Add pypi publishing workflow

.github/workflows/publish_to_pypi.yml

@@ -0,0 +1,109 @@
+name: Build and publish to PyPI
+
+on:
+  release:
+    types: [published]
+  workflow_dispatch:  # Allows you to run this workflow manually from the Actions tab
+
+jobs:
+  build_wheels:
+    name: Build wheels on ${{ matrix.os }}
+    runs-on: ${{ matrix.os }}
+    strategy:
+      fail-fast: false
+      matrix:
+        os: [ubuntu-latest, windows-latest, macos-latest]
+        python-version: ['3.8', '3.9', '3.10', '3.11', '3.12']
+        exclude:
+          # Add any exclusions if certain OS/Python combinations are problematic
+          # - os: macos-latest
+          #   python-version: '3.12'
+
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          fetch-depth: 0  # Gets all history for proper versioning
+
+      - name: Set up Python ${{ matrix.python-version }}
+        uses: actions/setup-python@v4
+        with:
+          python-version: ${{ matrix.python-version }}
+
+      - name: Install build dependencies
+        run: |
+          python -m pip install --upgrade pip
+          pip install build wheel setuptools
+
+      - name: Build wheels
+        run: |
+          python -m build --wheel --outdir dist/
+
+      - name: Upload wheels
+        uses: actions/upload-artifact@v3
+        with:
+          name: wheels-${{ matrix.os }}-${{ matrix.python-version }}
+          path: dist/*.whl
+
+  build_sdist:
+    name: Build source distribution
+    runs-on: ubuntu-latest
+
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          fetch-depth: 0  # Gets all history for proper versioning
+
+      - name: Set up Python
+        uses: actions/setup-python@v4
+        with:
+          python-version: '3.12'
+
+      - name: Install build dependencies
+        run: |
+          python -m pip install --upgrade pip
+          pip install build twine
+
+      - name: Build sdist
+        run: |
+          python -m build --sdist --outdir dist/
+
+      - name: Check metadata
+        run: |
+          twine check dist/*.tar.gz
+
+      - name: Upload sdist
+        uses: actions/upload-artifact@v3
+        with:
+          name: sdist
+          path: dist/*.tar.gz
+
+  publish:
+    name: Publish to PyPI
+    needs: [build_wheels, build_sdist]
+    runs-on: ubuntu-latest
+    # Only publish on release
+    if: github.event_name == 'release' && github.event.action == 'published'
+    environment:
+      name: pypi
+      url: https://pypi.org/project/syncode/
+    permissions:
+      id-token: write  # For PyPI trusted publishing
+
+    steps:
+      - name: Download all artifacts
+        uses: actions/download-artifact@v3
+        with:
+          path: dist
+
+      - name: Flatten dist directory
+        run: |
+          mkdir -p flat_dist
+          find dist -type f -name "*.whl" -o -name "*.tar.gz" -exec cp {} flat_dist \;
+          ls -la flat_dist
+
+      - name: Publish to PyPI
+        uses: pypa/gh-action-pypi-publish@release/v1
+        with:
+          packages-dir: flat_dist
+          verbose: true
+          # skip-existing: true  # Uncomment if you want to skip existing versions

.gitignore

@@ -8,3 +8,5 @@ tmp*
 cache/
 .ipynb_checkpoints/
 *.prof
+dist/
+syncode.egg-info/

README.md

@@ -48,27 +48,30 @@ Define your own grammar using simple EBNF syntax. Check out our [notebooks direc
 | 🎲 Sample with any existing decoding strategy (eg. greedy, beam search, nucleus sampling)                           |
 
 
-## 📖 More About **SynCode**
+## 🚀 Quick Start
+### Python Installation and Usage Instructions
 
-### How **SynCode** works?
+You can install SynCode via PyPI:
 
-<img width="750" alt="Screenshot 2024-03-21 at 2 22 15 AM" src="https://github.com/uiuc-focal-lab/syncode/assets/14147610/d9d73072-3c9b-47d4-a941-69d5cf8fb1bf">
+```bash
+pip install syncode
+```
 
