Merge pull request #59 from casperdcl/ci-fun

add CI build matrix, refactoring, linting

.github/workflows/ci.yml

@@ -1,28 +1,56 @@
 name: CI
-
-env:
-  DOCKER_BUILDKIT: 1
-  DOCKER_CLI_EXPERIMENTAL: "enabled"
-  COMPOSE_DOCKER_CLI_BUILD: 1
-
 on:
-  pull_request:
-    branches: ["main"]
-
   push:
-    branches: ["main"]
-
+  pull_request:
 jobs:
-  linter:
+  setup:
+    runs-on: ubuntu-latest
+    steps:
+    - uses: actions/checkout@v3
+      with:
+        fetch-depth: 0
+    - id: list
+      name: List modified models
+      run: |
+        git diff --stat HEAD~1
+        echo "modified=$(
+          for model in $(ls -d *-*); do
+            git diff --quiet HEAD~1 -- $model || echo $model
+          done | sort | jq -Rsc 'split("\n")[:-1]'
+        )" >> $GITHUB_OUTPUT
+    outputs:
+      models: ${{ steps.list.outputs.modified }}
+  build-test-push:
+    needs: setup
+    runs-on: [self-hosted, gpu]
+    strategy:
+      fail-fast: false  # TODO: remove
+      matrix:
+        model: ${{ fromJson(needs.setup.outputs.models) }}
+    env:
+      VERSION: '1.0.0'
+      DOCKER_BUILDKIT: 1
+      DOCKER_CLI_EXPERIMENTAL: enabled
+      COMPOSE_DOCKER_CLI_BUILD: 1
+    steps:
+    - uses: actions/checkout@v3
+    - name: Login to GitHub Container Registry
+      uses: docker/login-action@v2
+      with:
+        registry: ghcr.io
+        username: ${{ github.actor }}
+        password: ${{ secrets.GITHUB_TOKEN }}
+    - run: ./build.sh ${{ github.event_name == 'push' && startsWith(github.ref, 'refs/tags') && '--push' || '--load' }}
+      working-directory: ${{ matrix.model }}
+      env:
+        # can't load cross-CPU-platform
+        BUILDX_PLATFORM: ${{ github.event_name == 'push' && startsWith(github.ref, 'refs/tags') && 'linux/arm64,linux/amd64' || 'linux/amd64' }}
+        # can't test cross-CPU-platform
+        TESTS_SKIP_CPU: ${{ github.event_name == 'push' && startsWith(github.ref, 'refs/tags') && '1' || '' }}
+        # avoid OoM errors (TODO: remove when using larger GPU)
+        TESTS_SKIP_GPU: ${{ contains(fromJSON('["a2t-whisper","cht-dolly-v2","cht-gorilla","cht-llama-v2","cht-mpt","cht-xgen","dfs-dalle","dfs-diffusers"]'), matrix.model) && '1' || '' }}
+  pass:
+    needs: build-test-push
     runs-on: ubuntu-latest
     steps:
-      - name: Checkout Code Repository
-        uses: actions/checkout@v3
-
-      - name: Set up Python
-        uses: actions/setup-python@v4
-        with:
-          python-version: "3.10"
-
-      - name: Run pre-commit
-        uses: pre-commit/action@v3.0.0
+    - run: echo success

.pre-commit-config.yaml

@@ -1,37 +1,42 @@
-exclude: "^docs/|/migrations/"
 default_stages: [commit]
-
 repos:
-  - repo: https://github.com/pre-commit/pre-commit-hooks
-    rev: v4.4.0
-    hooks:
-      - id: trailing-whitespace
-      - id: end-of-file-fixer
-      - id: check-yaml
-      - id: end-of-file-fixer
-      - id: check-toml
-      - id: check-xml
-      - id: debug-statements
-      - id: check-builtin-literals
-      - id: check-case-conflict
-
-  - repo: https://github.com/psf/black
-    rev: 23.7.0
-    hooks:
-      - id: black
-
-  - repo: https://github.com/PyCQA/isort
-    rev: 5.12.0
-    hooks:
-      - id: isort
-
-  - repo: https://github.com/PyCQA/flake8
-    rev: 6.0.0
-    hooks:
-      - id: flake8
-        args: ["--config=setup.cfg"]
-        additional_dependencies: [flake8-isort]
-
+- repo: https://github.com/pre-commit/pre-commit-hooks
+  rev: v4.4.0
+  hooks:
+  - id: check-added-large-files
+  - id: check-builtin-literals
+  - id: check-case-conflict
+  - id: check-docstring-first
+  - id: check-executables-have-shebangs
+  - id: check-merge-conflict
+  - id: check-toml
+  - id: check-xml
+  - id: check-yaml
+  - id: debug-statements
+  - id: end-of-file-fixer
+  - id: mixed-line-ending
+  - id: sort-simple-yaml
+  - id: trailing-whitespace
+- repo: https://github.com/psf/black
+  rev: 23.7.0
+  hooks:
+  - id: black
+- repo: https://github.com/PyCQA/isort
+  rev: 5.12.0
+  hooks:
+  - id: isort
+- repo: https://github.com/PyCQA/flake8
+  rev: 6.0.0
+  hooks:
+  - id: flake8
+    additional_dependencies:
+    - flake8-broken-line
+    - flake8-bugbear
+    - flake8-comprehensions
+    - flake8-debugger
+    - flake8-isort
+    - flake8-pyproject
+    - flake8-string-format
 ci:
   autoupdate_schedule: weekly
   skip: []

