update title and env loading

ServiceNow · Sep 28, 2024 · ff7a0b8 · ff7a0b8
1 parent efc0b4e
commit ff7a0b8
Show file tree

Hide file tree

Showing 3 changed files with 454 additions and 431 deletions.
diff --git a/example.env b/example.env
@@ -3,5 +3,6 @@
 
 # OPENAI_API_KEY="" # put your https://platform.openai.com/ key here
 # OPENAI_ORGANIZATION="" # optional if you use your personal key
+
 # TOGETHER_API_KEY=""
 # HF_TOKEN=""
diff --git a/examples/intro_clean.ipynb b/examples/intro_clean.ipynb
@@ -89,22 +89,20 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "#### 5. Set your LLM API keys"
+    "#### 0.6. Set your LLM API keys"
    ]
   },
   {
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "In a `.env` file at the root of `TapeAgents` repository:\n",
+    "Set environement variable in a `.env` file at the root of `TapeAgents` repository or set them below\n",
     "\n",
     "```bash\n",
     "# .env file\n",
     "OPENAI_API_KEY=\"\" # put your https://platform.openai.com/ key here\n",
     "# OPENAI_ORGANIZATION=\"\" # optional if you use your personal key\n",
-    "```\n",
-    "\n",
-    "Then run the next block to set environment variable"
+    "```"
    ]
   },
   {
@@ -115,17 +113,8 @@
    "source": [
     "import dotenv\n",
     "\n",
-    "env_file = \"../.env\"\n",
-    "dotenv.load_dotenv(env_file, override=True)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "# You could also set them directly here\n",
+    "env_file = \".env\"\n",
+    "dotenv.load_dotenv(env_file, override=True)\n",
     "\n",
     "# os.environ[\"OPENAI_API_KEY\"] = \"\"\n",
     "# os.environ[\"OPENAI_ORGANIZATION\"] = \"\" # optional if you use your personal key"
@@ -422,7 +411,10 @@
     "# When the agent computes the view, it bumps up `top.next_node` every time it encounters a step with a new `prompt_id``.\n",
     "# The new prompt_id on the tape signals to the agent the current node has run.\n",
     "tape2 = DialogTape(\n",
-    "    steps=[UserStep(content=\"Hi, AI!\"), AssistantStep(metadata=StepMetadata(prompt_id=\"123\"), content=\"AI here, how I can help?\")]\n",
+    "    steps=[\n",
+    "        UserStep(content=\"Hi, AI!\"),\n",
+    "        AssistantStep(metadata=StepMetadata(prompt_id=\"123\"), content=\"AI here, how I can help?\"),\n",
+    "    ]\n",
     ")\n",
     "next_node2 = TapeViewStack.compute(tape2).top.next_node\n",
     "print(next_node2)\n",
@@ -648,7 +640,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 0. Setup your LLM key"
+    "## 3.0. Setup your LLM key"
    ]
   },
   {
@@ -657,7 +649,10 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "# Set more environment variables in the .env file or set them here\n",
+    "# Set more environment variables in the .env file or set them below\n",
+    "\n",
+    "dotenv.load_dotenv(env_file, override=True)\n",
+    "\n",
     "# os.environ[\"TOGETHER_API_KEY\"] = \"\" # put your https://together.ai/ key here\n",
     "# os.environ[\"HF_TOKEN\"] = \"\" # put your huggingface token here\n",
     "\n",
@@ -694,7 +689,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 1 Setup your Agents\n",
+    "## 3.1. Setup your Agents\n",
     "\n",
     "We've found that LLAMA3 function-calling is not yet battle-ready. We will use the structured output approach to make it call tools instead. We are also making this agent trainable by adding `make_llm_output` methods to each node. `Node.make_llm_output` defines how a node can reconstruct the LLM completion message that would be required to make the steps from the given tape at the given index. You can think of `Node.make_llm_output` as the inverse of `Node.generate_steps`.\n",
     "\n",
@@ -848,7 +843,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 2 Run your Agent"
+    "## 3.2. Run your Agent"
    ]
   },
   {