Upload folder using huggingface_hub

README.md

@@ -2,6 +2,9 @@
 
 A LoRA adapter for unsloth/gemma-2b that determines content indexing suitability using chain-of-thought reasoning.
 
+Note: This is used in a pipeline to determine if a context should be indexed or not.
+
+
 ## Technical Specifications
 
 ### Base Model
@@ -19,16 +22,21 @@ Input XML structure:
 <instruction>Determine true or false if the following content is suitable and should be indexed.</instruction>
 <suitable>
   <content>{input_text}</content>
-</suitable>
 ```
 
 Output XML structure:
 ```xml
-<thinking>{reasoning_process}</thinking>
-<category>{content_type}</category>
-<should_index>{true|false}</should_index>
+  <thinking>{reasoning_process}</thinking>
+  <category>{content_type}</category>
+  <should_index>{true|false}</should_index>
+</suitable>
+
 ```
 
+The model then expects an indefinite list of following <suitable> ... </suitable> so you can few shot it to correct a mistake or enhance the results.
+
+Your stop token should be `</suitable>` to ensure the model stops generating after the output.
+
 ## Deployment
 
 ### VLLM Server Setup