a2t-whisper/build.sh

@@ -0,0 +1,26 @@
+#!/bin/bash
+set -e
+export VERSION=1.0.1
+
+IMAGE=ghcr.io/premai-io/audio-to-text-whisper-tiny-cpu
+docker buildx build ${@:1} \
+    --file ./docker/cpu/Dockerfile \
+    --build-arg="MODEL_ID=tiny" \
+    --tag $IMAGE:latest \
+    --tag $IMAGE:$VERSION \
+    --platform ${BUILDX_PLATFORM:-linux/arm64,linux/amd64} \
+    .
+if test -z $TESTS_SKIP_CPU; then
+  docker run --rm $IMAGE:$VERSION pytest
+fi
+
+IMAGE=ghcr.io/premai-io/audio-to-text-whisper-large-v2-gpu
+docker buildx build ${@:1} \
+    --file ./docker/gpu/Dockerfile \
+    --build-arg="MODEL_ID=large-v2" \
+    --tag $IMAGE:latest \
+    --tag $IMAGE:$VERSION \
+    .
+if test -z $TESTS_SKIP_GPU; then
+  docker run --rm --gpus all $IMAGE:$VERSION pytest
+fi

cdr-replit/build.sh

@@ -0,0 +1,14 @@
+#!/bin/bash
+set -e
+export VERSION=1.0.0
+
+IMAGE=ghcr.io/premai-io/coder-replit-code-v1-3b-gpu
+docker buildx build ${@:1} \
+    --file ./docker/gpu/Dockerfile \
+    --build-arg="MODEL_ID=replit/replit-code-v1-3b" \
+    --tag $IMAGE:latest \
+    --tag $IMAGE:$VERSION \
+    .
+if test -z $TESTS_SKIP_GPU; then
+  docker run --rm --gpus all $IMAGE:$VERSION pytest
+fi

cdr-t5/build.sh

@@ -0,0 +1,15 @@
+#!/bin/bash
+set -e
+export VERSION=1.0.0
+
+IMAGE=ghcr.io/premai-io/coder-codet5p-220m-py-cpu
+docker buildx build ${@:1} \
+    --file ./docker/cpu/Dockerfile \
+    --build-arg="MODEL_ID=Salesforce/codet5p-220m-py" \
+    --tag $IMAGE:latest \
+    --tag $IMAGE:$VERSION \
+    --platform ${BUILDX_PLATFORM:-linux/arm64,linux/amd64} \
+    .
+if test -z $TESTS_SKIP_CPU; then
+  docker run --rm $IMAGE:$VERSION pytest
+fi

cht-dolly-v2/build.sh

@@ -0,0 +1,14 @@
+#!/bin/bash
+set -e
+export VERSION=1.0.3
+
+IMAGE=ghcr.io/premai-io/chat-dolly-v2-12b-gpu
+docker buildx build ${@:1} \
+    --file ./docker/gpu/Dockerfile \
+    --build-arg="MODEL_ID=databricks/dolly-v2-12b" \
+    --tag $IMAGE:latest \
+    --tag $IMAGE:$VERSION \
+    .
+if test -z $TESTS_SKIP_GPU; then
+  docker run --rm --gpus all $IMAGE:$VERSION pytest
+fi

cht-falcon/build.sh

@@ -0,0 +1,14 @@
+#!/bin/bash
+set -e
+export VERSION=1.0.0
+
+IMAGE=ghcr.io/premai-io/chat-falcon-7b-instruct-gpu
+docker buildx build ${@:1} \
+    --file ./docker/gpu/Dockerfile \
+    --build-arg="MODEL_ID=tiiuae/falcon-7b-instruct" \
+    --tag $IMAGE:latest \
+    --tag $IMAGE:$VERSION \
+    .
+if test -z $TESTS_SKIP_GPU; then
+  docker run --rm --gpus all $IMAGE:$VERSION pytest
+fi

cht-gorilla/build.sh

@@ -0,0 +1,25 @@
+#!/bin/bash
+set -e
+export VERSION=1.0.0
+
+IMAGE=ghcr.io/premai-io/chat-gorilla-falcon-7b-gpu
+docker buildx build ${@:1} \
+    --file ./docker/gpu/Dockerfile \
+    --build-arg="MODEL_ID=gorilla-llm/gorilla-falcon-7b-hf-v0" \
+    --tag $IMAGE:latest \
+    --tag $IMAGE:$VERSION \
+    .
+if test -z $TESTS_SKIP_GPU; then
+  docker run --rm --gpus all $IMAGE:$VERSION pytest
+fi
+
+IMAGE=ghcr.io/premai-io/chat-gorilla-mpt-7b-gpu
+docker buildx build ${@:1} \
+    --file ./docker/gpu/Dockerfile \
+    --build-arg="MODEL_ID=gorilla-llm/gorilla-mpt-7b-hf-v0" \
+    --tag $IMAGE:latest \
+    --tag $IMAGE:$VERSION \
+    .
+if test -z $TESTS_SKIP_GPU; then
+  docker run --rm --gpus all $IMAGE:$VERSION pytest
+fi

