| import os |
| from typing import Dict, Any, List |
|
|
| MASTER_SECRET = os.getenv("MASTER_SECRET", "nightglow-secret-key-2024") |
| REPO_ID = "MGFeng/Nightglow-AI" |
| BUCKET_ID = "MGFeng/Nightglow-AI-storage" |
| HEARTBEAT_TIMEOUT = 90 |
|
|
| DATA_SOURCES = [ |
| "nvidia/OpenCodeInstruct", |
| "Fsoft-AIC/the-vault-function", |
| "MGFeng/Nightglow-AI", |
| ] |
|
|
| MODEL_PRESETS: Dict[str, Dict[str, Any]] = { |
| "3B": { |
| "hidden_size": 3200, |
| "num_layers": 18, |
| "num_heads": 20, |
| "num_experts": 8, |
| "num_kv_heads": 20, |
| "intermediate_size": 12800, |
| "vocab_size": 49152, |
| "max_position_embeddings": 8192, |
| "dropout": 0.1, |
| "tasks": 1000, |
| }, |
| } |
|
|
| TRAINING_CONFIG = { |
| "batch_size": 2, |
| "max_seq_length": 512, |
| "learning_rate": 1e-4, |
| "gradient_clip": 1.0, |
| "weight_decay": 0.01, |
| "warmup_steps": 100, |
| "lr_scheduler": "cosine", |
| "use_fp16": True, |
| } |
|
|
| SORTED_SIZES = list(MODEL_PRESETS.keys()) |
|
|
| class ConfigManager: |
| _instance = None |
| def __new__(cls): |
| if cls._instance is None: |
| cls._instance = super().__new__(cls) |
| cls._instance._initialized = False |
| return cls._instance |
|
|
| def __init__(self): |
| if self._initialized: |
| return |
| self._config = { |
| "target_size": "3B", |
| "data_source": DATA_SOURCES[0], |
| "language": "python", |
| "total_steps": 1000, |
| "steps_per_task": 100, |
| **TRAINING_CONFIG, |
| } |
| self._initialized = True |
| print("✅ ConfigManager initialized") |
|
|
| def get(self, key: str, default=None): |
| return self._config.get(key, default) |
|
|
| def set(self, key: str, value: Any) -> bool: |
| if key not in self._config: |
| return False |
| self._config[key] = value |
| return True |
|
|
| def get_all(self) -> Dict[str, Any]: |
| return self._config.copy() |
|
|
| config = ConfigManager() |
