{
  "active_parameters_per_token": 50512000,
  "architecture": {
    "hidden_size": 896,
    "intermediate_size": 1024,
    "max_position_embeddings": 4096,
    "num_attention_heads": 14,
    "num_experts_per_tok": 1,
    "num_hidden_layers": 6,
    "num_key_value_heads": 14,
    "num_local_experts": 7,
    "vocab_size": 16384
  },
  "assistant_target_tokens": null,
  "completed_budget": true,
  "context_window": 4096,
  "dataset": "openbmb/Ultra-FineWeb (en, score>=0.70)",
  "elapsed_seconds": 14777.457427978516,
  "finished_at": "2026-08-26T07:11:20Z",
  "precision": "bf16 autocast with fp32 master weights",
  "qat": false,
  "qat_scheme": null,
  "sequence_length_trained": 1024,
  "stage": "pretrain",
  "steps": 50863,
  "tokens_trained": 5000036352,
  "total_parameters": 149602432
}