Add files using upload-large-folder tool

.gitattributes

@@ -33,3 +33,52 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__12_1.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__15_11.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__6_7.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__26_1.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__31_2.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__29_9.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__13_3.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__27_0.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__5_13.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__15_1.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__30_15.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__12_8.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__1_8.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__25_10.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__27_2.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__10_11.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__14_2.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__24_4.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__22_9.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__1_14.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__25_15.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__5_12.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__19_14.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__7_0.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__23_3.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__31_7.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__14_14.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__16_7.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__0_5.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__10_14.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__23_6.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__15_4.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__31_10.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__21_0.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__1_4.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__6_9.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__28_5.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__25_6.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__20_10.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__18_14.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__8_4.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__10_4.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__0_14.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__1_9.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__10_2.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__8_13.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__23_15.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__7_11.distcp filter=lfs diff=lfs merge=lfs -text
+model_and_optim/__0_6.distcp filter=lfs diff=lfs merge=lfs -text

config.yaml

@@ -0,0 +1,831 @@
+run_name: multitask_train
+model:
+  model_name: molmo
+  llm:
+    d_model: 4096
+    n_heads: 32
+    n_kv_heads: 8
+    head_dim: null
+    qkv_bias: false
+    clip_qkv: null
+    n_layers: 36
+    mlp_ratio: 4
+    mlp_hidden_size: 24576
+    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: true
+    attention_layer_norm_type: qwen3
+    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: 151936
+    additional_vocab_size: 128
+    weight_tying: false
+    embedding_size: 151936
+    use_position_ids: true
+    tokenizer:
+      identifier: Qwen/Qwen3-8B
