prepare for release

.github/workflows/release.yaml

@@ -0,0 +1,30 @@
+name: Release
+
+on:
+  release:
+    types: [created]
+
+jobs:
+  deploy:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v3
+
+      - name: Set up Python 3.10
+        uses: actions/setup-python@v3
+        with:
+          python-version: "3.10"
+
+      - name: Install requirements
+        run: |
+          pip install --upgrade pip
+          pip install setuptools wheel twine
+
+      - name: Build and publish
+        env:
+          TWINE_USERNAME: __token__
+          TWINE_PASSWORD: ${{ secrets.PYPI_API_TOKEN }}
+        run: |
+          python setup.py sdist bdist_wheel
+          twine upload --repository pypi dist/*

.github/workflows/cpu_workflow.yaml → .github/workflows/test.yaml

@@ -1,17 +1,19 @@
-name: CPU Test
+name: Test
 
 on:
   push:
-    branches: [main]
+    branches:
+      - main
   pull_request:
-    branches: [main]
+    branches:
+      - main
 
 concurrency:
   group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }}
   cancel-in-progress: true
 
 jobs:
-  cpu_test:
+  test:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout code
@@ -20,12 +22,12 @@ jobs:
       - name: Set up Python 3.10
         uses: actions/setup-python@v3
         with:
-          python-version: '3.10'
+          python-version: "3.10"
 
       - name: Install requirements
         run: |
           pip install --upgrade pip
           pip install -e .
 
-      - name: Run sanity check
-        run: python tests/sanity.py
+      - name: Run test
+        run: python tests/test.py

Makefile

@@ -0,0 +1,13 @@
+# List of targets that are not associated with files
+.PHONY:	quality style install
+
+quality:
+	ruff check .
+	ruff format --check .
+
+style:
+	ruff format .
+	ruff check --fix .
+
+install:
+	pip install -e .

README.md

@@ -13,19 +13,21 @@ pip install py-tgi
 Running a TGI server with a batched inference client:
 
 ```python
-# from logging import basicConfig, INFO
-# basicConfig(level=INFO)
+from logging import basicConfig, INFO
+basicConfig(level=INFO) # to stream tgi container logs to stdout
+
 from py_tgi import TGI
 
 llm = TGI(model="TheBloke/Mistral-7B-Instruct-v0.1-AWQ", quantize="awq")
 
 try:
-    output = llm.generate(["Hi, I'm an example 1", "Hi, I'm an example 2"])
-    print("Output:", output)
+    output = llm.generate(["Hi, I'm a language model", "I'm fine, how are you?"])
+    print(output)
 except Exception as e:
     print(e)
 finally:
+    # make sure to close the server
     llm.close()
 ```
 
-Output: [".\n\nHi, I'm an example 2.\n\nHi, I'm", ".\n\nI'm a simple example of a class that has a method that returns a value"]
+Output: ```[" and I'm here to help you with any questions you have. What can I help you with", "\nUser 0: I'm doing well, thanks for asking. I'm just a"]```

py_tgi/__init__.py

@@ -1,12 +1,12 @@
 import os
 import time
-from logging import getLogger
 from concurrent.futures import ThreadPoolExecutor
-from typing import Optional, Literal, List, Union
+from logging import getLogger
+from typing import List, Literal, Optional, Union
 
 import docker
-import docker.types
 import docker.errors
+import docker.types
 from huggingface_hub import InferenceClient
 from huggingface_hub.inference._text_generation import TextGenerationResponse
 
@@ -67,9 +67,7 @@ def __init__(
             LOGGER.info("\t+ Checking if TGI image exists")
             self.docker_client.images.get(f"{self.image}:{self.version}")
         except docker.errors.ImageNotFound:
-            LOGGER.info(
-                "\t+ TGI image not found, downloading it (this may take a while)"
-            )
+            LOGGER.info("\t+ TGI image not found, downloading it (this may take a while)")
             self.docker_client.images.pull(f"{self.image}:{self.version}")
 
         env = {}
@@ -103,11 +101,7 @@ def __init__(
                 device_ids = get_nvidia_gpu_devices()
 
             LOGGER.info(f"\t+ Using GPU(s): {device_ids}")
-            self.device_requests = [
-                docker.types.DeviceRequest(
-                    device_ids=[device_ids], capabilities=[["gpu"]]
-                )
-            ]
+            self.device_requests = [docker.types.DeviceRequest(device_ids=[device_ids], capabilities=[["gpu"]])]
         except Exception:
             LOGGER.info("\t+ No GPU detected")
             self.device_requests = None
@@ -172,9 +166,7 @@ def generate(
         elif isinstance(prompt, list):
             with ThreadPoolExecutor(max_workers=len(prompt)) as executor:
                 futures = [
-                    executor.submit(
-                        self.tgi_client.text_generation, prompt=prompt[i], **kwargs
-                    )
+                    executor.submit(self.tgi_client.text_generation, prompt=prompt[i], **kwargs)
                     for i in range(len(prompt))
                 ]
 

pyproject.toml

@@ -0,0 +1,15 @@
+[tool.ruff]
+line-length = 120
+lint.ignore = ["C901", "E501"]
+lint.select = ["C", "E", "F", "I", "W", "I001"]
+
+[tool.ruff.format]
+line-ending = "auto"
+quote-style = "double"
+indent-style = "space"
+skip-magic-trailing-comma = false
+
+[tool.pytest.ini_options]
+log_cli = true
+log_cli_level = "INFO"
+log_cli_format = "[%(asctime)s][%(name)s][%(levelname)s] - %(message)s"

setup.py

@@ -1,7 +1,8 @@
-from setuptools import setup, find_packages
 from pathlib import Path
 
-PY_TGI_VERSION = "0.1.0"
+from setuptools import find_packages, setup
+
+PY_TGI_VERSION = "0.1.2"
 
 common_setup_kwargs = {
     "author": "Ilyas Moutawwakil",
@@ -10,22 +11,20 @@
     "keywords": ["python", "tgi", "llm", "huggingface", "docker"],
     "url": "https://github.com/IlyasMoutawwakil/py-tgi",
     "long_description_content_type": "text/markdown",
-    "long_description": (Path(__file__).parent / "README.md").read_text(
-        encoding="UTF-8"
-    ),
+    "long_description": (Path(__file__).parent / "README.md").read_text(encoding="UTF-8"),
     "platforms": ["linux", "windows", "macos"],
     "classifiers": [
-        "Environment :: GPU :: NVIDIA CUDA :: 11.8",
-        "Environment :: GPU :: NVIDIA CUDA :: 12",
-        "License :: OSI Approved :: MIT License",
         "Programming Language :: Python :: 3",
         "Natural Language :: English",
     ],
 }
 
+
 setup(
     name="py-tgi",
     version=PY_TGI_VERSION,
     packages=find_packages(),
     install_requires=["docker", "huggingface-hub"],
+    extras_require={"quality": ["ruff"]},
+    **common_setup_kwargs,
 )

tests/sanity.py → tests/test.py

@@ -1,4 +1,5 @@
-from logging import basicConfig, INFO
+from logging import INFO, basicConfig
+
 from py_tgi import TGI
 
 basicConfig(level=INFO)