Update README.md

README.md

@@ -4,12 +4,19 @@ license: llama2
 
 Exllama 2 version of the model created by the work of cgato
 
-Original Model https://huggingface.co/cgato/Thespis-13b-v0.2
+Original Model https://huggingface.co/cgato/Thespis-13b-v0.3
 
----
+This model is a bit of a personal project. It uses a vanilla chat template and is focused on providing multiturn sfw and nsfw RP experience. 
+
+This model works best with internet style RP using standard markup with asterisks surrounding actions and no quotes around dialogue.
+
+It uses the following data:
+
+* 3000 samples from Claude Multiround Chat 30k dataset
+* 6000 samples from Pippa Dataset
+* 3000 samples from Puffin Dataset
+* 3800 samples of hand curated RP conversation with various characters.
 
-This model is a bit of a personal project. It uses a vanilla chat template. This current version uses 45k samples from Dolphin and 45k samples from Pippa with with no filtering.
-Then I did a second finetune on top with 2600 rounds of hand written RP conversation with various characters.
 
 Works with standard chat format for Ooba or SillyTavern.
 
@@ -17,9 +24,9 @@ Prompt Format: Chat
 ```
 {System Prompt}
 
-Username: {input}
+Username: {Input}
 BotName: {Response}
-Username: {input}
+Username: {Input}
 BotName: {Response}
 
 ```
@@ -28,5 +35,5 @@ BotName: {Response}
 Turn Template (for Ooba):
 You can either bake usernames into the prompt directly for ease of use or programatically add them if running through the API to use as a chatbot.
 ```
-<|user|>{Your Username}: <|user-message|>\n<|bot|>{BotName}: <|bot-message|>\n
+<|user|>{Username}: <|user-message|>\n<|bot|>{BotName}: <|bot-message|>\n
 ```