update docs

docs/decomp.md

@@ -0,0 +1,40 @@
+---
+layout: default
+title: "Task Decomposition"
+parent: "Paradigm"
+nav_order: 2
+---
+
+# Task Decomposition
+
+Many real-world tasks are too complex for one LLM call. The solution is to decompose them into multiple calls as a [Flow](./flow.md) of Nodes.
+
+### Example: Article Writing
+
+```python
+class GenerateOutline(Node):
+    def exec(self, topic):
+        prompt = f"Create a detailed outline for an article about {topic}"
+        return call_llm(prompt)
+
+class WriteSection(Node):
+    def exec(self, section):
+        prompt = f"Write content for this section: {section}"
+        return call_llm(prompt)
+
+class ReviewAndRefine(Node):
+    def exec(self, draft):
+        prompt = f"Review and improve this draft: {draft}"
+        return call_llm(prompt)
+
+# Connect nodes  
+outline = GenerateOutline()
+write = WriteSection()
+review = ReviewAndRefine()
+
+outline >> write >> review
+
+# Create flow
+writing_flow = Flow(start=outline)
+writing_flow.run({"topic": "AI Safety"})
+```

docs/index.md

@@ -41,19 +41,19 @@ We model the LLM workflow as a **Nested Directed Graph**:
 
 - [LLM Wrapper](./llm.md)
 - [Tool](./tool.md)
+- Chunking
 
-
+> We do not provide built-in implementations. 
-
+> Example implementations are provided as reference.
-> We do not provide built-in implementation for low-level details. Example implementations are provided as reference.
 {: .warning }
 
 
 ## High-Level Paradigm
 
 - [Structured Output](./structure.md)
-- Task Decomposition
+- [Task Decomposition](./decomp.md)
-- Map Reduce
+- [Map Reduce](./mapreduce.md)
-- RAG
+- [RAG](./rag.md)
 - Chat Memory
 - Agent
 - Multi-Agent
@@ -62,3 +62,4 @@ We model the LLM workflow as a **Nested Directed Graph**:
 ## Example Projects
 
 - [Summarization + QA agent for Paul Graham Essay](./essay.md)
+- More coming soon...

docs/llm.md

@@ -53,7 +53,7 @@ def call_llm(prompt):
     pass
 ```
 
-> ⚠️ May overlap with Node retries by caching LLM responses
+> ⚠️ Caching conflicts with Node retries, as retries yield the same result.
 {: .warning }
 
 

docs/mapreduce.md

@@ -0,0 +1,45 @@
+---
+layout: default
+title: "Map Reduce"
+parent: "Paradigm"
+nav_order: 3
+---
+
+# Map Reduce
+
+Process large inputs by splitting them into chunks using [BatchNode](./batch.md), then combining results.
+
+### Example: Document Summarization
+
+```python
+class MapSummaries(BatchNode):
+    def prep(self, shared):
+        text = shared["text"]
+        return [text[i:i+10000] for i in range(0, len(text), 10000)]
+
+    def exec(self, chunk):
+        return call_llm(f"Summarize this chunk: {chunk}")
+
+    def post(self, shared, prep_res, exec_res_list):
+        shared["summaries"] = exec_res_list
+
+class ReduceSummaries(Node):
+    def prep(self, shared):
+        return shared["summaries"]
+        
+    def exec(self, summaries):
+        return call_llm(f"Combine these summaries: {summaries}")
+        
+    def post(self, shared, prep_res, exec_res):
+        shared["final_summary"] = exec_res
+
+# Connect nodes
+map_node = MapSummaries()
+reduce_node = ReduceSummaries()
+
+map_node >> reduce_node
+
+# Create flow 
+summarize_flow = Flow(start=map_node)
+summarize_flow.run(shared)
+```

docs/rag.md

@@ -0,0 +1,47 @@
+---
+layout: default
+title: "RAG"
+parent: "Paradigm" 
+nav_order: 4
+---
+
+# RAG (Retrieval Augmented Generation)
+
+For certain LLM tasks like answering questions, providing context is essential.
+Use [vector search](./tool.md) to find relevant context for LLM responses.
+
+## Example: Question Answering
+
+```python
+class PrepareEmbeddings(Node):
+    def prep(self, shared):
+        texts = shared["texts"]
+        embeddings = [get_embedding(text) for text in texts]
+        shared["search_index"] = create_index(embeddings)
+
+class AnswerQuestion(Node):
+    def prep(self, shared):
+        question = input("Enter question: ")
+        query_embedding = get_embedding(question)
+        indices, _ = search_index(shared["search_index"], query_embedding, top_k=1)
+        relevant_text = shared["texts"][indices[0][0]]
+        return question, relevant_text
+
+    def exec(self, inputs):
+        question, context = inputs
+        prompt = f"Question: {question}\nContext: {context}\nAnswer: "
+        return call_llm(prompt)
+
+    def post(self, shared, prep_res, exec_res):
+        print(f"Answer: {exec_res}")
+
+# Connect nodes
+prep = PrepareEmbeddings()
+qa = AnswerQuestion()
+
+prep >> qa
+
+# Create flow
+qa_flow = Flow(start=prep)
+qa_flow.run(shared)
+```