+      tokenizer_dir: null
+    init_path: /weka/oe-training-default/mm-olmo/pretrained_llms/qwen3-8b.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: /weka/oe-training-default/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
+    compile_connector: dynamic
+    normalize_on_gpu: false
+  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
+    eval_system_prompt_mapping: null
+    timestamp_mode: 50-percent-seconds
+    p_choice_content_in_mc: 1.0
+  mm_preprocessor:
+    crop_mode: overlap-and-resize-c2
+    use_col_tokens: true
+    max_crops: 8
+    pooling_w: 2
+    pooling_h: 2
+    overlap_margins:
+    - 4
+    - 4
+    max_images: null
+    max_multi_image_crops: 4
+    max_answer_len: null
+    last_message_loss_only: false
+    loss_token_weighting: root_subsegments
+    max_text_tokens: null
+    image_padding_mask: false
+    legacy_image_mask: false
+  bi_directional_attn: null
+seed: 6198
+epoch: null
+dry_run: false
+ft_llm: true
+ft_vit: true
+ft_connector: true
+ft_embedding: lm_head
+optimizer:
+  name: adamw
+  learning_rate: 0.0001
+  weight_decay: 0.01
+  betas:
+  - 0.9
+  - 0.95
+  eps: 1.0e-05
+  connector_learning_rate: 5.0e-06
+  vit_learning_rate: 5.0e-06
+  llm_learning_rate: 1.0e-05
+  frame_selector_learning_rate: 0.0001
+  temporal_token_scorer_learning_rate: 0.0001
+  connector_weight_decay: 0.0
+  vit_weight_decay: 0.0
+  llm_weight_decay: 0.0
+  frame_selector_weight_decay: 0.01
+  temporal_token_scorer_weight_decay: 0.01
+  connector_betas:
+  - 0.9
+  - 0.95
+  vit_betas:
+  - 0.9
+  - 0.95
+  llm_betas:
+  - 0.9
+  - 0.95
+  frame_selector_betas:
+  - 0.9
+  - 0.95
+  temporal_token_scorer_betas:
+  - 0.9
+  - 0.95
+  connector_eps: 1.0e-06
+  vit_eps: 1.0e-06
+  llm_eps: 1.0e-06
+  frame_selector_eps: 1.0e-06
+  temporal_token_scorer_eps: 1.0e-06
+  metrics_log_interval: -1
+scheduler:
+  name: multimodal
+  units: steps
+  t_warmup: 100
+  t_max: null
+  alpha_f: 0.1
+  connector_t_warmup: 200
+  vit_t_warmup: 200
+  llm_t_warmup: 200
+  frame_selector_t_warmup: 200
+  temporal_token_scorer_t_warmup: 200
+  grad_clip_warmup_steps: null
+  grad_clip_warmup_factor: null
+  warmup_min_lr: 0.0
+data:
+  dataset: null
+  mixture: null
+  root_size_mixture:
+  - rate: 0.15
+    mixture:
+      pixmo_ask_model_anything: null
+      pixmo_cap: 50000.0
+      pixmo_cap_qa_as_user_qa: null
+      pixmo_pointing_explanations: null
+  - rate: 0.5
+    mixture:
+      coco_2014_vqa_multi: null
+      text_vqa: null
+      okvqa: null
+      chart_qa_weighted: null
+      doc_qa: null
+      info_qa: null
+      ai2_diagram_v2_mix_transparent: null
+      a_okvqa_mc: null
+      a_okvqa_da: null
+      android_control: null
+      science_qa_img: null
+      tabwmp_da: null
+      st_qa: null
+      tally_qa: null
+      pixmo_clocks: 250000.0
+      dv_qa: 10000.0
+      figure_qa: 10000.0
+      plot_qa: 20000.0
+      cosyn_chart_exp: null
+      cosyn_chemical_exp: null
+      cosyn_diagram_exp: null
+      cosyn_document: null
+      cosyn_math_exp: null
+      cosyn_music_exp: null
+      cosyn_table_exp: null
+  - rate: 0.35
+    mixture:
+      pixmo_points_train: null
+      pixmo_count_train: null
+      pixmo_points_high_freq_train: null
+      cosyn_point: null
+  kwargs_mixture: null
+  split: train
+  seed: 50189
+  pad: to_max
+  sequence_length: 2304
+  max_text_seq_len: null
+  shuffle: true
+  start_index: 0
+  packing: null
+  num_workers: 2
+  drop_last: true
+  pin_memory: true
+  prefetch_factor: null
+  persistent_workers: false
+  timeout: 0
+restore_dataloader: true
+fast_forward_batches: null
+evaluators: []
