metadata.json

{
  "note": "Дообучение с балансировкой 4000/класс",
  "data_sources": [
    "LMSYS Arena",
    "LMSYS Chat 1M",
    "GLM-5.1",
    "ShareChat",
    "WildChat",
    "Kimi K2.6"
  ],
  "balanced_samples_per_class": 4000,
  "total_balanced": 44000,
  "families": [
    "deepseek",
    "claude",
    "chatgpt",
    "glm",
    "kimi",
    "qwen",
    "mistral",
    "minimax",
    "gemini",
    "llama",
    "gemma"
  ],
  "family_counts": {
    "DeepSeek": 4000,
    "Claude": 4000,
    "ChatGPT": 4000,
    "GLM": 4000,
    "Kimi": 4000,
    "Qwen": 4000,
    "Mistral": 4000,
    "MiniMax": 4000,
    "Gemini": 4000,
    "Llama": 4000,
    "Gemma": 4000
  }
}