model_name: molmo
llm:
  d_model: 3584
  n_heads: 28
  n_kv_heads: 4
  head_dim: null
  qkv_bias: true
  clip_qkv: null
  n_layers: 28
  mlp_ratio: 4
  mlp_hidden_size: 37888
  activation_type: swiglu
  block_type: sequential
  rope: true
  rope_full_precision: true
  rope_theta: 1000000.0
  rope_type: default
  rope_factor: null
  rope_high_freq_factor: null
  rope_low_freq_factor: null
  rope_original_max_position_embeddings: null
  attention_type: sdpa
  float32_attention: true
  attention_dropout: 0.0
  attention_layer_norm: false
  attention_layer_norm_type: olmo
  residual_dropout: 0.1
  response_residual_dropout: 0.0
  layer_norm_type: rms
  layer_norm_with_affine: true
  layer_norm_eps: 1.0e-06
  attention_layer_norm_with_affine: true
  max_sequence_length: 4096
  max_position_embeddings: null
  include_bias: false
  bias_for_layer_norm: null
  norm_after: false
  moe_num_experts: 8
  moe_top_k: 2
  moe_mlp_impl: sparse
  moe_log_expert_assignment: false
  moe_shared_expert: false
  moe_lbl_in_fp32: false
  moe_interleave: false
  moe_loss_weight: 0.1
  moe_zloss_weight: null
  moe_dropless: true
  moe_capacity_factor: 1.25
  embedding_dropout: 0.0
  scale_logits: false
  vocab_size: 152064
  additional_vocab_size: 128
  weight_tying: false
  embedding_size: 152064
  use_position_ids: true
  tokenizer:
    identifier: Qwen/Qwen2.5-7B
    tokenizer_dir: null
    depth_tokens: true
  init_path: gs://mm-olmo/pretrained_llms/qwen2.5-7b.pt
  init_incremental: null
  new_embedding_init_range: 0.02
  initializer_range: 0.02
  normalize_input_embeds: false
  activation_checkpoint: whole_layer
  compile: blocks
  fix_pad_tokenizer: false
  init_std: 0.02
  init_fn: normal
  init_cutoff_factor: null
vision_backbone:
  vit:
    image_model_type: siglip
    image_default_input_size:
    - 378
    - 378
    image_patch_size: 14
    image_pos_patch_size: 14
    image_emb_dim: 1152
    image_num_heads: 16
    image_num_key_value_heads: 16
    image_num_layers: 27
    image_head_dim: 72
    image_mlp_dim: 4304
    image_mlp_activations: gelu_pytorch_tanh
    image_dropout_rate: 0.0
    image_num_pos: 729
    image_norm_eps: 1.0e-06
    attention_dropout: 0.0
    residual_dropout: 0.0
    initializer_range: 0.02
    float32_attention: true
    attention_type: sdpa
    activation_checkpointing: true
    init_path: gs://mm-olmo/pretrained_image_encoders/siglip2-so400m-14-384.pt
    resize_mode: siglip
    pad_value: 0.0
    normalize: siglip
  image_pooling_2d: attention_meanq
  pooling_attention_mask: false
  image_projector: mlp
  image_padding_embed: null
  vit_layers:
  - -3
  - -9
  skip_unused_layers: true
  image_feature_dropout: 0.0
  connector_activation_checkpointing: true
  compile_vit: blocks
data_formatter:
  prompt_templates: uber_model
  message_format: role
  system_prompt: demo_or_style
  always_start_with_space: false
  default_inference_len: 65
  select_answer: best
  debug: false
  image_last: false
  format_message_list: null
  p_one_message: 0.0
mm_preprocessor:
  crop_mode: overlap-and-resize-c2
  max_crops: 8
  max_images: 2
  max_multi_image_crops: 8
  pooling_w: 2
  pooling_h: 2
  overlap_margins:
  - 4
  - 4
  use_col_tokens: true
  loss_token_weighting: root_subsegments
  legacy_image_mask: false
  max_answer_len: null
  img_aug: true
bi_directional_attn: null
lora_enable: true
lora_rank: 32
lora_alpha: 16
lora_dropout: 0.0
lora_bias: none
norm_stats:
  libero_10_no_noops_modified:
    action:
      mean:
      - 0.01820324920117855
      - 0.05858374014496803
      - -0.05592384561896324
      - 0.004626928828656673
      - 0.00289608770981431
      - -0.007673131301999092
      - 0.5457824468612671
      std:
      - 0.2825464606285095
      - 0.35904666781425476
      - 0.3673802614212036
      - 0.03770702704787254
      - 0.05429719388484955
      - 0.08725254982709885
      - 0.49815231561660767
      max:
      - 0.9375
      - 0.9375
      - 0.9375
      - 0.30000001192092896
      - 0.29357144236564636
      - 0.375
      - 1.0
      min:
      - -0.9375
      - -0.9375
      - -0.9375
      - -0.23642857372760773
      - -0.3053571283817291
      - -0.3675000071525574
      - 0.0
      q01:
      - -0.6348214149475098
      - -0.7741071581840515
      - -0.7633928656578064
      - -0.09749999642372131
      - -0.14819999992847435
      - -0.2742857038974762
      - 0.0
      q99:
      - 0.7714285850524902
      - 0.8464285731315613
      - 0.9375
      - 0.13928571343421936
      - 0.15964286029338837
      - 0.3246428668498993
      - 1.0
    proprio:
      mean:
      - -0.04190658777952194
      - 0.03539430722594261
      - 0.8257141709327698
      - 2.908308267593384
      - -0.5562185049057007
      - -0.16649018228054047
      - 0.0
      - 0.028316624462604523
      - -0.028561657294631004
      std:
      - 0.10743364691734314
      - 0.14424669742584229
      - 0.2572328448295593
      - 0.3441362977027893
      - 1.234421730041504
      - 0.3579835891723633
      - 0.0
      - 0.013308707624673843
      - 0.013174631632864475
      max:
      - 0.21031762659549713
      - 0.39128610491752625
      - 1.3332009315490723
      - 3.6714255809783936
      - 3.560650587081909
      - 1.386339545249939
      - 0.0
      - 0.04160946607589722
      - 0.0013633022317662835
      min:
      - -0.4828203022480011
      - -0.3255046010017395
      - 0.445506751537323
      - 1.1321442127227783
      - -3.641430377960205
      - -1.842738389968872
      - 0.0
      - -0.0010040868073701859
      - -0.04111652821302414
      q01:
      - -0.3899900782108307
      - -0.2838300323486328
      - 0.44795057058334353
      - 1.8810229921340942
      - -2.886677579879761
      - -1.1599004411697387
      - 0.0
      - 0.002066459748893976
      - -0.04001387819647789
      q99:
      - 0.1530261474847791
      - 0.32915401458740223
      - 1.2546923208236693
      - 3.303542451858519
      - 2.7496529006957933
      - 0.6893712210655194
      - 0.0
      - 0.040048558115959164
      - -0.0017598449345678235
    num_transitions: 101469
    num_trajectories: 379