checkpoints: null
data: null
general:
  benchmark_csv_path: null
  consumed_train_samples: null
  ignore_sanity_checks: false
  project: openbmb
  run: MiniCPM-2B-dpo-bf16
  seed: 42
  step: 0
lighteval: null
logging: null
model:
  ddp_bucket_cap_mb: 25
  dtype: bfloat16
  init_method:
    std: 0.025
  make_vocab_size_divisible_by: 1
  model_config:
    attn_pdrop: 0.0
    bos_token_id: 1
    dim_model_base: 256
    eos_token_id: 2
    hidden_act: silu
    hidden_size: 2304
    initializer_range: 0.1
    intermediate_size: 5760
    max_position_embeddings: 2048
    num_attention_heads: 36
    num_hidden_layers: 40
    num_key_value_heads: 36
    pad_token_id: null
    pretraining_tp: 1
    rms_norm_eps: 1.0e-05
    rope_theta: 10000.0
    scale_depth: 1.4
    scale_emb: 12
    tie_word_embeddings: true
    use_cache: true
    vocab_size: 122753
optimizer: null
parallelism:
  dp: 1
  pp: 1
  pp_engine: 1f1b
  recompute_granularity: SELECTIVE
  tp: 1
  tp_linear_async_communication: true
  tp_mode: REDUCE_SCATTER
profiler: null
tokenizer:
  tokenizer_max_length: null
  tokenizer_name_or_path: openbmb/MiniCPM-2B-dpo-bf16
  tokenizer_revision: null
tokens: null