config.json

{
  "model_type": "wildnerve_tlm01",
  "architectures": ["Wildnerve_tlm01"],
  "SELECTED_MODEL": ["model_Custm.py", "model_PrTr.py"],
  "MODEL_NAME": "Wildnerve-tlm01_Hybrid_Model",
  "BASE_DATA_DIR": "data",
  "FILE_FORMATS": ["csv", "json", "txt"],
  "MAX_SEQ_LENGTH": 767,
  "SIMILARITY_THRESHOLD": 0.85,
  "DATASET_PATHS": {
    "general": ["data/general.json"],
    "programming_software_dev": ["data/programming_software_dev.json"],
    "python": [
      "data/python_mbpp.json",
      "data/python_programming.json",
      "data/python_transformer_model.json"
    ],
    "rust": [
      "data/rust_ai_language_model.json",
      "data/rust_blockchain.json",
      "data/rust_mbrp.json",
      "data/rust_programming.json"
    ],
    "solidity": ["data/solidity_programming.json"],
    "computer": [
      "data/computer_advanced_debugging.json",
      "data/computer_agenticAI.json",
      "data/computer_architecture.json",
      "data/computer_cloud_security.json",
      "data/computer_cloudCI-CD.json",
      "data/computer_creativity.json",
      "data/computer_crossplatform.json",
      "data/computer_cybersecurity.json",
      "data/computer_error_handling_examples.json",
      "data/computer_gitInstruct.json"
    ],
    "cpp": [
      "data/cpp_ai_language_model.json",
      "data/cpp_blockchain.json",
      "data/cpp_mbcppp.json",
      "data/cpp_programming.json"
    ],
    "go": [
      "data/golang_ai_language_model.json",
      "data/golang_mbgp.json",
      "data/golang_programming.json"
    ],
    "java": [
      "data/java_ai_language_model.json",
      "data/java_blockchain.json",
      "data/java_mbjp.json",
      "data/java_programming.json",
      "data/java_transformer_language_model.json"
    ],
    "javascript": [
      "data/javascript_chatbot.json",
      "data/javascript_n_Typescript_backend.json",
      "data/javascript_n_Typescript_frontend.json",
      "data/javascript_n_Typescript_programming.json"
    ],
    "mathematics": [
      "data/mathematics.json",
      "data/mathematics_training.json"
    ],
    "nim": [
      "data/nim_ai_language_model.json",
      "data/nim_blockchain.json",
      "data/nim_chatbot.json",
      "data/nim_conversation.json",
      "data/nim_mbnp.json",
      "data/nim_programming.json"
    ],
    "other_information": ["data/other_information.json"],
    "physics": [
      "data/physics_n_engineering.json",
      "data/physics_n_engineering_applied.json",
      "data/project_structure.json",
      "data/python_chatbot_guide.json"
    ]
  },
  "LAZY_LOADING_ENABLED": true,
  "MAX_INITIAL_SPECIALIZATIONS": 2,
  "train_file_path": "data/computer_advanced_debugging.json",
  "NUM_EPOCHS": 50,
  "LEARNING_RATE": 0.0001,
  "INPUT_SIZE": 768,
  "OUTPUT_SIZE": 768,
  "SPECIALIZATIONS": [
    "python",
    "rust",
    "solidity",
    "computer",
    "cpp",
    "go",
    "java",
    "javascript",
    "mathematics",
    "nim",
    "other_information",
    "physics"
  ],
  "PREPROCESSING": {
    "LOWERCASE": true,
    "REMOVE_SPECIAL_CHARACTERS": true,
    "REPLACE_MULTIPLE_SPACES": true
  },
  "STDP_CONFIG": {
    "WEIGHT_THRESHOLD": 0.5,
    "ACTIVATION_THRESHOLD": 0.2,
    "USE_SNN": true,
    "ALPHA": 0.1,
    "BETA": 0.2,
    "BASE_DIR": "checkpoints",
    "SNN_FILENAME_FORMAT": "snn_model_{specialization}_{epoch}.pt",
    "STDPLearningRate": 0.01,
    "STDPMemDecay": 0.9,
    "SpikeThreshold": 0.5,
    "firing_rate": 10,
    "MAX_SEQ_LENGTH": 2048,
    "STDP_PRETRAIN_EPOCHS": 5,
    "STDP_FINETUNE_EPOCHS": 3,
    "BATCH_SIZE_PRETRAIN": 32,
    "BATCH_SIZE_FINETUNE": 16,
    "NUM_NEURONS": 1024,
    "MAX_RATE": 100
  },
  "TRAINING_CONFIG": {
    "PATIENCE": 3,
    "DELTA": 0.001,
    "VERBOSE": true,
    "NUM_EPOCHS": 20,
    "LEARNING_RATE": 0.0001,
    "TRANSFORMER_LEARNING_RATE": 5e-5,
    "TRANSFORMER_NUM_EPOCHS": 20
  },
  "CHECKPOINT_CONFIG": {
    "PATH": "checkpoints",
    "BASE_DIR": "checkpoints",
    "TRANSFORMER_FILENAME_FORMAT": "transformer_model_{specialization}_{epoch}.pt",
    "SNN_FILENAME_FORMAT": "snn_model_{specialization}_{epoch}.pt"