+eval_interval: 2000
+inf_evaluators:
+- label: chart_qa
+  data:
+    dataset: chart_qa
+    mixture: null
+    root_size_mixture: null
+    kwargs_mixture: null
+    split: validation
+    seed: 691203
+    pad: to_max
+    sequence_length: 1792
+    max_text_seq_len: null
+    shuffle: true
+    start_index: 0
+    packing: null
+    num_workers: 2
+    drop_last: true
+    pin_memory: true
+    prefetch_factor: null
+    persistent_workers: true
+    timeout: 0
+  evaluator:
+    n_to_log: 0
+    num_wandb_examples: 32
+    save_predictions: null
+    save_tokens: false
+    vqa_eval: relaxed_correctness,scifi_relaxed_correctness,em
+    pointing_eval: false
+    count_eval: false
+    point_count_eval: false
+    android_eval: false
+    clock_eval: false
+    clock_bench_eval: false
+    math_vista_eval: false
+    temp_compass_eval: ''
+    temp_compass_disable_api: false
+    video_mme_eval: ''
+    mlvu_gen_eval: false
+    long_video_bench_eval: false
+    plm_fgqa_eval: false
+    long_video_bench_caption_eval: false
+    vinoground_eval: false
+    vixmo_caption_eval: false
+    refexp_eval: false
+    coco_caption_eval: false
+    qv_highlights_eval: false
+    tomato: false
+    temporal_bench: false
+  max_new_tokens: 12
+  device_batch_size: 4
+  subset_num_batches: null
+  max_examples: 2048
+  console_log_interval: 20
+  include_image: false
+- label: chart_qa_exp
+  data:
+    dataset: chart_qa_exp
+    mixture: null
+    root_size_mixture: null
+    kwargs_mixture: null
+    split: validation
+    seed: 691203
+    pad: to_max
+    sequence_length: 1792
+    max_text_seq_len: null
+    shuffle: true
+    start_index: 0
+    packing: null
+    num_workers: 2
+    drop_last: true
+    pin_memory: true
+    prefetch_factor: null
+    persistent_workers: true
+    timeout: 0
+  evaluator:
+    n_to_log: 0
+    num_wandb_examples: 32
+    save_predictions: null
+    save_tokens: false
+    vqa_eval: relaxed_correctness,scifi_relaxed_correctness,em
+    pointing_eval: false
+    count_eval: false
+    point_count_eval: false
+    android_eval: false
+    clock_eval: false
+    clock_bench_eval: false
+    math_vista_eval: false
+    temp_compass_eval: ''
+    temp_compass_disable_api: false
+    video_mme_eval: ''
+    mlvu_gen_eval: false
+    long_video_bench_eval: false
+    plm_fgqa_eval: false
+    long_video_bench_caption_eval: false
+    vinoground_eval: false
+    vixmo_caption_eval: false
+    refexp_eval: false
+    coco_caption_eval: false
+    qv_highlights_eval: false
+    tomato: false
+    temporal_bench: false
+  max_new_tokens: 256
+  device_batch_size: 4
+  subset_num_batches: null
+  max_examples: 2048
+  console_log_interval: 20
+  include_image: false
+- label: info_qa
+  data:
+    dataset: info_qa
+    mixture: null
+    root_size_mixture: null
+    kwargs_mixture: null
+    split: validation
+    seed: 691203
+    pad: to_max
+    sequence_length: 1792
+    max_text_seq_len: null
+    shuffle: true
+    start_index: 0
+    packing: null
+    num_workers: 2
+    drop_last: true
+    pin_memory: true
+    prefetch_factor: null
+    persistent_workers: true
+    timeout: 0
+  evaluator:
+    n_to_log: 0
+    num_wandb_examples: 32
+    save_predictions: null
+    save_tokens: false
+    vqa_eval: ansl,em
+    pointing_eval: false
+    count_eval: false
+    point_count_eval: false
+    android_eval: false
+    clock_eval: false
+    clock_bench_eval: false
+    math_vista_eval: false
+    temp_compass_eval: ''
+    temp_compass_disable_api: false
+    video_mme_eval: ''
+    mlvu_gen_eval: false
+    long_video_bench_eval: false
+    plm_fgqa_eval: false
