162 lines
3.7 KiB
YAML
162 lines
3.7 KiB
YAML
ENV:
|
|
BACKEND: nccl
|
|
SEED: 2024
|
|
#
|
|
SOLVER:
|
|
NAME: ACESolver
|
|
RESUME_FROM:
|
|
LOAD_MODEL_ONLY: True
|
|
USE_FSDP: False
|
|
SHARDING_STRATEGY:
|
|
USE_AMP: True
|
|
DTYPE: float16
|
|
CHANNELS_LAST: True
|
|
MAX_STEPS: 500
|
|
MAX_EPOCHS: -1
|
|
NUM_FOLDS: 1
|
|
ACCU_STEP: 1
|
|
EVAL_INTERVAL: 50
|
|
RESCALE_LR: False
|
|
#
|
|
WORK_DIR: ./cache/save_data/ace_0.6b_512
|
|
LOG_FILE: std_log.txt
|
|
#
|
|
FILE_SYSTEM:
|
|
- NAME: "HuggingfaceFs"
|
|
TEMP_DIR: ./cache/cache_data
|
|
- NAME: "LocalFs"
|
|
TEMP_DIR: ./cache/cache_data
|
|
- NAME: "ModelscopeFs"
|
|
TEMP_DIR: ./cache/cache_data
|
|
|
|
#
|
|
MODEL:
|
|
NAME: LatentDiffusionACE
|
|
PRETRAINED_MODEL:
|
|
IGNORE_KEYS: [ ]
|
|
SCALE_FACTOR: 0.18215
|
|
SIZE_FACTOR: 8
|
|
DECODER_BIAS: 0.5
|
|
DEFAULT_N_PROMPT:
|
|
USE_EMA: True
|
|
EVAL_EMA: False
|
|
TEXT_IDENTIFIER: [ '{image}', '{image1}', '{image2}', '{image3}', '{image4}', '{image5}', '{image6}', '{image7}', '{image8}', '{image9}' ]
|
|
USE_TEXT_POS_EMBEDDINGS: True
|
|
#
|
|
DIFFUSION:
|
|
NAME: BaseDiffusion
|
|
PREDICTION_TYPE: eps
|
|
MIN_SNR_GAMMA:
|
|
NOISE_SCHEDULER:
|
|
NAME: LinearScheduler
|
|
NUM_TIMESTEPS: 1000
|
|
BETA_MIN: 0.0001
|
|
BETA_MAX: 0.02
|
|
#
|
|
DIFFUSION_MODEL:
|
|
NAME: ACE
|
|
PRETRAINED_MODEL: ms://iic/ACE-0.6B-512px@models/dit/ace_0.6b_512px.pth
|
|
IGNORE_KEYS: [ ]
|
|
PATCH_SIZE: 2
|
|
IN_CHANNELS: 4
|
|
HIDDEN_SIZE: 1152
|
|
DEPTH: 28
|
|
NUM_HEADS: 16
|
|
MLP_RATIO: 4.0
|
|
PRED_SIGMA: True
|
|
DROP_PATH: 0.0
|
|
WINDOW_DIZE: 0
|
|
Y_CHANNELS: 4096
|
|
MAX_SEQ_LEN: 1024
|
|
QK_NORM: True
|
|
USE_GRAD_CHECKPOINT: True
|
|
ATTENTION_BACKEND: flash_attn
|
|
#
|
|
FIRST_STAGE_MODEL:
|
|
NAME: AutoencoderKL
|
|
EMBED_DIM: 4
|
|
PRETRAINED_MODEL: ms://iic/ACE-0.6B-512px@models/vae/vae.bin
|
|
IGNORE_KEYS: []
|
|
#
|
|
ENCODER:
|
|
NAME: Encoder
|
|
CH: 128
|
|
OUT_CH: 3
|
|
NUM_RES_BLOCKS: 2
|
|
IN_CHANNELS: 3
|
|
ATTN_RESOLUTIONS: [ ]
|
|
CH_MULT: [ 1, 2, 4, 4 ]
|
|
Z_CHANNELS: 4
|
|
DOUBLE_Z: True
|
|
DROPOUT: 0.0
|
|
RESAMP_WITH_CONV: True
|
|
#
|
|
DECODER:
|
|
NAME: Decoder
|
|
CH: 128
|
|
OUT_CH: 3
|
|
NUM_RES_BLOCKS: 2
|
|
IN_CHANNELS: 3
|
|
ATTN_RESOLUTIONS: [ ]
|
|
CH_MULT: [ 1, 2, 4, 4 ]
|
|
Z_CHANNELS: 4
|
|
DROPOUT: 0.0
|
|
RESAMP_WITH_CONV: True
|
|
GIVE_PRE_END: False
|
|
TANH_OUT: False
|
|
#
|
|
COND_STAGE_MODEL:
|
|
NAME: T5EmbedderHF
|
|
PRETRAINED_MODEL: ms://iic/ACE-0.6B-512px@models/text_encoder/t5-v1_1-xxl/
|
|
TOKENIZER_PATH: ms://iic/ACE-0.6B-512px@models/tokenizer/t5-v1_1-xxl
|
|
LENGTH: 120
|
|
T5_DTYPE: bfloat16
|
|
ADDED_IDENTIFIER: [ '{image}', '{caption}', '{mask}', '{ref_image}', '{image1}', '{image2}', '{image3}', '{image4}', '{image5}', '{image6}', '{image7}', '{image8}', '{image9}' ]
|
|
CLEAN: whitespace
|
|
USE_GRAD: False
|
|
LOSS:
|
|
NAME: ReconstructLoss
|
|
LOSS_TYPE: l2
|
|
#
|
|
SAMPLE_ARGS:
|
|
SAMPLER: ddim
|
|
SAMPLE_STEPS: 20
|
|
GUIDE_SCALE: 4.5
|
|
GUIDE_RESCALE: 0.5
|
|
#
|
|
OPTIMIZER:
|
|
NAME: AdamW
|
|
LEARNING_RATE: 1e-7
|
|
EPS: 1e-10
|
|
WEIGHT_DECAY: 5e-4
|
|
#
|
|
TRAIN_DATA:
|
|
NAME: ImageTextPairMSDatasetForACE
|
|
MODE: train
|
|
MS_DATASET_NAME: cache/datasets/hed_pair
|
|
MS_DATASET_NAMESPACE: ""
|
|
MS_DATASET_SPLIT: "train"
|
|
MS_DATASET_SUBNAME: ""
|
|
PROMPT_PREFIX: ""
|
|
REPLACE_STYLE: False
|
|
MAX_SEQ_LEN: 1024
|
|
PIN_MEMORY: True
|
|
BATCH_SIZE: 1
|
|
NUM_WORKERS: 1
|
|
SAMPLER:
|
|
NAME: LoopSampler
|
|
#
|
|
TRAIN_HOOKS:
|
|
-
|
|
NAME: BackwardHook
|
|
PRIORITY: 0
|
|
-
|
|
NAME: LogHook
|
|
LOG_INTERVAL: 50
|
|
-
|
|
NAME: CheckpointHook
|
|
INTERVAL: 100
|
|
-
|
|
NAME: ProbeDataHook
|
|
PROB_INTERVAL: 100
|