-In the SynCode workflow, the LLM takes partial code _C<sub>k</sub>_ and generates a distribution for the next token _t<sub>k+1</sub>_. The incremental parser processes _C<sub>k</sub>_ to generate accept sequences _A_, the sequences of terminals that can follow partial code called accept sequences. Simultaneously, the incremental parser computes a remainder _r_ from the partial code, representing the suffix that may change its terminal type in subsequent generations. The backbone of SynCode is the offline construction of a DFA mask store, a lookup table derived from regular expressions representing the terminals of the language grammar. The DFA mask store facilitates efficient traversal of DFA states, enabling the retrieval of masks mapped to each state and accept sequence. SynCode walks over the DFA using the remainder and uses the mask store to compute the mask specific to each accept sequence. By unifying masks for each accept sequence SynCode gets the set of syntactically valid tokens. The LLM iteratively generates a token _t<sub>k+1</sub>_ using the distribution and the mask, appending it to _C<sub>k</sub>_ to create the updated code _C<sub>k+1</sub>_. The process continues until the LLM returns the final code _C<sub>n</sub>_ based on the defined stop condition.
+Alternatively, you can install the latest development version directly from GitHub:
 
-## 🚀 Quick Start
-### Python Installation and Usage Instructions
-Simply install SynCode via PyPi using the following command:
-``` bash
+```bash
 pip install git+https://github.com/uiuc-focal-lab/syncode.git
 ```
 
-Note: SynCode depends on HuggingFace [transformers](https://github.com/huggingface/transformers):
-| SynCode version | Recommended transformers version |
-| -------------- | -------------------------------- |
-| `v0.1.4` (latest) | `v4.44.0`                         |
-| `v0.1.2`          | `v4.42.0`                         |
+#### Version Compatibility
+
+SynCode depends on HuggingFace [transformers](https://github.com/huggingface/transformers):
 
+| SynCode version | Required transformers version | Python version |
+| -------------- | ----------------------------- | -------------- |
+| `v0.4.1` (latest) | `v4.44.0`                    | 3.6 - 3.12     |
+
+**Note:** Python 3.13 is not currently supported due to dependency constraints.
 
 ### Usage option 1:
 SynCode can be used as a simple logit processor with HuggingFace [transformers](https://github.com/huggingface/transformers) library interface. Check this [notebook](./notebooks/example_logits_processor.ipynb) for example.
@@ -426,6 +429,14 @@ print(f"Syncode augmented LLM output:\n{output}")
 }
 ```
 
+## 📖 More About **SynCode**
+
+### How **SynCode** works?
+
+<img width="750" alt="Screenshot 2024-03-21 at 2 22 15 AM" src="https://github.com/uiuc-focal-lab/syncode/assets/14147610/d9d73072-3c9b-47d4-a941-69d5cf8fb1bf">
+
+In the SynCode workflow, the LLM takes partial code _C<sub>k</sub>_ and generates a distribution for the next token _t<sub>k+1</sub>_. The incremental parser processes _C<sub>k</sub>_ to generate accept sequences _A_, the sequences of terminals that can follow partial code called accept sequences. Simultaneously, the incremental parser computes a remainder _r_ from the partial code, representing the suffix that may change its terminal type in subsequent generations. The backbone of SynCode is the offline construction of a DFA mask store, a lookup table derived from regular expressions representing the terminals of the language grammar. The DFA mask store facilitates efficient traversal of DFA states, enabling the retrieval of masks mapped to each state and accept sequence. SynCode walks over the DFA using the remainder and uses the mask store to compute the mask specific to each accept sequence. By unifying masks for each accept sequence SynCode gets the set of syntactically valid tokens. The LLM iteratively generates a token _t<sub>k+1</sub>_ using the distribution and the mask, appending it to _C<sub>k</sub>_ to create the updated code _C<sub>k+1</sub>_. The process continues until the LLM returns the final code _C<sub>n</sub>_ based on the defined stop condition.
+
 ## Contact
 For questions, please contact [Shubham Ugare](mailto:shubhamdugare@gmail.com).
 

pyproject.toml

@@ -4,7 +4,8 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "syncode"
-version = "0.4.0"
+version="0.4.1"
+requires-python = ">=3.6,<3.13"
 description = "Grammar-guided code generation tool"
 readme = "README.md"
 authors = [

requirements.txt

@@ -3,6 +3,6 @@ interegular
 regex==2023.8.8
 torch
 tqdm
-transformers==4.44.0
+transformers==4.44.0; python_version < "3.13"
 datasets
 jsonschema

setup.py

@@ -1,4 +1,5 @@
 import setuptools
+python_requires=">=3.6,<3.13",
 
 with open("README.md", "r", encoding="utf-8") as fh:
     long_description = fh.read()
@@ -17,7 +18,7 @@
 
 setuptools.setup(
     name="syncode",
-    version="0.4.0",
+    version="0.4.1",
     author="Shubham Ugare",
     author_email="shubhamugare@gmail.com",
     description="This package provides the tool for grammar augmented LLM generation.",