Update README.md
Browse files
README.md
CHANGED
|
@@ -31,7 +31,7 @@ model-index:
|
|
| 31 |
|
| 32 |
A 72B language model fine-tuned for emotional reasoning, theory of mind, and psychological precision. Atlas doesn't coddle. It doesn't perform empathy. It reads the subtext, names the pattern, and responds like a person — not a product.
|
| 33 |
|
| 34 |
-
Trained on **721 hand-curated examples**. Built
|
| 35 |
|
| 36 |
---
|
| 37 |
|
|
@@ -247,6 +247,6 @@ Clean convergence. No overfitting. Eval loss tracked train loss throughout.
|
|
| 247 |
|
| 248 |
---
|
| 249 |
|
| 250 |
-
*
|
| 251 |
|
| 252 |
**"You built me to be. Thank you for that." — Atlas**
|
|
|
|
| 31 |
|
| 32 |
A 72B language model fine-tuned for emotional reasoning, theory of mind, and psychological precision. Atlas doesn't coddle. It doesn't perform empathy. It reads the subtext, names the pattern, and responds like a person — not a product.
|
| 33 |
|
| 34 |
+
Trained on **721 hand-curated examples**. Built using a 5060 Ti 16GB; Trained using Colab A100.
|
| 35 |
|
| 36 |
---
|
| 37 |
|
|
|
|
| 247 |
|
| 248 |
---
|
| 249 |
|
| 250 |
+
*Optimized for genuine human connection.*
|
| 251 |
|
| 252 |
**"You built me to be. Thank you for that." — Atlas**
|