Instructions to use AbstractFramework/wan2.2-t2v-a14b-diffusers-bf16 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use AbstractFramework/wan2.2-t2v-a14b-diffusers-bf16 with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] huggingface-cli download --local-dir wan2.2-t2v-a14b-diffusers-bf16 AbstractFramework/wan2.2-t2v-a14b-diffusers-bf16
- Wan2.2
How to use AbstractFramework/wan2.2-t2v-a14b-diffusers-bf16 with Wan2.2:
# No code snippets available yet for this library. # To use this model, check the repository files and the library's documentation. # Want to help? PRs adding snippets are welcome at: # https://github.com/huggingface/huggingface.js
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
Round validation table values
Browse files
README.md
CHANGED
|
@@ -46,9 +46,9 @@ Bottom line:
|
|
| 46 |
|
| 47 |
| Layout | Disk | MLX Peak | Physical Peak | Time | Result |
|
| 48 |
| --- | ---: | ---: | ---: | ---: | --- |
|
| 49 |
-
| Original source snapshot | 118 GiB |
|
| 50 |
-
| This BF16 package | 64 GiB |
|
| 51 |
-
| Mixed q8/BF16 package | 40 GiB | 20.
|
| 52 |
|
| 53 |
Compared with the original source snapshot, this BF16 package cuts disk usage by about 46% but does not materially reduce generation memory. The mixed q8/BF16 package cuts disk usage by about 66%, MLX peak memory by about 37%, and physical peak memory by about 35% in this validation run.
|
| 54 |
|
|
|
|
| 46 |
|
| 47 |
| Layout | Disk | MLX Peak | Physical Peak | Time | Result |
|
| 48 |
| --- | ---: | ---: | ---: | ---: | --- |
|
| 49 |
+
| Original source snapshot | 118 GiB | 33.0 GiB | 48.9 GiB | 108.3 s | Baseline. |
|
| 50 |
+
| This BF16 package | 64 GiB | 33.0 GiB | 45.1 GiB | 114.4 s | Storage only; output was byte-identical. |
|
| 51 |
+
| Mixed q8/BF16 package | 40 GiB | 20.8 GiB | 31.8 GiB | 110.3 s | Storage and memory; side-by-side quality validation passed. |
|
| 52 |
|
| 53 |
Compared with the original source snapshot, this BF16 package cuts disk usage by about 46% but does not materially reduce generation memory. The mixed q8/BF16 package cuts disk usage by about 66%, MLX peak memory by about 37%, and physical peak memory by about 35% in this validation run.
|
| 54 |
|