# Data Monitor Configuration

# Data directories (auto-detect best location)
data_base_paths:
  - ~/data/suno_data_monitor
  - ~/tmp/suno_data_monitor
  - /tmp/suno_data_monitor

datasets:
  v6_sfx:
    train: /app2/suno/data/auk_v0/metas_t6_tr_w_sfx.jsonl
    val: /app2/suno/data/auk_v0/metas_v6_val.jsonl
  sfx_v3:
    train: /app2/suno/data/diffusion/sfx/v3/metas_tr.jsonl
    val: /app2/suno/data/diffusion/sfx/v3/metas_val.jsonl
  v9:
    train: /app2/suno/data/auk_v0/metas_v9_tr.jsonl
    val: /app2/suno/data/auk_v0/metas_v8_val.jsonl
    sft_ids: /app2/suno/data/auk_v0/ids_keep_sets_v13.json  # To be confirmed
  v8:
    train: /app2/suno/data/auk_v0/metas_v8_tr.jsonl
    val: /app2/suno/data/auk_v0/metas_v8_val.jsonl
    sft_ids: /app2/suno/data/auk_v0/ids_keep_sets_v13.json  # To be confirmed
  v7:
    train: /app2/suno/data/auk_v0/metas_v7_tr.jsonl
    val: /app2/suno/data/auk_v0/metas_v7_val.jsonl
    sft_ids: /app2/suno/data/auk_v0/ids_keep_sets_v12.json  # To be confirmed
  v6:
    train: /app2/suno/data/auk_v0/metas_v6_tr.jsonl
    val: /app2/suno/data/auk_v0/metas_v6_val.jsonl
    sft_ids: /app2/suno/data/auk_v0/ids_keep_sets_v12.json  # To be confirmed
  v5:
    train: /app2/suno/data/auk_v0/metas_v5_tr.jsonl
    val: /app2/suno/data/auk_v0/metas_v5_val.jsonl
    sft_ids: /app2/suno/data/auk_v0/ids_keep_sets_v12.json  # To be confirmed
  v4:
    train: /app2/suno/data/auk_v0/metas_v4_tr.jsonl
    val: /app2/suno/data/auk_v0/metas_v4_val.jsonl
    sft_ids: /app2/suno/data/auk_v0/ids_keep_sets_v12.json  # To be confirmed
  v3:
    train: /app2/suno/data/auk_v0/metas_v3_tr.jsonl
    val: /app2/suno/data/auk_v0/metas_v3_val.jsonl
    sft_ids: /app2/suno/data/auk_v0/ids_keep_sets_v12.json  # To be confirmed
  v2:
    train: /app2/suno/data/auk_v0/metas_v2_tr.jsonl
    val: /app2/suno/data/auk_v0/metas_v2_val.jsonl
    sft_ids: /app2/suno/data/auk_v0/ids_keep_sets_v12.json  # To be confirmed
  v1:
    train: /app2/suno/data/auk_v0/metas_v1_tr.jsonl
    val: /app2/suno/data/auk_v0/metas_v1_val.jsonl
    sft_ids: /app2/suno/data/auk_v0/ids_keep_sets_v12.json  # To be confirmed
  v0:
    train: /app2/suno/data/auk_v0/metas_tr.jsonl
    val: /app2/suno/data/auk_v0/metas_val.jsonl
    sft_ids: /app2/suno/data/auk_v0/ids_keep_sets_v12.json  # To be confirmed

processing:
  chunk_size: 1_000_000  # ~1M records per chunk
  parallel_workers: 64
  memory_limit_gb: 16
  test_size: 100_000     # Size for test runs

# Known value_tags (will discover more dynamically)
known_value_tags:
  - quality
  - spectral_centroid
  - bass
  - mid
  - high
  - stereo_width
  - spectral_flatness
  - crest_factor
  - loudness

# Output formats
output_formats:
  - json
  - csv
  - markdown