diff --git a/docs/_config.yml b/docs/_config.yml index 0786a65..ee908de 100644 --- a/docs/_config.yml +++ b/docs/_config.yml @@ -31,5 +31,4 @@ callouts: color: blue best-practice: title: Best Practice - color: green - + color: green \ No newline at end of file diff --git a/docs/core_abstraction/async.md b/docs/core_abstraction/async.md index fd55fa7..087860a 100644 --- a/docs/core_abstraction/async.md +++ b/docs/core_abstraction/async.md @@ -10,12 +10,9 @@ nav_order: 5 **Async** Nodes implement `prep_async()`, `exec_async()`, `exec_fallback_async()`, and/or `post_async()`. This is useful for: 1. **prep_async()**: For *fetching/reading data (files, APIs, DB)* in an I/O-friendly way. - 2. **exec_async()**: Typically used for async LLM calls. - 3. **post_async()**: For *awaiting user feedback*, *coordinating across multi-agents* or any additional async steps after `exec_async()`. - **Note**: `AsyncNode` must be wrapped in `AsyncFlow`. `AsyncFlow` can also include regular (sync) nodes. ### Example diff --git a/docs/core_abstraction/communication.md b/docs/core_abstraction/communication.md index c15af32..85c0276 100644 --- a/docs/core_abstraction/communication.md +++ b/docs/core_abstraction/communication.md @@ -15,7 +15,6 @@ Nodes and Flows **communicate** in two ways: - Great for data results, large content, or anything multiple nodes need. - You shall design the data structure and populate it ahead. - 2. **Params (only for [Batch](./batch.md))** - Each node has a local, ephemeral `params` dict passed in by the **parent Flow**, used as an identifier for tasks. Parameter keys and values shall be **immutable**. - Good for identifiers like filenames or numeric IDs, in Batch mode. @@ -85,7 +84,6 @@ Here: - **Set** via `set_params()`. - **Cleared** and updated each time a parent Flow calls it. - > Only set the uppermost Flow params because others will be overwritten by the parent Flow. > > If you need to set child node params, see [Batch](./batch.md). @@ -125,6 +123,4 @@ flow = Flow(start=node) # 5) Set Flow params (overwrites node params) flow.set_params({"filename": "doc2.txt"}) flow.run(shared) # The node summarizes doc2, not doc1 -``` - ---- +``` \ No newline at end of file diff --git a/docs/core_abstraction/flow.md b/docs/core_abstraction/flow.md index 70dad46..b5e193b 100644 --- a/docs/core_abstraction/flow.md +++ b/docs/core_abstraction/flow.md @@ -86,7 +86,6 @@ flowchart TD - `node.run(shared)`: Just runs that node alone (calls `prep->exec->post()`), returns an Action. - `flow.run(shared)`: Executes from the start node, follows Actions to the next node, and so on until the flow can't continue. - > `node.run(shared)` **does not** proceed to the successor. > This is mainly for debugging or testing a single node. > @@ -108,7 +107,6 @@ A **Flow** is also a **Node**, so it will run `prep()` and `post()`. However: - It **won't** run `exec()`, as its main logic is to orchestrate its nodes. - `post()` always receives `None` for `exec_res` and should instead get the flow execution results from the shared store. - ### Basic Flow Nesting Here's how to connect a flow to another node: @@ -177,5 +175,4 @@ flowchart LR paymentFlow --> inventoryFlow inventoryFlow --> shippingFlow end -``` - +``` \ No newline at end of file diff --git a/docs/core_abstraction/core_abstraction.md b/docs/core_abstraction/index.md similarity index 100% rename from docs/core_abstraction/core_abstraction.md rename to docs/core_abstraction/index.md diff --git a/docs/core_abstraction/node.md b/docs/core_abstraction/node.md index 834e4e4..ce2c23c 100644 --- a/docs/core_abstraction/node.md +++ b/docs/core_abstraction/node.md @@ -13,7 +13,6 @@ A **Node** is the smallest building block. Each Node has 3 steps `prep->exec->po - 1. `prep(shared)` - **Read and preprocess data** from `shared` store. - Examples: *query DB, read files, or serialize data into a string*. @@ -31,14 +30,11 @@ A **Node** is the smallest building block. Each Node has 3 steps `prep->exec->po - Examples: *update DB, change states, log results*. - **Decide the next action** by returning a *string* (`action = "default"` if *None*). - - > **Why 3 steps?