Hugging Face's logo

THUDM
/
CogVideoX1.5-5B-I2V

Image-to-Video
Diffusers
Safetensors
English
CogVideoXImageToVideoPipeline
video-generation
thudm
Model card Files Community
7
zR commited on
Commit
53543e5
•
1 Parent(s): 35e3f31

test

Browse files
Files changed (3) hide show
  1. README.md +3 -2
  2. README_zh.md +2 -2
  3. vae/config.json +2 -1
README.md CHANGED
@@ -41,7 +41,7 @@ Below is a table listing information on the video generation models available in
41
  <tr>
42
  <td style="text-align: center;">Video Resolution</td>
43
  <td colspan="1" style="text-align: center;">1360 * 768</td>
44
- <td colspan="1" style="text-align: center;">256 <= W <=1360<br> 256 <= H <=768<br> W, H % 16 == 0</td>
45
  </tr>
46
  <tr>
47
  <td style="text-align: center;">Inference Precision</td>
@@ -53,7 +53,8 @@ Below is a table listing information on the video generation models available in
53
  </tr>
54
  <tr>
55
  <td style="text-align: center;">Multi-GPU Inference Memory Consumption</td>
56
- <td colspan="2" style="text-align: center;"><b>BF16: 15GB* </b><br></td>
 
57
  </tr>
58
  <tr>
59
  <td style="text-align: center;">Inference Speed<br>(Step = 50, FP/BF16)</td>
 
41
  <tr>
42
  <td style="text-align: center;">Video Resolution</td>
43
  <td colspan="1" style="text-align: center;">1360 * 768</td>
44
+ <td colspan="1" style="text-align: center;"> Min(W, H) = 768 <br> 768 ≤ Max(W, H) ≤ 1360 <br> Max(W, H) % 16 = 0 </td>
45
  </tr>
46
  <tr>
47
  <td style="text-align: center;">Inference Precision</td>
 
53
  </tr>
54
  <tr>
55
  <td style="text-align: center;">Multi-GPU Inference Memory Consumption</td>
56
+ <td colspan="2" style="text-align: center;"><b>BF16: 24GB* using diffusers</b><br></td>
57
+
58
  </tr>
59
  <tr>
60
  <td style="text-align: center;">Inference Speed<br>(Step = 50, FP/BF16)</td>
README_zh.md CHANGED
@@ -27,7 +27,7 @@ CogVideoX是 [清影](https://chatglm.cn/video?fr=osm_cogvideo) 同源的开源
27
  <tr>
28
  <td style="text-align: center;">视频分辨率</td>
29
  <td colspan="1" style="text-align: center;">1360 * 768</td>
30
- <td colspan="1" style="text-align: center;">256 <= W <=1360<br> 256 <= H <=768<br> W,H % 16 == 0</td>
31
  </tr>
32
  <tr>
33
  <td style="text-align: center;">推理精度</td>
@@ -39,7 +39,7 @@ CogVideoX是 [清影](https://chatglm.cn/video?fr=osm_cogvideo) 同源的开源
39
  </tr>
40
  <tr>
41
  <td style="text-align: center;">多GPU显存消耗</td>
42
- <td colspan="2" style="text-align: center;"><b>BF16: 15GB* </b><br></td>
43
  </tr>
44
  <tr>
45
  <td style="text-align: center;">推理速度<br>(Step = 50, FP/BF16)</td>
 
27
  <tr>
28
  <td style="text-align: center;">视频分辨率</td>
29
  <td colspan="1" style="text-align: center;">1360 * 768</td>
30
+ <td colspan="1" style="text-align: center;"> Min(W, H) = 768 <br> 768 ≤ Max(W, H) ≤ 1360 <br> Max(W, H) % 16 = 0 </td>
31
  </tr>
32
  <tr>
33
  <td style="text-align: center;">推理精度</td>
 
39
  </tr>
40
  <tr>
41
  <td style="text-align: center;">多GPU显存消耗</td>
42
+ <td colspan="2" style="text-align: center;"><b>BF16: 24GB* using diffusers</b><br></td>
43
  </tr>
44
  <tr>
45
  <td style="text-align: center;">推理速度<br>(Step = 50, FP/BF16)</td>
vae/config.json CHANGED
@@ -35,5 +35,6 @@
35
  "CogVideoXUpBlock3D"
36
  ],
37
  "use_post_quant_conv": false,
38
- "use_quant_conv": false
 
39
  }
 
35
  "CogVideoXUpBlock3D"
36
  ],
37
  "use_post_quant_conv": false,
38
+ "use_quant_conv": false,
39
+ "invert_scale_latents": true
40
  }