements.",
difficulty_level=student_level,
learning_objectives=fallback_analysis["learning_objectives"],
estimated_time_to_improve=fallback_analysis["estimated_time_to_improve"]
)
def main():
"""Main function to demonstrate the system with fine-tuned model"""
print("Generative AI for Programming Education")
print("Using Fine-tuned CodeLlama-7B Model")
print("=" * 50)
# System information
print(f"Available GPUs: {torch.cuda.device_count()}")
if torch.cuda.is_available():
print("GPU Memory before loading:")
get_gpu_memory()
else:
print("System Memory before loading:")
get_system_memory()
# Initialize the system with your fine-tuned model path
# Update this path to point to your actual fine-tuned model
model_path = r"C:\Users\farou\OneDrive - Aston University\finetunning"
ai_tutor = ProgrammingEducationAI(model_path)
try:
# Load the fine-tuned model
print("Loading fine-tuned model...")
ai_tutor.load_model()
print("✓ Model loaded successfully!")
# Clear cache after loading
clear_cuda_cache()
if torch.cuda.is_available():
print("GPU Memory after loading:")
get_gpu_memory()
else:
print("System Memory after loading:")
get_system_memory()
# Example student code for testing
student_code = """
def find_duplicates(numbers):
x = []
for i in range(len(numbers)):
for j in range(i+1, len(numbers)):
if numbers[i] == numbers[j]:
x.append(numbers[i])
return x
# Test the function
result = find_duplicates([1, 2, 3, 2, 4, 5, 3])
print(result)
"""
print(f"\nAnalyzing student code:\n{student_code}")
# Get feedback using fine-tuned model
feedback_list = ai_tutor.analyze_student_code(student_code, "beginner")
print("\n" + "="*50)
print("FINE-TUNED MODEL FEEDBACK:")
print("="*50)
for i, feedback in enumerate(feedback_list, 1):
print(f"\n{i}. {feedback.feedback_type.upper()}:")
print(f" {feedback.feedback_message}")
if feedback.suggested_improvement:
print(f" Suggestion: {feedback.suggested_improvement}")
print(
f" Learning objectives: {', '.join(feedback.learning_objectives)}")
# Demonstrate direct model calls
print("\n" + "="*50)
print("DIRECT MODEL GENERATION:")
print("="*50)
# Code review
print("\n1. CODE REVIEW:")
code_review = ai_tutor.generate_code_review(student_code, "beginner")
print(code_review)
# Educational feedback
print("\n2. EDUCATIONAL FEEDBACK:")
educational_feedback = ai_tutor.generate_educational_feedback(
student_code, "beginner")
print(educational_feedback)
# Demonstrate comprehensive feedback system
print("\n" + "="*50)
print("COMPREHENSIVE EDUCATIONAL FEEDBACK SYSTEM:")
print("="*50)
comprehensive_feedback = ai_tutor.generate_comprehensive_feedback(
student_code, "beginner")
# Display comprehensive feedback
print("\n📊 CODE ANALYSIS:")
print("="*30)
print("\n✅ STRENGTHS:")
for i, strength in enumerate(comprehensive_feedback.strengths, 1):
print(f" {i}. {strength}")
print("\n❌ WEAKNESSES:")
for i, weakness in enumerate(comprehensive_feedback.weaknesses, 1):
print(f" {i}. {weakness}")
print("\n⚠️ ISSUES:")
for i, issue in enumerate(comprehensive_feedback.issues, 1):
print(f" {i}. {issue}")
print("\n📝 STEP-BY-STEP IMPROVEMENT GUIDE:")
print("="*40)
for i, step in enumerate(comprehensive_feedback.step_by_step_improvement, 1):
print(f" Step {i}: {step}")
print("\n🎓 LEARNING POINTS:")
print("="*25)
for i, point in enumerate(comprehensive_feedback.learning_points, 1):
print(f" {i}. {point}")
print("\n📋 REVIEW SUMMARY:")
print("="*20)
print(f" {comprehensive_feedback.review_summary}")
print("\n❓ COMPREHENSION QUESTION:")
print("="*30)
print(f" Question: {comprehensive_feedback.comprehension_question}")
print(f" Answer: {comprehensive_feedback.comprehension_answer}")
print(f" Explanation: {comprehensive_feedback.explanation}")
print("\n🔧 IMPROVED CODE:")
print("="*20)
print(comprehensive_feedback.improved_code)
print("\n💡 FIX EXPLANATION:")
print("="*20)
print(f" {comprehensive_feedback.fix_explanation}")
print("\n📊 METADATA:")
print("="*15)
print(f" Student Level: {comprehensive_feedback.student_level}")
print(
f" Learning Objectives: {', '.join(comprehensive_feedback.learning_objectives)}")
print(
f" Estimated Time to Improve: {comprehensive_feedback.estimated_time_to_improve}")
except Exception as e:
print(f"Error: {e}")
print(
"Make sure to update the model_path variable to point to your fine-tuned model.")
if __name__ == "__main__":
main()
|