{
  "output_dir": "/scratch/11079/antonxue/cache/ADLMC/diffuqwen35_yolo25k_4b_20260731_204150_50k",
  "base_model": "/scratch/11079/antonxue/cache/huggingface/hub/models--Qwen--Qwen3.5-4B/snapshots/851bf6e806efd8d0a36b00ddf55e13ccb7b8cd0a",
  "data_dir": "/scratch/11079/antonxue/adlmc_pretrain",
  "mixture": "v3",
  "max_length": 4096,
  "stable_steps": 44000,
  "decay_steps": 5000,
  "short_batch_probability": 0.0,
  "truncate_probability": 0.1,
  "truncate_min_frac": 0.5,
  "batch_size": 4,
  "gradient_accumulation_steps": 4,
  "learning_rate": 2e-05,
  "warmup_steps": 1000,
  "weight_decay": 0.01,
  "logging_steps": 5,
  "save_steps": 1000,
  "save_total_limit": 7,
  "seed": 42,
  "dataloader_num_workers": 8,
  "gradient_checkpointing": true,
  "resume_from_checkpoint": "/scratch/11079/antonxue/cache/ADLMC/diffuqwen35_yolo25k_4b_20260731_204150_50k"
}