Update README.md
Browse files
README.md
CHANGED
|
@@ -13,8 +13,12 @@ It was trained on closed source datasets containing synthetically generated and
|
|
| 13 |
The model contains the following layout:
|
| 14 |
|
| 15 |
Size: 80B Parameters, A3B (3 Billion Active)
|
|
|
|
| 16 |
Depth: 48 layers
|
|
|
|
| 17 |
Hybrid layout: 12 * (3 * (Gated DeltaNet -> MoE) -> 1 * (Gated Attention -> MoE))
|
|
|
|
| 18 |
MoE: 512 experts, 10 activated experts, 1 shared expert
|
|
|
|
| 19 |
Context Length: 262,144 native
|
| 20 |
|
|
|
|
| 13 |
The model contains the following layout:
|
| 14 |
|
| 15 |
Size: 80B Parameters, A3B (3 Billion Active)
|
| 16 |
+
|
| 17 |
Depth: 48 layers
|
| 18 |
+
|
| 19 |
Hybrid layout: 12 * (3 * (Gated DeltaNet -> MoE) -> 1 * (Gated Attention -> MoE))
|
| 20 |
+
|
| 21 |
MoE: 512 experts, 10 activated experts, 1 shared expert
|
| 22 |
+
|
| 23 |
Context Length: 262,144 native
|
| 24 |
|