Update package settings in pyproject.toml and LICENSE

.github/pull_request_template.md

@@ -0,0 +1,28 @@
+# PR Description
+
+Fixes # (issue number)
+
+## Summary
+
+Provide a concise and clear summary of the changes and their relation to the reported issue.
+
+## Motivation and Context
+
+Explain the reasoning behind the changes and provide relevant context for a better understanding of the modifications.
+
+## Dependencies
+
+Enumerate any new dependencies necessary for implementing these changes.
+
+## Type of Change
+
+Specify the appropriate type of change and remove irrelevant options.
+
+- Bug fix
+- New feature
+- Breaking change
+- Documentation update
+
+## Testing Procedure
+
+Please describe the testing process undertaken to validate these changes.

LICENSE

@@ -1,5 +1,3 @@
-Copyright (c) 2022-present, FriendliAI Inc. All rights reserved.
-
                                  Apache License
                            Version 2.0, January 2004
                         http://www.apache.org/licenses/

docs/docs/cli/intro.mdx

@@ -2,7 +2,7 @@
 sidebar_position: 1
 ---
 
-# 🤗 Welcome to PeriFlow
+# 🚀 Welcome to PeriFlow
 
 PeriFlow is the fastest engine for serving generative AI models such as GPT-3.
 With PeriFlow, a company can significantly reduce the cost and environmental impact of running its generative AI models.

docs/docs/tutorials/intro.mdx

@@ -0,0 +1,151 @@
+---
+sidebar_position: 1
+---
+
+# How to Use Your Checkpoint?
+
+## Introduction
+
+PeriFlow allows you to easily upload your own AI model checkpoint from your local machine to the desired PeriFlow project.
+This guide will walk you through the steps to upload the checkpoint using the PeriFlow Client.
+
+## Prerequisites
+
+Before you begin, make sure you have the PeriFlow Client installed.
+You can install it with the following command, which also includes the necessary machine learning library (`mllib`) dependencies:
+
+```sh
+pip install periflow-client[mllib]
+```
+
+:::info
+### Converting Hugging Face Checkpoint (Optional)
+
+If you have a Hugging Face checkpoint and want to convert it to a PeriFlow-compatible format, you need to install the package with the following command:
+
+```sh
+pip install periflow-client[mllib]
+```
+
+If you don't need to convert the checkpoint, you can install the package without the `mllib` dependencies:
+
+```sh
+pip install periflow-client
+```
+
+Note that the PeriFlow Client package requires Python version >= 3.8.
+:::
+
+## Step 1. Signing in with PeriFlow
+
+To use the PeriFlow CLI and upload your AI model checkpoint, you need to sign in with your PeriFlow account. Follow these steps:
+
+1. Open a terminal shell.
+2. Enter the following command:
+
+```sh
+pf login
+```
+
+3. When prompted, provide your PeriFlow account credentials, including your ID and password. After successful login, you will see the following output:
+
+```txt
+Enter Username: <YOUR_ID>
+Enter Password:
+
+
+Login success!
+Welcome back to...
+ _____          _  _____ _
+|  __ \___ _ __(_)|  ___| | _____      __
+|  ___/ _ \ '__| || |__ | |/ _ \ \ /\ / /
+| |  |  __/ |  | ||  __|| | (_) | V  V /
+|_|   \___|_|  |_||_|   |_|\___/ \_/\_/
+
+```
+
+## Step 2. Switching to Your Project
+
+Now that you are logged in, you can switch to the PeriFlow project where you want to upload the checkpoint. Follow these steps:
+
+1. Open a terminal shell (if not already open).
+2. Use the following command to switch to your desired project, replacing `$YOUR_PROJECT_NAME` with the actual project name:
+
+```sh
+pf project switch $YOUR_PROJECT_NAME
+```
+
+:::info
+If you are not sure about the available projects, you can list them using the following command:
+
+```sh
+pf project list
+```
+:::
+
+## Step 3. Converting Hugging Face Checkpoint (Optional)
+
+If you want to serve your AI model with PeriFlow, you'll need to convert it to the PeriFlow-compatible format.
+Here's how you can easily convert a Hugging Face model checkpoint to the PeriFlow format:
+
+1. Open a terminal shell (if not already open).
+2. Use the following command to convert the Hugging Face model checkpoint to the PeriFlow-compatible format. Replace `$MODEL_NAME_OR_PATH`, `$OUTPUT_DIR`, and `$DATA_TYPE` with the actual values you want.
+
+```sh
+pf checkpoint convert \
+    --model-name-or-path $MODEL_NAME_OR_PATH \
+    --output-dir $OUTPUT_DIR \
+    --data-type $DATA_TYPE
+```
+
+:::info
+You have two options for `$MODEL_NAME_OR_PATH`:
+
+- Enter the local checkpoint path.
+- Use the Hugging Face model name (e.g., `gpt2`, `EleutherAI/gpt-j-6b`).
+
+For the `$DATA_TYPE` parameter, you can choose from three options:
+
+- `fp16`: 16-bit floating-point format.
+- `fp32`: 32-bit floating-point format.
+- `bf16`: bfloat16 format.
+:::
+
+3. After executing the above command, the following files will be created at $OUTPUT_DIR:
+    - `model.h5`: The converted model checkpoint file with HDF5 format.
+    - `tokenizer.json`: Tokenizer file used for tokenizing/detokenizing the model inputs/ouputs.
+    - `attr.yaml`: The checkpoint attributes file containing the model configurations and generation configurations.
+
+:::caution
+PeriFlow CLI will try to configure `tokenizer.json` and `attr.yaml` automatically, but the automatic configuration might fail in the following cases:
+
+- If the model does support the [**Fast tokenizer**](https://huggingface.co/learn/nlp-course/chapter6/3), which is compatible with PeriFlow, `tokenizer.json` will not be created.
+- If the model config or generation config published to Hugging Face does not contain the required information, the value will be left blank (marked with "FILL ME") in `attr.yaml`.
+- If there is a conflict between the model config and the generation config published to the Hugging Face, the value will be left blank (marked with "FILL ME") in `attr.yaml`.
+:::
+
+After completing this step, your Hugging Face model checkpoint will be converted to the PeriFlow-compatible format, and you'll be able to use it with PeriFlow seamlessly.
+
+## Step 4. Uploading Checkpoint
+
+Once your checkpoint is prepared, you can upload it to PeriFlow using the following steps:
+
+1. Open a terminal shell (if not already open).
+2. Use the following command to upload the model checkpoint file (`model.h5`) and the tokenizer file (`tokenizer.json`).
+
+```sh
+pf checkpoint upload \
+    --name $CHECKPOINT_NAME \
+    --source-path $SOURCE_PATH \
+    --attr-file $ATTR_FILE_PATH
+```
+
+:::info
+Replace the placeholders with the appropriate values:
+
+- `$CHECKPOINT_NAME`: Enter the desired name for your checkpoint.
+- `$SOURCE_PATH`: Provide the path to the directory containing `model.h5` and `tokenizer.json`.
+- `$ATTR_FILE_PATH`: Specify the path to `attr.yaml`, which contains the model configurations and generation configurations.
+:::
+
+After following these steps, your model checkpoint will be successfully uploaded to your PeriFlow project, ready to be used for AI model serving.

