{
  "model": "AppleMind 1.0 Mini",
  "parameters": 1020480,
  "trainable_parameters": 1020480,
  "final_steps": 2288,
  "final_tokens": 299892736,
  "target_tokens": 300000000,
  "actual_tokens_per_parameter": 293.8741925368454,
  "dataset": {
    "fineweb_edu": 100000000,
    "fineweb_hq": 100000000,
    "smollm": 100000000
  },
  "model_config": {
    "n_layer": 2,
    "n_head": 2,
    "n_embd": 20,
    "context_length": 256,
    "vocab_size": 50260
  },
  "training": {
    "batch_size": 64,
    "gradient_accumulation": 8,
    "tokens_per_micro_batch": 16384,
    "tokens_per_step": 131072,
    "learning_rate": 3e-05,
    "weight_decay": 0.1,
    "gradient_clip": 1.0,
    "bf16": true,
    "tf32": true,
    "torch_compile": false
  },
  "repositories": {
    "checkpoints": "AppleMind-AI/AppleMind-1.0-Mini-Checkpoints",
    "final": "AppleMind-AI/AppleMind-1.0-Mini"
  }
}