Add files using upload-large-folder tool
Browse files
README.md
CHANGED
@@ -113,13 +113,14 @@ We apply Huffman coding to the exponent bits of BFloat16 model weights, which ar
|
|
113 |
|
114 |
4. To run without CPU offloading (40GB VRAM required):
|
115 |
```bash
|
116 |
-
python t2v.py
|
117 |
```
|
118 |
|
119 |
-
|
120 |
-
|
121 |
-
|
122 |
-
|
|
|
123 |
|
124 |
### 📄 Learn More
|
125 |
|
|
|
113 |
|
114 |
4. To run without CPU offloading (40GB VRAM required):
|
115 |
```bash
|
116 |
+
PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True python t2v.py
|
117 |
```
|
118 |
|
119 |
+
To run with CPU offloading (22.5GB VRAM required):
|
120 |
+
```bash
|
121 |
+
PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True python t2v.py --cpu_offload
|
122 |
+
```
|
123 |
+
> Setting `PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True` is strongly recommended to prevent out-of-memory errors caused by GPU memory fragmentation.
|
124 |
|
125 |
### 📄 Learn More
|
126 |
|