[Cookbook] Add Anyscale Endpoint cookbook

cookbooks/Anyscale/README.md

@@ -0,0 +1,29 @@
+# Anyscale Endpoints with AIConfig
+
+[![colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1JgGjJ2YglyaT5GHQNswkPOyB5oHGbOcv?usp=sharing)
+
+[Anyscale Endpoints](https://www.anyscale.com/endpoints) support optimized inference for many open source models, including the LLaMA2 family of models (7B, 13B, 70B, CodeLLaMA, LLaMA Guard) and Mistral (7B, Mixtral 8x7B).
+
+This cookbook shows how to use any [Anyscale Endpoints](https://www.anyscale.com/endpoints) model with AIConfig using the same simple API.
+
+We cover:
+
+- Inference using Anyscale Endpoints
+- Prompt chains with multiple models
+
+Read more about [AIConfig for prompt and model management](https://github.com/lastmile-ai/aiconfig).
+
+## Models supported with Anyscale Endpoints
+
+For the complete list, please see https://app.endpoints.anyscale.com/docs
+
+- LLaMA-7B: meta-llama/Llama-2-7b-chat-hf
+- LLaMA-13B: meta-llama/Llama-2-13b-chat-hf
+- LLaMA-70B: meta-llama/Llama-2-70b-chat-hf
+- LLaMA Guard: Meta-Llama/Llama-Guard-7b
+- CodeLLaMA: codellama/CodeLlama-34b-Instruct-hf
+- Mistral-7B (OpenOrca): Open-Orca/Mistral-7B-OpenOrca
+- Mistral-7B: mistralai/Mistral-7B-Instruct-v0.1
+- Mixtral-8x7B: mistralai/Mixtral-8x7B-Instruct-v0.1
+- Zephyr: HuggingFaceH4/zephyr-7b-beta
+- GTE: thenlper/gte-large

cookbooks/Anyscale/travel.aiconfig.json

@@ -0,0 +1,55 @@
+{
+  "name": "NYC Trip Planner",
+  "description": "Intrepid explorer with ChatGPT and AIConfig",
+  "schema_version": "latest",
+  "metadata": {
+    "model_parsers": {
+      "meta-llama/Llama-2-7b-chat-hf": "AnyscaleEndpoint",
+      "meta-llama/Llama-2-13b-chat-hf": "AnyscaleEndpoint",
+      "Meta-Llama/Llama-Guard-7b": "AnyscaleEndpoint",
+      "meta-llama/Llama-2-70b-chat-hf": "AnyscaleEndpoint",
+      "Open-Orca/Mistral-7B-OpenOrca": "AnyscaleEndpoint",
+      "codellama/CodeLlama-34b-Instruct-hf": "AnyscaleEndpoint",
+      "HuggingFaceH4/zephyr-7b-beta": "AnyscaleEndpoint",
+      "mistralai/Mistral-7B-Instruct-v0.1": "AnyscaleEndpoint",
+      "mistralai/Mixtral-8x7B-Instruct-v0.1": "AnyscaleEndpoint",
+      "thenlper/gte-large": "AnyscaleEndpoint"
+    },
+    "models": {
+      "meta-llama/Llama-2-7b-chat-hf": {
+        "model": "meta-llama/Llama-2-7b-chat-hf",
+        "top_p": 1,
+        "temperature": 1
+      },
+      "meta-llama/Llama-2-70b-chat-hf": {
+        "model": "meta-llama/Llama-2-70b-chat-hf",
+        "max_tokens": 3000,
+        "system_prompt": "You are an expert travel coordinator with exquisite taste."
+      }
+    },
+    "default_model": "meta-llama/Llama-2-7b-chat-hf"
+  },
+  "prompts": [
+    {
+      "name": "get_activities",
+      "input": "Tell me 10 fun attractions to do in NYC."
+    },
+    {
+      "name": "gen_itinerary",
+      "input": "Generate an itinerary ordered by {{order_by}} for these activities: {{get_activities.output}}.",
+      "metadata": {
+        "model": {
+          "name": "AnyscaleEndpoint",
+          "settings": {
+            "model": "mistralai/Mixtral-8x7B-Instruct-v0.1",
+            "max_tokens": 3000,
+            "system_prompt": "You are an expert travel coordinator with exquisite taste."
+          }
+        },
+        "parameters": {
+          "order_by": "geographic location"
+        }
+      }
+    }
+  ]
+}

