Hollow_knight__Wan_2.2

Hollow knight [Wan 2.2]

Creator: Mantissa_Hub Type: LORA Base Model: Wan Video 2.2 TI2V-5B Version: ti2v-5B Trigger Words: hollow_knight

Civitai Model ID: 1836374 Civitai Version ID: 2078128

Stats (at time of fetch for this version):

• Downloads: 95
• Rating: 0 (0 ratings)
• Favorites: N/A

---

📄 Description (Parent Model)

You can find detailed information about the versions in these articles: [Wan 2.2 TI2V-5B] [Wan 2.2 T2V-A14B]

Version Notes (ti2v-5B)

Training Details dataset.toml resolutions = [[ 1280, 704]] enable_ar_bucket = true min_ar = 0.5 max_ar = 2.0 num_ar_buckets = 7 ar_buckets = [[1280, 704]] frame_buckets = [ 1, 81] [[directory]] path = "/home/user/hollow_knight_dataset/videos" num_repeats = 5 train.toml output_dir = "/home/user/hollow_knight/5B" dataset = "/home/user/hollow_knight_dataset.toml" epochs = 40 micro_batch_size_per_gpu = 1 pipeline_stages = 1 gradient_accumulation_steps = 1 gradient_clipping = 1 warmup_steps = 100 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 = 2 activation_checkpointing = true partition_method = "parameters" save_dtype = "bfloat16" caching_batch_size = 1 steps_per_print = 10 video_clip_mode = "single_beginning"

[model] type = "wan" ckpt_path = "/home/user/Wan2.2-TI2V-5B" dtype = "bfloat16" transformer_dtype = "float8" timestep_sample_method = "logit_normal"

[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

---

Civitai Links

• 🔗 View This Version on Civitai →
• View Full Model Page →
• View Creator Profile →

---

File Information

• Filename: hollow_knight_r032_5e-05_wan5B.safetensors
• Size: 153.82 MB
• Hash (AutoV2): D28A0F6D3E
• Hash (SHA256): D28A0F6D3E760AE046ECE83C2F8103C1F75299C6CFD6038EF34EC312B13F3A03