You can find detailed information about the versions in these articles:<ul><li><a target="_blank" rel="ugc" href="https://civarchive.com/articles/18990">[Wan 2.1 T2V-14B]</a></li><li><a target="_blank" rel="ugc" href="https://civarchive.com/articles/18999">[Wan 2.2 TI2V-5B]</a></li><li><a rel="ugc" href="https://civarchive.com/articles/23124">[Wan 2.2 T2V-A14B]</a></li></ul>

Beauty of rain [Wan 2.1/2.2]

dataset.toml<pre><code>resolutions = [ [ 342, 192,], [ 854, 480,], [1067, 600]]
enable_ar_bucket = true
min_ar = 0.5
max_ar = 2.0
num_ar_buckets = 7
frame_buckets = [ 1, 16, 30, 33]
[[directory]]
path = "/home/beauty_of_rain_dataset/videos"
num_repeats = 2</code></pre>train.toml<pre><code>output_dir = "/home/beauty_of_rain_dataset/A14B"
dataset = "/home/dataset_v000_854_480_A14B.toml"
epochs = 100
micro_batch_size_per_gpu = 1
pipeline_stages = 1
gradient_accumulation_steps = 1
gradient_clipping = 1
warmup_steps = 10
eval_every_n_epochs = 1
eval_before_first_step = true
eval_micro_batch_size_per_gpu = 1
eval_gradient_accumulation_steps = 1
save_every_n_epochs = 6
activation_checkpointing = true
partition_method = "parameters"
save_dtype = "bfloat16"
caching_batch_size = 1
steps_per_print = 10
video_clip_mode = "single_beginning"
blocks_to_swap = 32

[model]
type = "wan"
ckpt_path = "/home/Wan2.2-T2V-A14B"
dtype = "bfloat16"
transformer_dtype = "float8"
timestep_sample_method = "logit_normal"
transformer_path = "/home/Wan2.2-T2V-A14B/low_noise_model"
min_t = 0
max_t = 0.875

[adapter]
type = "lora"
rank = 32
dtype = "bfloat16"

[optimizer]
type = "adamw_optimi"
lr = 5e-5
betas = [ 0.9, 0.99,]
weight_decay = 0.01
eps = 1e-8</code></pre>

Files with SHA256: 1b6d14bffa9ac5cf5469922e289ab2ae391d15c91aa65ae73a81a0b0a7ece571