docs: 📜 Updated TODO

TODO

@@ -1,10 +1,21 @@
-Test new embeddings 
 
-Benchmark / rate embeddings & vectors 
+context engineering, - only include vector hits that are x distance?
 
-Is RAG still the "thing"? - What is the cutting edge 
+AI in the middle - make the ai generate the string for vector search
- - "Context Engineering" is the current evolution, although GraphRaG has been a thing?
+
- - Context Engineering seems to be finding the balance of how to provide just the right amount of context to get best results. 
+instruction tuned embeddings?
-    Too little context and the llm doesnt have enough info to give an accurate answer 
+
-    Too much conflicting context (poison)
+entity chunking & re-ranking
-    too much context (confusion)
+
+bredth vs depth = separate workflows
+
+examples into prompts & better prompts
+
+common model attributes - temp & top-k
+
+QA specific embedding models? 
+
+
+
+Evaluation metrics, how good is it doing? 
+rate my response!?