+    long_video_bench_caption_eval: false
+    vinoground_eval: false
+    vixmo_caption_eval: false
+    refexp_eval: false
+    coco_caption_eval: false
+    qv_highlights_eval: false
+    tomato: false
+    temporal_bench: false
+  max_new_tokens: 12
+  device_batch_size: 4
+  subset_num_batches: null
+  max_examples: 2048
+  console_log_interval: 20
+  include_image: false
+- label: doc_qa
+  data:
+    dataset: doc_qa
+    mixture: null
+    root_size_mixture: null
+    kwargs_mixture: null
+    split: validation
+    seed: 691203
+    pad: to_max
+    sequence_length: 1792
+    max_text_seq_len: null
+    shuffle: true
+    start_index: 0
+    packing: null
+    num_workers: 2
+    drop_last: true
+    pin_memory: true
+    prefetch_factor: null
+    persistent_workers: true
+    timeout: 0
+  evaluator:
+    n_to_log: 0
+    num_wandb_examples: 32
+    save_predictions: null
+    save_tokens: false
+    vqa_eval: ansl,em
+    pointing_eval: false
+    count_eval: false
+    point_count_eval: false
+    android_eval: false
+    clock_eval: false
+    clock_bench_eval: false
+    math_vista_eval: false
+    temp_compass_eval: ''
+    temp_compass_disable_api: false
+    video_mme_eval: ''
+    mlvu_gen_eval: false
+    long_video_bench_eval: false
+    plm_fgqa_eval: false
+    long_video_bench_caption_eval: false
+    vinoground_eval: false
+    vixmo_caption_eval: false
+    refexp_eval: false
+    coco_caption_eval: false
+    qv_highlights_eval: false
+    tomato: false
+    temporal_bench: false
+  max_new_tokens: 12
+  device_batch_size: 4
+  subset_num_batches: null
+  max_examples: 2048
+  console_log_interval: 20
+  include_image: false
+- label: ai2_diagram
+  data:
+    dataset: ai2_diagram_v2_mix_transparent
+    mixture: null
+    root_size_mixture: null
+    kwargs_mixture: null
+    split: validation
+    seed: 691203
+    pad: to_max
+    sequence_length: 1792
+    max_text_seq_len: null
+    shuffle: true
+    start_index: 0
+    packing: null
+    num_workers: 2
+    drop_last: true
+    pin_memory: true
+    prefetch_factor: null
+    persistent_workers: true
+    timeout: 0
+  evaluator:
+    n_to_log: 0
+    num_wandb_examples: 32
+    save_predictions: null
+    save_tokens: false
+    vqa_eval: mc_ai2d_opaque,mc_ai2d_transparent
+    pointing_eval: false
+    count_eval: false
+    point_count_eval: false
+    android_eval: false
+    clock_eval: false
+    clock_bench_eval: false
+    math_vista_eval: false
+    temp_compass_eval: ''
+    temp_compass_disable_api: false
+    video_mme_eval: ''
+    mlvu_gen_eval: false
+    long_video_bench_eval: false
+    plm_fgqa_eval: false
+    long_video_bench_caption_eval: false
+    vinoground_eval: false
+    vixmo_caption_eval: false
+    refexp_eval: false
+    coco_caption_eval: false
+    qv_highlights_eval: false
+    tomato: false
+    temporal_bench: false
+  max_new_tokens: 12
+  device_batch_size: 4
+  subset_num_batches: null
+  max_examples: 2048
+  console_log_interval: 20
+  include_image: false
+- label: coco_2014_vqa
+  data:
+    dataset: coco_2014_vqa
+    mixture: null
+    root_size_mixture: null
+    kwargs_mixture: null
+    split: validation
+    seed: 691203
+    pad: to_max
+    sequence_length: 1792
+    max_text_seq_len: null
+    shuffle: true
+    start_index: 0
+    packing: null
+    num_workers: 2
+    drop_last: true
+    pin_memory: true
+    prefetch_factor: null
+    persistent_workers: true
+    timeout: 0
+  evaluator:
+    n_to_log: 0
+    num_wandb_examples: 32
+    save_predictions: null
+    save_tokens: false
+    vqa_eval: vqa_score
+    pointing_eval: false
+    count_eval: false
