Update README.md

More comparison updates

README.md

@@ -28,6 +28,9 @@ This model was designed for roleplaying and storytelling and I think it does wel
     <td>
       <strong>Model Version</strong>
     </td>
+    <td>
+      <strong>EXL2 Perplexity (Default dataset)</strong>
+    </td>
     <td>
       <strong>Description</strong>
     </td>
@@ -37,7 +40,10 @@ This model was designed for roleplaying and storytelling and I think it does wel
       Evathene-v1.0
     </td>
     <td>
-      The original Evathene release based on Athene-V2-Chat and EVA-Qwen2.5-72B-v0.1.
+      ~8.6 @ 4.5 bpw
+    </td>
+    <td>
+      The original Evathene release based on Athene-V2-Chat and EVA-Qwen2.5-72B-v0.1. It's quite solid, but I think the newer versions are better.
     </td>
   </tr>
   <tr>
@@ -45,13 +51,19 @@ This model was designed for roleplaying and storytelling and I think it does wel
       Evathene-v1.1
     </td>
     <td>
-      Updated Evathene release based on Athene-V2-Chat and EVA-Qwen2.5-72B-v0.2. Uses the same recipe as v0.1 but I think it came out a litte better thanks to EVA-v0.2. It's smart and writes competently.
+      ~8.61 @ 4.5 bpw
+    </td>
+    <td>
+      Updated Evathene release based on Athene-V2-Chat and EVA-Qwen2.5-72B-v0.2. Uses the same recipe as v0.1 but I think it came out a litte better thanks to EVA-v0.2. It's smart and writes competently. I think v1.2 improves on its prose, but some users might prefer v1.1's "formal" style, and people might want to use it in their own LLM merge recipes.
     </td>
   </tr>
   <tr>
     <td>
       Evathene-v1.2
     </td>
+    <td>
+      ~8.66 @ 4.5 bpw
+    </td>
     <td>
       A merge of Evathene-v1.1 with Evathene-v1.4. It combines the essence of both models and is the version I recommend for most use cases. It has plenty of personality, is quite smart, and will teach you new words while you're RPing. (You've been warned: its vocabulary is impressive.) With some prompting, you can also get it to channel some of v1.4's energy and writing style, but you should check out v1.4 if you prefer a less formal, more "crazy" experience.
     </td>
@@ -60,6 +72,9 @@ This model was designed for roleplaying and storytelling and I think it does wel
     <td>
       Evathene-v1.4
     </td>
+    <td>
+      ~9.36 @ 4.5 bpw
+    </td>
     <td>
       Evathene based on Athene-V2-Chat and EVA-Qwen2.5-72B-v0.1, but I inverted their relationship in the recipe used for v1.0. The result was a model that has a lot of personality and is great fun in the right context. If you like a lewd ERP writing style or intend to RP with some characters who have big personalities, you'll want to check this one out. You might have to reroll responses more often than with the other versions, but you won't regret it.
     </td>