Wan 2.2 TI2V 5b Diffusers BF16
This repository contains BF16 MLX-Gen saved weights for Wan-AI/Wan2.2-TI2V-5B-Diffusers. It is designed for local Apple Silicon inference with mlx-gen.
Modellquelle
- Art
- Other
- Dateigröße
- 1,31–4,64 GB
- Basismodell
- Wan-AI/Wan2.2-TI2V-5B-Diffusers
- Version
- vRevision 6738b2c
- Aktualisiert
- 31. Okt. 2026
- Downloads
- 0
- Quelle
- Hugging Face
Quellenbeschreibung
wan2.2-ti2v-5b-diffusers-bf16
This repository contains BF16 MLX-Gen saved weights for
Wan-AI/Wan2.2-TI2V-5B-Diffusers.
It is designed for local Apple Silicon inference with
mlx-gen.
It uses the mflux/MLX saved-weight layout. It is not a Diffusers or Transformers
from_pretrained() checkpoint.
Source Model
Original model: Wan-AI/Wan2.2-TI2V-5B-Diffusers.
This prepared derivative follows the Apache 2.0 license of the source model.
Precision
The upstream TI2V-5B source snapshot is not uniformly 16-bit on disk: the transformer and VAE safetensors are FP32, while the UMT5 text encoder is BF16. MLX-Gen loads Wan transformer/VAE weights at BF16 runtime precision, so this prepared BF16 package reduces storage and download size but is not a runtime-memory optimization versus source generation.
Use this package when you want a smaller reusable MLX-Gen folder that preserves source behavior. Use the mixed q8/BF16 package when you want a smaller model footprint.
Measurements
Measured on 2026-06-04 with mlx-gen 0.18.10 on an Apple M5 Max with 128 GiB unified memory.
Validation profile: 1280x704, 17 frames, 20 denoising steps, guidance 5, 24 fps, seed 321,
explicit empty negative prompt.
| Layout | Storage | Logical Model | Full-Process Physical Peak | Max RSS | MLX Peak | Total Time | Output |
|---|---|---|---|---|---|---|---|
| Upstream source snapshot | 31.9 GiB | 10.6 GiB | 102.7 GiB | 13.7 GiB | 58.5 GiB | 216.2 s | base-source.mp4 |
| This BF16 package | 21.2 GiB | 10.6 GiB | 102.6 GiB | 14.5 GiB | 58.5 GiB | 261.6 s | prepared-bf16.mp4 |
| Mixed q8/BF16 package | 16.9 GiB | 6.3 GiB | 103.7 GiB | 13.8 GiB | 54.2 GiB | 243.4 s | mixed-q8-bf16.mp4 |
The source and this BF16 package produced byte-identical decoded MP4 frames. The mixed q8/BF16
package stayed visually in the same family with mean frame MAE 1.66 versus source/BF16.
Storage is the Hugging Face repository total. Logical Model is the loaded Wan transformer plus
VAE tensor footprint measured from MLX arrays. Full-Process Physical Peak is Darwin
phys_footprint sampled from model initialization through MP4 save and health validation.
Validation assets:
- contact-sheet.png
- metrics.json
Usage
python -m pip install -U mlx-gen
mlxgen download --model AbstractFramework/wan2.2-ti2v-5b-diffusers-bf16
mlxgen generate \
--model AbstractFramework/wan2.2-ti2v-5b-diffusers-bf16 \
--prompt "A short cinematic video of a glowing orange glass sphere floating above calm teal water, soft reflections, gentle camera movement" \
--negative-prompt "" \
--width 1280 \
--height 704 \
--frames 17 \
--steps 20 \
--guidance 5 \
--fps 24 \
--seed 321 \
--output video.mp4
TI2V-5B also supports first-frame image-to-video in MLX-Gen when one input image is supplied.
Attribution
MLX-Gen is based on mflux by Filip Strand and the original mflux contributors.
Prepared and contributed by @lpalbou.
Quellen
1 Quelle
Quellen
1 QuelleVerifiziert 31. Okt.
Download files
text_encoder · 3 weight files
All files listed in the provider index are present.
- text_encoder/model.safetensors.index.json · provider index
- text_encoder/config.json
- text_encoder/model-00001-of-00003.safetensors · individual shard
- text_encoder/model-00002-of-00003.safetensors · individual shard
- text_encoder/model-00003-of-00003.safetensors · individual shard
transformer · 5 weight files
All files listed in the provider index are present.
- transformer/model.safetensors.index.json · provider index
- transformer/0.safetensors · individual shard
- transformer/1.safetensors · individual shard
- transformer/2.safetensors · individual shard
- transformer/3.safetensors · individual shard
- transformer/4.safetensors · individual shard
vae · 1 weight files
All files listed in the provider index are present.