+    point_count_eval: false
+    android_eval: false
+    clock_eval: false
+    clock_bench_eval: false
+    math_vista_eval: false
+    temp_compass_eval: ''
+    temp_compass_disable_api: false
+    video_mme_eval: ''
+    mlvu_gen_eval: false
+    long_video_bench_eval: false
+    plm_fgqa_eval: false
+    long_video_bench_caption_eval: false
+    vinoground_eval: false
+    vixmo_caption_eval: false
+    refexp_eval: false
+    coco_caption_eval: false
+    qv_highlights_eval: false
+    tomato: false
+    temporal_bench: false
+  max_new_tokens: 12
+  device_batch_size: 4
+  subset_num_batches: null
+  max_examples: 2048
+  console_log_interval: 20
+  include_image: false
+- label: pixmo_clocks
+  data:
+    dataset: pixmo_clocks
+    mixture: null
+    root_size_mixture: null
+    kwargs_mixture: null
+    split: validation
+    seed: 691203
+    pad: to_max
+    sequence_length: 1792
+    max_text_seq_len: null
+    shuffle: true
+    start_index: 0
+    packing: null
+    num_workers: 2
+    drop_last: true
+    pin_memory: true
+    prefetch_factor: null
+    persistent_workers: true
+    timeout: 0
+  evaluator:
+    n_to_log: 0
+    num_wandb_examples: 32
+    save_predictions: null
+    save_tokens: false
+    vqa_eval: ''
+    pointing_eval: false
+    count_eval: false
+    point_count_eval: false
+    android_eval: false
+    clock_eval: true
+    clock_bench_eval: false
+    math_vista_eval: false
+    temp_compass_eval: ''
+    temp_compass_disable_api: false
+    video_mme_eval: ''
+    mlvu_gen_eval: false
+    long_video_bench_eval: false
+    plm_fgqa_eval: false
+    long_video_bench_caption_eval: false
+    vinoground_eval: false
+    vixmo_caption_eval: false
+    refexp_eval: false
+    coco_caption_eval: false
+    qv_highlights_eval: false
+    tomato: false
+    temporal_bench: false
+  max_new_tokens: 12
+  device_batch_size: 4
+  subset_num_batches: null
+  max_examples: 2048
+  console_log_interval: 20
+  include_image: false
+- label: android_control_ll
+  data:
+    dataset: android_control_ll
+    mixture: null
+    root_size_mixture: null
+    kwargs_mixture: null
+    split: validation
+    seed: 691203
+    pad: to_max
+    sequence_length: 1792
+    max_text_seq_len: null
+    shuffle: true
+    start_index: 0
+    packing: null
+    num_workers: 2
+    drop_last: true
+    pin_memory: true
+    prefetch_factor: null
+    persistent_workers: true
+    timeout: 0
+  evaluator:
+    n_to_log: 0
+    num_wandb_examples: 32
+    save_predictions: null
+    save_tokens: false
+    vqa_eval: ''
+    pointing_eval: false
+    count_eval: false
+    point_count_eval: false
+    android_eval: true
+    clock_eval: false
+    clock_bench_eval: false
+    math_vista_eval: false
+    temp_compass_eval: ''
+    temp_compass_disable_api: false
+    video_mme_eval: ''
+    mlvu_gen_eval: false
+    long_video_bench_eval: false
+    plm_fgqa_eval: false
+    long_video_bench_caption_eval: false
+    vinoground_eval: false
+    vixmo_caption_eval: false
+    refexp_eval: false
+    coco_caption_eval: false
+    qv_highlights_eval: false
+    tomato: false
+    temporal_bench: false
+  max_new_tokens: 16
+  device_batch_size: 4
+  subset_num_batches: null
+  max_examples: 2048
+  console_log_interval: 20
+  include_image: false
+- label: pointing_eval
+  data:
+    dataset: pointing_eval
+    mixture: null
+    root_size_mixture: null
+    kwargs_mixture: null
+    split: test
+    seed: 691203
+    pad: to_max
+    sequence_length: 1792
+    max_text_seq_len: null
+    shuffle: true
+    start_index: 0
+    packing: null
+    num_workers: 2
+    drop_last: true
+    pin_memory: true
