[AIC-py] usability fixes for llama cookbook

Summary
- expose CLI args for paths for easier usage
- use extension pypi package
- add readme for installation + usage

cookbooks/llama/python/README.md

@@ -0,0 +1,16 @@
+Example app using llama aiconfig extension.
+
+1. Install (example using anaconda)
+   `conda create -n aiconfig-llama-cookbook`
+   `conda activate aiconfig-llama-cookbook`
+   `conda install pip`
+   `pip install -r python/requirements.txt`
+
+2. Download a model, e.g.
+   `curl -L https://huggingface.co/TheBloke/Llama-2-7B-Chat-GGUF/resolve/main/llama-2-7b-chat.Q4_K_M.gguf --output ./models/llama-2-7b-chat.Q4_K_M.gguf`
+3. cd into cookbook root dir
+   `$ pwd`
+   aiconfig/cookbooks/llama
+4. Create an AIConfig like this: https://github.com/lastmile-ai/aiconfig/blob/e92e5a3c80b9c2b74a9432f0441318a951d54d0c/cookbooks/llama/llama-aiconfig.json
+5. Run with your local paths:
+   `python python/ask_llama.py --aiconfig-path='../llama/llama-aiconfig.json' --model-path='../../models/llama-2-7b-chat.Q4_K_M.gguf' 2> ask-llama.err`

cookbooks/llama/python/ask_llama.py

@@ -1,15 +1,33 @@
 import asyncio
+import sys
 
 from aiconfig.model_parser import InferenceOptions
 from llama import LlamaModelParser
 
 from aiconfig import AIConfigRuntime
+import argparse
 
 
 async def main():
-    llama_model_parser = LlamaModelParser(
-        model_path="models/llama-2-7b-chat.Q4_K_M.gguf"
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "--aiconfig-path",
+        type=str,
+        required=True,
+        help="Relative or absolute path to aiconfig json, e.g. cookbooks/llama/llama-aiconfig.json",
     )
+    parser.add_argument(
+        "--model-path",
+        type=str,
+        required=True,
+        help="Relative or absolute path to model",
+    )
+    args = parser.parse_args()
+    return await run(args.aiconfig_path, args.model_path)
+
+
+async def run(aiconfig_path: str, model_path: str):
+    llama_model_parser = LlamaModelParser(model_path=model_path)
 
     for lm in [
         "llama-2-7b-chat",
@@ -18,7 +36,7 @@ async def main():
     ]:
         AIConfigRuntime.register_model_parser(llama_model_parser, lm)
 
-    config = AIConfigRuntime.load("cookbooks/llama/llama-aiconfig.json")
+    config = AIConfigRuntime.load(aiconfig_path)
 
     def stream_callback(data, accumulated_message, index):
         print(data, end="", flush=True)
@@ -38,9 +56,9 @@ def stream_callback(data, accumulated_message, index):
     await config.run("prompt13b", params={}, options=inference_options)
 
     print("\n\nRunning prompt13b_code...")
-    code_res = await config.run("prompt13b_code", params={}, options=inference_options)
-    print(f"\n\n\n\nCode response:\n{code_res}")
+    await config.run("prompt13b_code", params={}, options=inference_options)
 
 
 if __name__ == "__main__":
-    asyncio.run(main())
+    res = asyncio.run(main())
+    sys.exit(res)

cookbooks/llama/python/pyproject.toml

@@ -0,0 +1,43 @@
+[build-system]
+requires = ["setuptools", "wheel"]
+
+[project]
+name = "python-aiconfig-llama-cookbook"
+version = "0.0.1"
+authors = [
+    { name="Jonathan Lessinger", email="[email protected]" },
+]
+description = "LLama cookbook using AIConfig Library"
+readme = "README.md"
+requires-python = ">=3.7"
+classifiers = [
+  "Intended Audience :: Developers",
+  "Programming Language :: Python :: 3",
+]
+dynamic = ["dependencies"]
+
+[tool.setuptools.dynamic]
+dependencies = {file = ["requirements.txt"]}
+
+[project.urls]
+"Homepage" = "https://github.com/lastmile-ai/aiconfig"
+"Bug Tracker" = "https://github.com/lastmile-ai/aiconfig/issues"
+
+# Black formatting
+[tool.black]
+line-length = 99
+include = '\.pyi?$'
+exclude = '''
+/(
+      .eggs         # exclude a few common directories in the
+    | .git          # root of the project
+    | .hg
+    | .mypy_cache
+    | .tox
+    | venv
+    | _build
+    | buck-out
+    | build
+    | dist
+  )/
+'''

cookbooks/llama/python/requirements.txt

@@ -0,0 +1,2 @@
+python-aiconfig-llama
+python-aiconfig