docs/docusaurus.config.js

@@ -41,6 +41,9 @@ const SECTIONS = [
   defineSection('sdk', {
     label: latestVersions['sdk'],
   }),
+  defineSection('tutorials', {
+    label: latestVersions['tutorials'],
+  }),
 ];
 
 /** @type {import('@docusaurus/types').Config} */
@@ -116,11 +119,21 @@ const config = {
             to: 'sdk/intro',
             position: 'left',
           },
+          {
+            label: 'Tutorials',
+            to: 'tutorials/intro',
+            position: 'left',
+          },
           {
             href: 'https://github.com/friendliai/periflow-client/',
             label: 'GitHub',
             position: 'right',
           },
+          {
+            label: 'Quick Start',
+            href: 'https://periflow.ai/quick-start/',
+            position: 'left',
+          },
         ],
       },
       footer: {
@@ -137,6 +150,14 @@ const config = {
                 label: 'Python SDK',
                 to: 'sdk/intro',
               },
+              {
+                label: 'Tutorials',
+                to: 'tutorials/intro',
+              },
+              {
+                label: 'Quick Start',
+                href: 'https://periflow.ai/quick-start/'
+              }
             ],
           },
           {

periflow/cli/main.py

@@ -27,7 +27,7 @@
 from periflow.utils.version import get_installed_version
 
 app = typer.Typer(
-    help="Welcome to PeriFlow 🤗",
+    help="Supercharge Generative AI Serving 🚀",
     no_args_is_help=True,
     context_settings={"help_option_names": ["-h", "--help"]},
     add_completion=False,

pyproject.toml

@@ -1,9 +1,17 @@
 [tool.poetry]
 name = "periflow-client"
-version = "0.1.0"
-description = "PeriFlow is a reliable, speedy, and efficient service for training and serving your own Generative AI models on any data of your choice."
+version = "0.1.1"
+description = "Client of PeriFlow, the fastest generative AI serving available."
+license = "Apache-2.0"
 authors = ["PeriFlow teams <eng@friendli.ai>"]
-packages = [{include = "periflow"}]
+packages = [
+    { include = "periflow" },
+]
+readme = "README.md"
+homepage = "https://docs.periflow.ai/"
+repository = "https://github.com/friendliai/periflow-client"
+documentation = "https://docs.periflow.ai/"
+keywords = ["generative-ai", "serving", "llm"]
 
 [tool.poetry.scripts]
 pf = "periflow.cli:app"