cht-llama-cpp/build.sh

@@ -0,0 +1,27 @@
+#!/bin/bash
+set -e
+export VERSION=1.0.4
+
+IMAGE=ghcr.io/premai-io/chat-gpt4all-lora-q4-cpu
+docker buildx build ${@:1} \
+    --file ./docker/cpu/Dockerfile \
+    --build-arg="MODEL_ID=gpt4all-lora-q4" \
+    --tag $IMAGE:latest \
+    --tag $IMAGE:$VERSION \
+    --platform ${BUILDX_PLATFORM:-linux/arm64,linux/amd64} \
+    .
+if test -z $TESTS_SKIP_CPU; then
+  docker run --rm $IMAGE:$VERSION pytest
+fi
+
+IMAGE=ghcr.io/premai-io/chat-vicuna-7b-q4-cpu
+docker buildx build ${@:1} \
+    --file ./docker/cpu/Dockerfile \
+    --build-arg="MODEL_ID=vicuna-7b-q4" \
+    --tag $IMAGE:latest \
+    --tag $IMAGE:$VERSION \
+    --platform ${BUILDX_PLATFORM:-linux/arm64,linux/amd64} \
+    .
+if test -z $TESTS_SKIP_CPU; then
+  docker run --rm $IMAGE:$VERSION pytest
+fi

cht-llama-cpp/models.py

@@ -7,6 +7,7 @@
     "gpt4all-lora-q4": {"modelWeightsName": "gpt4all-lora-q4.bin", "ctxMaxTokens": 512},
     "vicuna-7b-q4": {"modelWeightsName": "vicuna-7b-q4.bin", "ctxMaxTokens": 512},
 }
+DEFAULT_N_THREADS = max(multiprocessing.cpu_count() // 2, 1)
 
 
 def get_model_info() -> dict:
@@ -43,10 +44,12 @@ def generate(
         n: int = 1,
         stream: bool = False,
         max_tokens: int = 256,
-        stop: list = [],
-        n_threads: int = max(multiprocessing.cpu_count() // 2, 1),
+        stop: list = None,
+        n_threads: int = DEFAULT_N_THREADS,
         **kwargs,
     ):
+        if stop is None:
+            stop = []
         messages = cls.reduce_number_of_messages(messages[::-1], max_tokens)[::-1]
         cls.model.n_threads = n_threads
         return cls.model.create_chat_completion(

cht-llama-v2/build.sh

@@ -0,0 +1,47 @@
+#!/bin/bash
+set -e
+export VERSION=1.0.0
+
+IMAGE=ghcr.io/premai-io/chat-llama-2-7b-gpu
+docker buildx build ${@:1} \
+    --file ./docker/gpu/Dockerfile \
+    --build-arg="MODEL_ID=llama-2-7b-hf" \
+    --tag $IMAGE:latest \
+    --tag $IMAGE:$VERSION \
+    .
+if test -z $TESTS_SKIP_GPU; then
+  docker run --rm --gpus all $IMAGE:$VERSION pytest
+fi
+
+IMAGE=ghcr.io/premai-io/chat-llama-2-7b-chat-gpu
+docker buildx build ${@:1} \
+    --file ./docker/gpu/Dockerfile \
+    --build-arg="MODEL_ID=llama-2-7b-chat-hf" \
+    --tag $IMAGE:latest \
+    --tag $IMAGE:$VERSION \
+    .
+if test -z $TESTS_SKIP_GPU; then
+  docker run --rm --gpus all $IMAGE:$VERSION pytest
+fi
+
+IMAGE=ghcr.io/premai-io/chat-llama-2-13b-gpu
+docker buildx build ${@:1} \
+    --file ./docker/gpu/Dockerfile \
+    --build-arg="MODEL_ID=llama-2-13b-hf" \
+    --tag $IMAGE:latest \
+    --tag $IMAGE:$VERSION \
+    .
+if test -z $TESTS_SKIP_GPU; then
+  docker run --rm --gpus all $IMAGE:$VERSION pytest
+fi
+
+IMAGE=ghcr.io/premai-io/chat-llama-2-13b-chat-gpu
+docker buildx build ${@:1} \
+    --file ./docker/gpu/Dockerfile \
+    --build-arg="MODEL_ID=llama-2-13b-chat-hf" \
+    --tag $IMAGE:latest \
+    --tag $IMAGE:$VERSION \
+    .
+if test -z $TESTS_SKIP_GPU; then
+  docker run --rm --gpus all $IMAGE:$VERSION pytest
+fi