Update model card (step 78750)
Browse files
README.md
CHANGED
|
@@ -22,19 +22,19 @@ Fuses CLIP-L, CLIP-G, and decoupled T5-XL scales into a unified latent space.
|
|
| 22 |
|
| 23 |
- **Fusion Strategy**: adaptive_cantor
|
| 24 |
- **Latent Dimension**: 2048
|
| 25 |
-
- **Training Steps**:
|
| 26 |
-
- **Best Loss**: 0.
|
| 27 |
|
| 28 |
## Learned Parameters
|
| 29 |
|
| 30 |
**Alpha (Visibility):**
|
| 31 |
- clip_g: 0.7291
|
| 32 |
-
- clip_l: 0.
|
| 33 |
- t5_xl_g: 0.7244
|
| 34 |
-
- t5_xl_l: 0.
|
| 35 |
|
| 36 |
**Beta (Capacity):**
|
| 37 |
-
- clip_l_t5_xl_l: 0.
|
| 38 |
- clip_g_t5_xl_g: 0.5744
|
| 39 |
|
| 40 |
|
|
|
|
| 22 |
|
| 23 |
- **Fusion Strategy**: adaptive_cantor
|
| 24 |
- **Latent Dimension**: 2048
|
| 25 |
+
- **Training Steps**: 78,750
|
| 26 |
+
- **Best Loss**: 0.2336
|
| 27 |
|
| 28 |
## Learned Parameters
|
| 29 |
|
| 30 |
**Alpha (Visibility):**
|
| 31 |
- clip_g: 0.7291
|
| 32 |
+
- clip_l: 0.7280
|
| 33 |
- t5_xl_g: 0.7244
|
| 34 |
+
- t5_xl_l: 0.7161
|
| 35 |
|
| 36 |
**Beta (Capacity):**
|
| 37 |
+
- clip_l_t5_xl_l: 0.5726
|
| 38 |
- clip_g_t5_xl_g: 0.5744
|
| 39 |
|
| 40 |
|