  },
  "GENERATION_CONFIG": {
    "temperature": 0.7,
    "top_p": 0.9,
    "num_return_sequences": 1
  },
  "TOKENIZER_CONFIG": {
    "MODEL_NAME": "gpt2", 
    "MAX_SEQ_LENGTH": 1024,
    "POOLING_MODE": "last"
  },
  "DATA_LOADER_CONFIG": {
    "BATCH_SIZE": 32,
    "NUM_WORKERS": 0,
    "SHUFFLE": true,
    "INCLUDE_CRAWL": true
  },
  "ATTENTION_CONFIG": {
    "WINDOW_SIZE": 256,
    "STRIDE": 128,
    "MEMORY_SIZE": 64,
    "NUM_HEADS": 8,
    "ATTENTION_DROPOUT": 0.1,
    "ATTENTION_TYPES": {
      "SLIDING": true,
      "HIERARCHICAL": true,
      "GLOBAL": true
    },
    "PROMPT_THRESHOLDS": {
      "LENGTH_THRESHOLD": 500,
      "COMPLEXITY_THRESHOLD": 0.7,
      "PERPLEXITY_THRESHOLD": 50
    },
    "ATTENTION_WEIGHTS": {
      "SHORT_COMPLEX": {
        "SLIDING": 0.4,
        "HIERARCHICAL": 0.6
      },
      "LONG_CONTEXT": {
        "SLIDING": 0.3,
        "HIERARCHICAL": 0.4,
        "GLOBAL": 0.3
      }
    }
  },
  "TRANSFORMER_CONFIG": {
    "TEST_MODE": false,
    "LOGGING_LEVEL": "INFO",
    "LOG_FILE": "logs/training.log",
    "SAVE_CHECKPOINTS": true,
    "BASE_DIR": "checkpoints",
    "TRANSFORMER_FILENAME_FORMAT": "transformer_model_{specialization}_{epoch}.pt",
    "MODEL_NAME": "gpt2",
    "MAX_SEQ_LENGTH": 767,
    "NUM_EPOCHS": 20,
    "LEARNING_RATE": 5e-5,
    "BATCH_SIZE": 32,
    "EMBEDDING_DIM": 768,
    "NUM_HEADS": 12,
    "HIDDEN_DIM": 768,
    "NUM_LAYERS": 12,
    "DROPOUT": 0.1,
    "specialization1": "cpp",
    "specialization2": "java",
    "specialization3": "go",
    "specialization4": "javascript",
    "specialization5": "nim",
    "specialization6": "python",
    "specialization7": "rust",
    "specialization8": "solidity",
    "specialization9": "computer",
    "specialization10": "mathematics",
    "specialization11": "physics",
    "specialization12": "other_information", 
    "DATASET_PATH": "data/cpp_ai_language_model.json",
    "OUTPUT_SIZE": 768,
    "POOLING_MODE": "mean",
    "VOCAB_SIZE": 50257,
    "MAX_RATE": 100,
    "MODE": "pretrained",
    "MODE2": "custom",
    "SHUFFLE": true,
    "SIMILARITY_THRESHOLD": 0.85,
    "USE_PRETRAINED_ENCODER": true,
    "ATTENTION_MECHANISM": {
      "TYPE": "hybrid",
      "WINDOW_SIZE": 1024,
      "STRIDE": 512,
      "USE_MEMORY": true
    }
  },
  "DUAL_ENCODER_CONFIG": {
    "USE_PRETRAINED_ENCODER": true,
    "USE_CUSTOM_ENCODER": true,
    "DEBUG": false
  },
  "PROMPT_ANALYZER_CONFIG": {
    "MODEL_NAME": "gpt2",
    "DATASET_PATH": null,
    "SPECIALIZATION": null,
    "HIDDEN_DIM": 768,
    "MAX_CACHE_SIZE": 10
  },
  "MODEL_PRIORITY": {
    "PRIMARY": "model_Custm",
    "SECONDARY": "model_PrTr",
    "USE_PRETRAINED_FALLBACK": true
  },
  "MAX_ACTIVE_MODELS": 5,
  "MODEL_IDLE_THRESHOLD": 600,
  "MAX_MEMORY_USAGE": 0.8,
  "TOP_K": 3,
  "TOPIC_KEYWORDS": {
    "programming": ["python", "java", "javascript", "typescript", "rust", "go", "golang", 
                    "c++", "cpp", "c#", "ruby", "php", "perl", "swift", "kotlin", "scala",
                    "html", "css", "sql", "database", "code", "function", "class", "method",
                    "algorithm", "data structure", "api", "http", "rest", "graphql", "json"],
    "computer_science": ["algorithm", "complexity", "big o", "data structure", "graph", "tree", "hash table",
                        "linked list", "stack", "queue", "heap", "binary search", "sorting", "recursion",
                        "dynamic programming", "greedy algorithm", "backtracking", "computational", "turing"],
    "software_engineering": ["design pattern", "architecture", "solid principles", "agile", "scrum", "kanban",
                            "waterfall", "ci/cd", "continuous integration", "continuous deployment", "testing",
                            "unit test", "integration test", "e2e test", "test-driven", "tdd", "bdd"],
    "web_development": ["frontend", "backend", "full-stack", "web app", "spa", "pwa", "responsive",
                        "mobile-first", "css framework", "react", "angular", "vue", "svelte", "next.js"]
  }
}