+    prefetch_factor: null
+    persistent_workers: true
+    timeout: 0
+  evaluator:
+    n_to_log: 0
+    num_wandb_examples: 32
+    save_predictions: null
+    save_tokens: false
+    vqa_eval: ''
+    pointing_eval: true
+    count_eval: false
+    point_count_eval: false
+    android_eval: false
+    clock_eval: false
+    clock_bench_eval: false
+    math_vista_eval: false
+    temp_compass_eval: ''
+    temp_compass_disable_api: false
+    video_mme_eval: ''
+    mlvu_gen_eval: false
+    long_video_bench_eval: false
+    plm_fgqa_eval: false
+    long_video_bench_caption_eval: false
+    vinoground_eval: false
+    vixmo_caption_eval: false
+    refexp_eval: false
+    coco_caption_eval: false
+    qv_highlights_eval: false
+    tomato: false
+    temporal_bench: false
+  max_new_tokens: 192
+  device_batch_size: 4
+  subset_num_batches: null
+  max_examples: 2048
+  console_log_interval: 20
+  include_image: false
+inf_eval_interval: 2000
+eval_on_last_step: true
+eval_on_load: false
+save_folder: /weka/oe-training-default/sanghol/molmo/models/uber-v1/uber3.4-synthetic-siglip2-qwen3_8b
+checkpointer_config:
+  save_thread_count: null
+  load_thread_count: null
+  pre_download: false
+  work_dir: null
+  throttle_uploads: false
+canceled_check_interval: 50
+save_interval: 1000
+save_at: null
+save_final_optim: true
+save_num_checkpoints_to_keep: 1
+save_final_unsharded_checkpoint: false
+save_interval_ephemeral: null
+save_overwrite: true
+load_path: null
+reset_optimizer_state: false
+reset_trainer_state: false
+initial_model_checkpoint: /weka/oe-training-default/sanghol/molmo/models/dense-cap-v1/captioner-qwen3_8b/step22347
+allow_resume: true
+max_duration: 30000
+global_train_batch_size: 256
+device_train_microbatch_size: 4
+max_grad_norm: 1.0
+multi_component_grad_norm: true
+batch_divisor: global_batch
+max_grad_norm_ratio: null
+precision: amp_bf16
+wandb:
+  project: molmo2-dev
+  entity: prior-ai2
+  group: uber-v1
+  name: uber3.4-synthetic-siglip2-qwen3_8b
+  tags:
+  - watching
+  log_artifacts: false
+  rank_zero_only: true
+  log_interval: 20
+  allow_resume: false
+beaker_log_interval: 50
+speed_monitor:
+  window_size: 20
+  gpu_flops_available: null
+console_log_interval: 20
+gen1_gc_interval: 1
+compile:
+  mode: default
+  fullgraph: false
+  dynamic: false
+  backend: inductor
+activation_checkpointing: true
+fsdp:
+  fsdp2: true
+  precision: float
+  use_orig_params: true
+  wrapping_strategy: by_block_and_size
+  sharding_strategy: FULL_SHARD
+  hybrid_sharding_num_model_replicas: null
+softmax_auxiliary_loss: true
+softmax_auxiliary_loss_scale: 0.0001
+saliency_score_loss_wt: null
+frame_score_loss_wt: null
+frame_score_loss_type: mse
+frame_score_loss_target: 0.7
+time_limit: null
+extra_steps_after_cancel: 10
+python_profiling: false
+torch_profiling: false
+stop_at: 30000
+stop_after: null
+fused_loss: false
+compile_loss: true
+runtime_data:
+  args: /gantry-runtime/launch_scripts/train_multitask_model.py 3.4-synthetic /weka/oe-training-default/sanghol/molmo/models/dense-cap-v1/captioner-qwen3_8b/step22347
+    --save_overwrite --save_interval=1000 --wandb.group=uber-v1 --wandb.name=uber3.4-synthetic-siglip2-qwen3_8b
+    --save_folder=/weka/oe-training-default/sanghol/molmo/models/uber-v1/uber3.4-synthetic-siglip2-qwen3_8b
+  hostname: jupiter-cs-aus-109.reviz.ai2.in
+  date: 09/13/2025, 07:25
+  world_size: 32
+  resuming_from: null
+  beaker_experiment_id: 01K50Z8VKCVN9Z4X917WEEP9VQ
+  beaker_experiment_url: https://beaker.org/ex/01K50Z8VKCVN9Z4X917WEEP9VQ
+  wandb_id: xhs8xbpk
+  wandb_url: https://wandb.ai/prior-ai2/molmo2-dev/runs/xhs8xbpk