- vae/model.safetensors.index.json · provider index
- vae/0.safetensors · individual shard
Repository files
File inventory only; completeness is unverified.
Modellartefakte
9 Artefakte
Modellartefakte
9 Artefaktetext_encoder/model-00001-of-00003.safetensors
safetensors · 4,60 GB · SHA-256 a8e861969c74…8f58 · Hugging Face
Download shardtext_encoder/model-00002-of-00003.safetensors
safetensors · 4,64 GB · SHA-256 d57d948ece48…02d7 · Hugging Face
Download shardtext_encoder/model-00003-of-00003.safetensors
safetensors · 1,34 GB · SHA-256 0da9ee284e21…7c5f · Hugging Face
Download shardtransformer/0.safetensors
safetensors · MIXED (unverified) · 2,00 GB · SHA-256 eaf5e225a4db…f23d · Hugging Face
Download filetransformer/1.safetensors
safetensors · BF16 (unverified) · 1,97 GB · SHA-256 79c132d54f16…e904 · Hugging Face
Download filetransformer/2.safetensors
safetensors · 1,99 GB · SHA-256 f9b23d7a502a…d602 · Hugging Face
Download filetransformer/3.safetensors
safetensors · 1,97 GB · SHA-256 44685a00aa01…f9d8 · Hugging Face
Download filetransformer/4.safetensors
safetensors · 1,38 GB · SHA-256 ebd6cce0f41d…d4c3 · Hugging Face
Download fileQuellenauszüge
2 Auszüge
Quellenauszüge
2 Auszüge--- license: apache-2.0 base_model: Wan-AI/Wan2.2-TI2V-5B-Diffusers pipeline_tag: text-to-video library_name: mlx-gen tags: - mlx - mlx-gen - mflux - apple-silicon - bf16 - wan - wan2.2 - video-generation - text-to-video - image-to-video --- # wan2.2-ti2v-5b-diffusers-bf16 This repository contains BF16 MLX-Gen saved weights for [`Wan-AI/Wan2.2-TI2V-5B-Diffusers`](https://huggingface.co/Wan-AI/Wan2.2-TI2V-5B-Diffusers). It is designed for local Apple Silicon inference with [`mlx-gen`](https://github.com/lpalbou/mlx-gen). It uses the mflux/MLX saved-weight layout. It is not a Diffusers or Transformers `from_pretrained()` checkpoint. ## Source Model Original model: [`Wan-AI/Wan2.2-TI2V-5B-Diffusers`](https://huggingface.co/Wan-AI/Wan2.2-TI2V-5B-Diffusers). This prepared derivative follows the Apache 2.0 license of the source model. ## Precision The upstream TI2V-5B source snapshot is not uniformly 16-bit on disk: the transformer and VAE safetensors are FP32, while the UMT5 text encoder is BF16. MLX-Gen loads Wan transformer/VAE weights at BF16 runtime precision, so this prepared BF16 package reduces storage and download size but is not a runtime-memory optimization versus source generation. Use this package when you want a smaller reusable MLX-Gen folder that preserves source behavior. Use the mixed q8/BF16 package when you want a smaller model footprint. ## Measurements Measured on 2026-06-04 with `mlx-gen 0.18.10` on an Apple M5 Max with 128 GiB unified memory. Validation profile: `1280x704`, 17 frames, 20 denoising steps, guidance `5`, 24 fps, seed `321`, explicit empty negative prompt. | Layout | Storage | Logical Model | Full-Process Physical Peak | Max RSS | MLX Peak | Total Time | Output | | --- | ---: | ---: | ---: | ---: | ---: | ---: | --- | | Upstream source snapshot | 31.9 GiB | 10.6 GiB | 102.7 GiB | 13.7 GiB | 58.5 GiB | 216.2 s | [base-source.mp4](validation/ti2v5b-clean/base-source.mp4) | | This BF16 package | 21.2 GiB | 10.6 GiB | 102.6 GiB | 14.5 GiB | 58.5 GiB | 261.6 s | [prepared-bf16.mp4](validation/ti2v5b-clean/prepared-bf16.mp4) | | Mixed q8/BF16 package | 16.9 GiB | 6.3 GiB | 103.7 GiB | 13.8 GiB | 54.2 GiB | 243.4 s | [mixed-q8-bf16.mp4](validation/ti2v5b-clean/mixed-q8-bf16.mp4) | The source and this BF16 package produced byte-identical decoded MP4 frames. The mixed q8/BF16 package stayed visually in the same family wi...
Source context: 0 downloads · 0 likes · Pipeline text-to-video · Library mlx-gen · Repo AbstractFramework/wan2.2-ti2v-5b-diffusers-bf16
