Davids048andClaude Opus 5.5 c4bb13a650 [refactor]: decide the v1 preprocessing VAE precision during resolution ...

The fp32 VAE that every v1 preprocessing task except text_only needs is a
resolution step (derive_video_preprocess_vae_precision) instead of an
override that the entry point applied after materialization, so the
PipelineConfig carries it too. Sections of a resolved config pickle on
their own, so a DataLoader worker can receive one.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012M91wnVFmPEvJ9h39r5BH7

2026-10-03 06:25:18 +00:00

..

test_components_parquet_saver.py

[self-forcing] [4/n] Preprocessing for collecting ODE trajectory (#788)

2025-09-15 17:54:42 -07:00

test_kandinsky5_e2e_media_oracle.py

[feat] Add Kandinsky5 QAD training pipeline: data preprocessing, QAT finetune, QAT-aware DMD distillation (#1601)

2026-07-26 02:59:21 -07:00

test_kandinsky5_overfit_caption.py

[misc] yapf pass under CI's interpreter (3.12) + pin hook language_version (#1702)

2026-08-08 22:24:17 -07:00

test_kandinsky5_overfit_load_video.py

[feat] Add Kandinsky5 QAD training pipeline: data preprocessing, QAT finetune, QAT-aware DMD distillation (#1601)

2026-07-26 02:59:21 -07:00

test_ltx2_overfit_preprocess.py

[feat] Add LTX-2 and LTX-2.3 fine-tuning to the modular trainer (#1624)

2026-07-20 11:58:39 -07:00

test_minimax_h3_overfit_preprocess.py

[misc] yapf pass under CI's interpreter (3.12) + pin hook language_version (#1702)

2026-08-08 22:24:17 -07:00

test_v1_preprocess_config.py

[refactor]: decide the v1 preprocessing VAE precision during resolution

2026-10-03 06:25:18 +00:00