Update README.md

README.md

@@ -35,7 +35,7 @@ The B, C, and D classes are derived from the tokens per model ratio from LLaMA,
 | **[GerbilLab/GerbilBlender-A-77m](https://hf.co/GerbilLab/GerbilBlender-A-77m)** | 77m | A-Class | 20 | 1520M | 262K | 3.3334 | 26.06 | 1908.0661 | 18.2 | 52.09 |
 | **[GerbilLab/GerbilBlender-B-star-77m](https://hf.co/GerbilLab/GerbilBlender-B-star-77m)** | 77m | B*-Class | 40 | 3040M | 262K-524K | 3.1879 | 26.33 | 1766.5002 | 18.24 | 53.43 |
 | **[GerbilLab/GerbilBlender-C-star-77m](https://hf.co/GerbilLab/GerbilBlender-C-star-77m)** | 77m | C*-Class | 60 | 4560M | 262K-524K | coming soon |
-| **[GerbilLab/GerbilBlender-A-104m](https://hf.co/GerbilLab/GerbilBlender-A-104m)** | 104m | A-Class | 20 | 2060M | 1M | 3.592 | 26.41 | 2972.4260 | 17.31 | 49.6 |
+| **[GerbilLab/GerbilBlender-A-104m](https://hf.co/GerbilLab/GerbilBlender-A-104m)** | 104m | A-Class | 20 | 2060M | 1M (too big, would outperform a-77 if was using 524k)| 3.592 | 26.41 | 2972.4260 | 17.31 | 49.6 |
 | --- | --- | --- | --- | --- |  --- |  --- | 
 <!---
 | [GerbilLab/T5Blender-A-24m](https://hf.co/GerbilLab/T5Blender-A-24m) | 24m | A-class | 20 | 460M | 131K | 5.5642 | 25.85 | 57122770.9237 | 0 | 52.25 |