model_and_optim/__0_14.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:04ebf7580a878e6b6b25f9bf03dc8597d8ecf9bfbd0c3dd4bb0aa193860bb2c1
+size 193681551

model_and_optim/__0_5.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7c4d8ee380591eda64ade07e10fbc2915838318d75ebc412993b85fcaed0c229
+size 193561739

model_and_optim/__0_6.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ccbdb8a6f3a059c2a3da60e3320b5f0f19af2315b3302d6367f5fe0c0d2dd5e1
+size 193581259

model_and_optim/__10_11.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e14545b3df6bb160f9b7348806c2d8e711277125ecda2d9dc57725eaecb99e3c
+size 193704827

model_and_optim/__10_14.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:148c41bb91946fae036bc2b2f9f882c677035354a45141ae05cfdb5fc3684a5f
+size 193578101

model_and_optim/__10_2.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fb7f0caa2ef2eeb61931f0bbf6974c13f1c07d5f8601285a9162e01d2352d4a9
+size 193461027

model_and_optim/__10_4.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5c6f6820e1adb60e03289103e69762c7acb878fb01e5a96382551a83b26b508f
+size 193461636

model_and_optim/__12_1.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:73b07b57f992282ab233063cd08032a803ce6c6021b5cfa6811782d98b6592d2
+size 193467212

model_and_optim/__12_8.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b856c459d4c6897d915b713cf70df4e8dffb0a49a276d3399be8f14c25418558
+size 193401014

model_and_optim/__13_3.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:466278110b0058f39efbf5fe08cb40453f5b6173563e9f01caa48befb99e0e1b
+size 193451811

model_and_optim/__14_14.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a70b1182d865ca32ef338c936bdc259c44d25c2cf6ef09ff85a5a94ecae6853a
+size 193578101

model_and_optim/__14_2.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ed0c549dee6732cd6f129c43e6455eb7e6fc55c78fa9a4d3321883c8d95e0648
+size 193461027

model_and_optim/__15_1.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:35bec598dee2e6c081d1623745a076f867c727dbd5630e06ad3bf05b2699770c
+size 193467212

model_and_optim/__15_11.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4e79da0e258b60db86b8025c07b993ed35c2ebc79dd99f00da0a2e74d08820f4
+size 193704827

model_and_optim/__15_4.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d82d9a894c446de2bfc3d9ba35dfd32bc15e0c0194db01967d1bcada94e2c568
+size 193461636

model_and_optim/__16_7.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ec66a54d4ef5180acba44c094e6110a175ffdc7b4187ed7ef257d5c6f74d2682
+size 193440900

model_and_optim/__18_14.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c71d00cfee908af783bb5c157595795986a93e59a5a8b4cdc4d4252cf4267cd8
+size 193578101

model_and_optim/__19_14.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:75826ba0bcab9c19075e4f4c1b40b8d904c1db80c8873fff3b968a42c3887cd2
+size 193578101

model_and_optim/__1_14.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3aa839e1d98da1656b85afa368d740a95c62311670936111f7d321b470b6a195
+size 193578101

model_and_optim/__1_4.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a8dface0174b05bb668161ac4e61e7bb64abc6ad4c8ca7d3f122c84aa18c5a9c
+size 193461636

model_and_optim/__1_8.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5bbcc23e3f583f537bb5b40706bf137304efb040091b8e52a7186a2db12481a7
+size 193401014

model_and_optim/__1_9.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e3f6682e1dba102c645e0c6ccc9a9c758b9d4854a715d13fc9bc3dbff7920e97
+size 193394102

model_and_optim/__20_10.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c9c17d9af2524a42c9a9bdd1b531bdf5d75ecc6e5b4688377ae4f4a7f17de5da
+size 193427326

model_and_optim/__21_0.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dfff6720de93ee9c2ff8f1efde3229d6070dcedd51f9dfa07d45a6075acf1f9c
+size 193405798