cookbooks/Anyscale/travel.aiconfig.yaml

@@ -0,0 +1,53 @@
+description: Intrepid explorer with ChatGPT and AIConfig
+metadata:
+  default_model: meta-llama/Llama-2-7b-chat-hf
+  model_parsers:
+    HuggingFaceH4/zephyr-7b-beta: AnyscaleEndpoint
+    Meta-Llama/Llama-Guard-7b: AnyscaleEndpoint
+    Open-Orca/Mistral-7B-OpenOrca: AnyscaleEndpoint
+    codellama/CodeLlama-34b-Instruct-hf: AnyscaleEndpoint
+    meta-llama/Llama-2-13b-chat-hf: AnyscaleEndpoint
+    meta-llama/Llama-2-70b-chat-hf: AnyscaleEndpoint
+    meta-llama/Llama-2-7b-chat-hf: AnyscaleEndpoint
+    mistralai/Mistral-7B-Instruct-v0.1: AnyscaleEndpoint
+    mistralai/Mixtral-8x7B-Instruct-v0.1: AnyscaleEndpoint
+    thenlper/gte-large: AnyscaleEndpoint
+  models:
+    meta-llama/Llama-2-70b-chat-hf:
+      max_tokens: 3000
+      model: meta-llama/Llama-2-70b-chat-hf
+      system_prompt: You are an expert travel coordinator with exquisite taste.
+    meta-llama/Llama-2-7b-chat-hf:
+      model: meta-llama/Llama-2-7b-chat-hf
+      temperature: 1
+      top_p: 1
+  parameters: {}
+name: NYC Trip Planner
+prompts:
+- input: Tell me 10 fun attractions to do in NYC.
+  name: get_activities
+- input: 'Generate an itinerary ordered by {{order_by}} for these activities: {{get_activities.output}}.'
+  metadata:
+    model:
+      name: AnyscaleEndpoint
+      settings:
+        max_tokens: 3000
+        model: mistralai/Mixtral-8x7B-Instruct-v0.1
+        system_prompt: You are an expert travel coordinator with exquisite taste.
+    parameters:
+      order_by: geographic location
+  name: gen_itinerary
+- input: What should I bring to {{location}}?
+  metadata:
+    model:
+      name: gpt-3.5-turbo
+      settings:
+        model: gpt-3.5-turbo
+        system_prompt:
+          content: You provide a bulleted list of items to pack for a week long trip.
+          role: system
+    parameters:
+      location: nyc
+    remember_chat_context: true
+  name: gen_packing_list
+schema_version: latest

cookbooks/Function-Calling-OpenAI/README.md

@@ -4,6 +4,6 @@
 
 This example is taken from [OpenAI's function calling demo](https://github.com/openai/openai-node/blob/v4/examples/function-call-stream.ts) and modified to show how the same functionality can be done more simply with AIConfig.
 
-TThis notebook serves as a practical guide for leveraging AIConfig and function calling with OpenAI models. We start with a mock database of books and functions to list, search, and retrieve books. Function calling is enabled so the LLM can interpret a user's question, determine the appropriate function to call, and execute the function. Read more about [Function Calling with Open AI](https://openai.com/blog/function-calling-and-other-api-updates) and [AIConfig for prompt and model management](https://github.com/lastmile-ai/aiconfig).
+This notebook serves as a practical guide for leveraging AIConfig and function calling with OpenAI models. We start with a mock database of books and functions to list, search, and retrieve books. Function calling is enabled so the LLM can interpret a user's question, determine the appropriate function to call, and execute the function. Read more about [Function Calling with Open AI](https://openai.com/blog/function-calling-and-other-api-updates) and [AIConfig for prompt and model management](https://github.com/lastmile-ai/aiconfig).
 
 [Google Colab notebook](https://colab.research.google.com/drive/1RZ5s2XmD-Gkg64QlS80lgwOFT3F7J346)

python/src/aiconfig/Config.py

@@ -5,6 +5,7 @@
 
 import requests
 from aiconfig.callback import CallbackEvent, CallbackManager
+from aiconfig.default_parsers.anyscale_endpoint import DefaultAnyscaleEndpointParser
 from aiconfig.default_parsers.openai import DefaultOpenAIParser
 from aiconfig.default_parsers.palm import PaLMChatParser, PaLMTextParser
 from aiconfig.model_parser import InferenceOptions, ModelParser
@@ -35,6 +36,7 @@
 ]
 for model in gpt_models:
     ModelParserRegistry.register_model_parser(DefaultOpenAIParser(model))
+ModelParserRegistry.register_model_parser(DefaultAnyscaleEndpointParser("AnyscaleEndpoint"))
 ModelParserRegistry.register_model_parser(PaLMChatParser())
 ModelParserRegistry.register_model_parser(PaLMTextParser())
 ModelParserRegistry.register_model_parser(HuggingFaceTextGenerationParser())