Files

T

leffffandClaude Sonnet 5 2dbb0c94a9 [feat] Add Kandinsky5 QAD training pipeline: data preprocessing, QAT finetune, QAT-aware DMD distillation

Adds the training-side counterpart to the existing Kandinsky-5 T2V/I2V
inference pipeline (#1471): a full quantization-aware distillation (QAD)
recipe for Kandinsky-5.0 Lite T2V at 512x768/121 frames on the new
fastvideo/train/ stack.

- Data preprocessing: preprocess_kandinsky5_overfit.py encodes raw
  video + captions (Kandinsky5's shared HunyuanVideo VAE + dual
  Qwen/Reason1 + CLIP text encoders) into the t2v parquet schema.
- Stage 1 (Attn-QAT finetune): fastvideo.train.models.kandinsky5.Kandinsky5Model
  wires Kandinsky5 into the modular trainer; FASTVIDEO_ATTENTION_BACKEND=
  ATTN_QAT_TRAIN routes dense/local attention through a fake-quantized
  straight-through-estimator kernel during finetuning so the DiT learns
  to absorb quantization error. No weight quantization during training.
- Stage 2 (QAT-aware DMD2 distillation): Kandinsky5DmdDenoisingStage +
  Kandinsky5DMDPipeline distill the stage-1 checkpoint from many-step
  teacher sampling down to a 4-step student, with teacher/critic masked
  back to full-precision dense attention regardless of the student's
  QAT env var (component_loader.py's _loading_teacher_critic_model gate).

See examples/train/configs/fine_tuning/kandinsky5/README.md for the full
data -> stage-1 -> stage-2 -> inference (NVFP4/FP8) walkthrough.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

2026-07-26 02:29:52 -07:00

configs

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

2026-07-26 02:29:52 -07:00

scenario

[feat] add qat docs (#1621 )

2026-07-19 18:37:47 -07:00

README.md

[feat] add qat docs (#1621 )

2026-07-19 18:37:47 -07:00

run_slurm.sh

[misc]: reorganize training configs and add documentation (#1177 )

2026-03-26 16:38:31 -07:00

run.sh

[feat] Add minimal LoRA finetuning support to the YAML training stack (#1242 )

2026-05-17 14:49:13 -07:00

README.md

Training

Quick Start

Single-node

bash examples/train/run.sh <config.yaml> [--dotted.key value ...]

The script auto-detects available GPUs. Override with NUM_GPUS:

NUM_GPUS=4 bash examples/train/run.sh \
    examples/train/configs/fine_tuning/wan/t2v.yaml

Multi-node (Slurm)

bash examples/train/run_slurm.sh <config.yaml> <num_nodes> [--dotted.key value ...]

bash examples/train/run_slurm.sh \
    examples/train/configs/fine_tuning/wan/t2v.yaml 4 \
    --training.distributed.hsdp_shard_dim 8 \
    --training.distributed.hsdp_replicate_dim 4

Slurm environment variables:

Variable	Default	Description
`PARTITION`	`main`	Slurm partition
`NUM_GPUS`	`8`	GPUs per node
`CPUS_PER_TASK`	`128`	CPUs per task
`MEM`	`1440G`	Memory per node
`EXCLUDE`	`""`	Nodes to exclude

Overriding config values

Any config field can be overridden from the command line using dotted keys:

bash examples/train/run.sh examples/train/configs/fine_tuning/wan/t2v.yaml \
    --training.optimizer.learning_rate 1e-5 \
    --training.loop.max_train_steps 1000 \
    --training.checkpoint.output_dir outputs/my_experiment

Resuming from a checkpoint

bash examples/train/run.sh examples/train/configs/fine_tuning/wan/t2v.yaml \
    --training.checkpoint.resume_from_checkpoint outputs/my_experiment/checkpoint-500

W&B Logging

Training metrics and validation videos are logged to Weights & Biases. Set your API key before launching:

export WANDB_API_KEY=your_key_here

To disable W&B (e.g. for local debugging):

export WANDB_MODE=offline

The project name and run name are set in the config under training.tracker:

training:
  tracker:
    project_name: my_project
    run_name: my_run

Directory Layout

examples/train/
├── configs/      # Single-step training configs (by method and model)
├── scenario/     # Multi-step end-to-end training pipelines
├── run.sh        # Single-node launcher
└── run_slurm.sh  # Multi-node Slurm launcher

See configs/README.md and scenario/README.md for details.

The featured QAD Wan2.1 MixKit scenario runs Attn-QAT supervised fine-tuning, checkpoint export, and three-step DMD2. See the Attn-QAT training guide.