** To enforce the principle of *separation of concerns*. The data storage and data processing are operated separately. > > All steps are *optional*. E.g., you can only implement `prep` and `post` if you just need to process data. {: .note } - ### Fault Tolerance & Retries You can **retry** `exec()` if it raises an exception via two parameters when define the Node: @@ -106,5 +102,4 @@ action_result = summarize_node.run(shared) print("Action returned:", action_result) # "default" print("Summary stored:", shared["summary"]) -``` - +``` \ No newline at end of file diff --git a/docs/core_abstraction/parallel.md b/docs/core_abstraction/parallel.md index 747dfbd..ff1664c 100644 --- a/docs/core_abstraction/parallel.md +++ b/docs/core_abstraction/parallel.md @@ -19,7 +19,6 @@ nav_order: 6 > - **Consider Single-Node Batch APIs**: Some LLMs offer a **batch inference** API where you can send multiple prompts in a single call. This is more complex to implement but can be more efficient than launching many parallel requests and mitigates rate limits. {: .best-practice } - ## AsyncParallelBatchNode Like **AsyncBatchNode**, but run `exec_async()` in **parallel**: @@ -54,4 +53,4 @@ class SummarizeMultipleFiles(AsyncParallelBatchFlow): sub_flow = AsyncFlow(start=LoadAndSummarizeFile()) parallel_flow = SummarizeMultipleFiles(start=sub_flow) await parallel_flow.run_async(shared) -``` +``` \ No newline at end of file diff --git a/docs/design_pattern/agent.md b/docs/design_pattern/agent.md index ee699cb..78c1faf 100644 --- a/docs/design_pattern/agent.md +++ b/docs/design_pattern/agent.md @@ -98,5 +98,4 @@ search - "decide" >> decide # Loop back flow = Flow(start=decide) flow.run({"query": "Who won the Nobel Prize in Physics 2024?"}) -``` - +``` \ No newline at end of file diff --git a/docs/design_pattern/paradigm.md b/docs/design_pattern/index.md similarity index 100% rename from docs/design_pattern/paradigm.md rename to docs/design_pattern/index.md diff --git a/docs/design_pattern/mapreduce.md b/docs/design_pattern/mapreduce.md index 7b161ce..fb767a1 100644 --- a/docs/design_pattern/mapreduce.md +++ b/docs/design_pattern/mapreduce.md @@ -14,7 +14,6 @@ MapReduce is a design pattern suitable when you have either: and there is a logical way to break the task into smaller, ideally independent parts. You first break down the task using [BatchNode](../core_abstraction/batch.md) in the map phase, followed by aggregation in the reduce phase. - ### Example: Document Summarization ```python @@ -36,4 +35,4 @@ map_node >> reduce_node # Create flow summarize_flow = Flow(start=map_node) summarize_flow.run(shared) -``` +``` \ No newline at end of file diff --git a/docs/design_pattern/memory.md b/docs/design_pattern/memory.md index e94fbbf..36b23eb 100644 --- a/docs/design_pattern/memory.md +++ b/docs/design_pattern/memory.md @@ -122,4 +122,4 @@ reply - "continue" >> retrieve flow = Flow(start=retrieve) shared = {} flow.run(shared) -``` +``` \ No newline at end of file diff --git a/docs/design_pattern/multi_agent.md b/docs/design_pattern/multi_agent.md index 3f5937f..d427586 100644 --- a/docs/design_pattern/multi_agent.md +++ b/docs/design_pattern/multi_agent.md @@ -70,7 +70,6 @@ Agent received: Network connectivity: stable | timestamp_2 Agent received: Processing load: optimal | timestamp_3 ``` - ### Interactive Multi-Agent Example: Taboo Game Here's a more complex example where two agents play the word-guessing game Taboo. diff --git a/docs/design_pattern/workflow.md b/docs/design_pattern/workflow.md index d7bee94..5601867 100644 --- a/docs/design_pattern/workflow.md +++ b/docs/design_pattern/workflow.md @@ -9,7 +9,6 @@ nav_order: 2 Many real-world tasks are too complex for one LLM call. The solution is to decompose them into a [chain](../core_abstraction/flow.md) of multiple Nodes. - > - You don't want to make each task **too coarse**, because it may be *too complex for one LLM call*. > - You don't want to make each task **too granular**, because then *the LLM call doesn't have enough context* and results are *not consistent across nodes*. > @@ -47,4 +46,4 @@ shared = {"topic": "AI Safety"} writing_flow.run(shared) ``` -For *dynamic cases*, consider using [Agents](./agent.md). +For *dynamic cases*, consider using [Agents](./agent.md). \ No newline at end of file diff --git a/docs/index.md b/docs/index.md index ca019ff..2e635be 100644 --- a/docs/index.md +++ b/docs/index.md @@ -11,7 +11,6 @@ nav_order: 1 A [100-line](https://github.com/the-pocket/PocketFlow/blob/main/pocketflow/__init__.py) minimalist LLM framework for *Agents, Task Decomposition, RAG, etc*. - We model the LLM workflow as a **Graph + Shared Store**: - **Nodes** handle simple (LLM) tasks. - Nodes connect through **Actions** (labeled edges) for *Agents*. @@ -21,17 +20,13 @@ We model the LLM workflow as a **Graph + Shared Store**: - **Batch** Nodes/Flows for data-intensive tasks. - **Async** Nodes/Flows allow waits for asynchronous tasks. -
- - > Have questions? Chat with [AI Assistant](https://chatgpt.com/g/g-677464af36588191b9eba4901946557b-mini-llm-flow-assistant) {: .note } - ## Core Abstraction - [Node](./core_abstraction/node.md) diff --git a/docs/utility_function/llm.md b/docs/utility_function/llm.md index 96cd8c1..b3bc588 100644 --- a/docs/utility_function/llm.md +++ b/docs/utility_function/llm.md @@ -77,7 +77,6 @@ class SummarizeNode(Node): return call_llm(f"Summarize: {text}", self.cur_retry==0) ``` - - Enable logging: ```python @@ -93,4 +92,4 @@ def call_llm(prompt): I believe it is a **bad practice** to provide LLM-specific implementations in a general framework: - **LLM APIs change frequently**. Hardcoding them makes maintenance a nightmare. - You may need **flexibility** to switch vendors, use fine-tuned models, or deploy local LLMs. -- You may need **optimizations** like prompt caching, request batching, or response streaming. +- You may need **optimizations** like prompt caching, request batching, or response streaming. \ No newline at end of file diff --git a/docs/utility_function/tool.md b/docs/utility_function/tool.md index e4dbe19..20664b8 100644 --- a/docs/utility_function/tool.md +++ b/docs/utility_function/tool.md @@ -68,7 +68,6 @@ def execute_sql(query): return result ``` - > ⚠️ Beware of SQL injection risk {: .warning } diff --git a/docs/utility_function/viz.md b/docs/utility_function/viz.md index 26f6c80..c46e1b7 100644 --- a/docs/utility_function/viz.md +++ b/docs/utility_function/viz.md @@ -137,6 +137,4 @@ data_science_flow = DataScienceFlow(start=data_prep_node) data_science_flow.run({}) ``` -The output would be: `Call stack: ['EvaluateModelNode', 'ModelFlow', 'DataScienceFlow']` - - +The output would be: `Call stack: ['EvaluateModelNode', 'ModelFlow', 'DataScienceFlow']` \ No newline at end of file