model_and_optim/__22_9.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cbc6f09fd3b0c8d36f73710f8188c71ae3600e2c0434d609345c244c5951c66a
+size 193394102

model_and_optim/__23_15.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dc0f98a106de9cad78141f361b19b51c21b341b41aac1c70c2df0a18cb938590
+size 193567308

model_and_optim/__23_3.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d87d535f75281a382008936bc27387d2ab2d4ff3465382bcff7b61be6d58a783
+size 193451811

model_and_optim/__23_6.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1e93f60ea866036c45c0929faeaa65668d3a5b1a25f4cb2c14cc5bd71e08b5c1
+size 193445508

model_and_optim/__24_4.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f410c1c76b8672a70b374d69ea6678f4b82b78c56468acc07ddd570a83712604
+size 193461636

model_and_optim/__25_10.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2e99a2f1f661117de8cfdf2298d02e6b863ae48fa36c19ee7ff6f8fa67cf40f0
+size 193427326

model_and_optim/__25_15.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f7d75adac4dca2715d8d8b252ac39afd9f6b8a268ad7a4156e32ed7f23184dba
+size 193567308

model_and_optim/__25_6.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9786ada91a5609b13330d09ca950716c1fb4280b5e106626e68b97444784fb1e
+size 193445508

model_and_optim/__26_1.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:916dd8cfcbe1200ce346c68d070fb00ca9e28c573a841546c792282b4899adf1
+size 193467212

model_and_optim/__27_0.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d059390b063cdbb68876351d35f82feb83799490b8fae054395e0b0cb3d13225
+size 193405798

model_and_optim/__27_2.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d51c741c4cfad841a1d5f37c6f7c90c3ff95f63fea30984d28937312460b4cdc
+size 193461027

model_and_optim/__28_5.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:84aef054ca4035e2ca0032722a7058de8ba4f64030338524812070a0fadcb3e5
+size 193457028

model_and_optim/__29_9.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c4655890b30b0ce7ff2af03b8378cfa84d5ad7237b7abf063aae8efde98bbb61
+size 193394102

model_and_optim/__30_15.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:23822029b0e6d4a7f6e74bda0831e7b64df8a546afb1e9f651037a087a2902ac
+size 193567308

model_and_optim/__31_10.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ff5295ce59390a8b2842e1e879a1c774f67bec154f8dfb2813982a18ccd5b895
+size 192763774

model_and_optim/__31_2.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0cfa6c1a8fe5add6ab11d8fa4c4b3a3010684f866cccbd2cb34668a8257b5f06
+size 193166115

model_and_optim/__31_7.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:81cce0c5804b272b2efa38e2fc4c4335f5e09c695463f60616a5f66f240e7f03
+size 193440900

model_and_optim/__5_12.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0d5cf25e01585e94eb927d3082bad45dec8f94f2f7a8643b9c163bc198447e1d
+size 193697915

model_and_optim/__5_13.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a5eacba755bb6c4756617c3653321edd3229d4de5d44b89a1d83aef2828faad5
+size 193691003

model_and_optim/__6_7.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5d9d1f8a6bca0796bce4c7315b14f585ba088e68efb2d44696dae03c55b57836
+size 193440900

model_and_optim/__6_9.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9454663677d49650300a9b109ed9b85a5d782aa49375ac3d515ed8e0458f375c
+size 193394102

model_and_optim/__7_0.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d0b38a1471fa25b831aa8f8f4bbaae15614b7ce3985e6283a0990670e7e59bcb
+size 193405798

model_and_optim/__7_11.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:245c3ef961fa27a43f67a434e3dd2fbd6cf139dbdeb0952ace67bf82d223b2df
+size 193704827

model_and_optim/__8_13.distcp

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:875775552bc1c90ca51ffb4624128576a63d49a3dc709dd9946adbf5b1418438
+size 193691003