ggml-python 0.0.20__tar.gz → 0.0.22__tar.gz

{ggml_python-0.0.20 → ggml_python-0.0.22}/.github/workflows/test.yaml

@@ -14,7 +14,7 @@ jobs:
     runs-on: ubuntu-latest
     strategy:
       matrix:
-        python-version: ["3.8", "3.9", "3.10", "3.11"]
+        python-version: ["3.8", "3.9", "3.10", "3.11", "3.12"]
 
     steps:
       - uses: actions/checkout@v3
@@ -37,7 +37,7 @@ jobs:
     runs-on: windows-latest
     strategy:
       matrix:
-        python-version: ["3.8", "3.9", "3.10", "3.11"]
+        python-version: ["3.8", "3.9", "3.10", "3.11", "3.12"]
 
     steps:
       - uses: actions/checkout@v3
@@ -61,7 +61,7 @@ jobs:
     runs-on: macos-latest
     strategy:
       matrix:
-        python-version: ["3.8", "3.9", "3.10", "3.11"]
+        python-version: ["3.8", "3.9", "3.10", "3.11", "3.12"]
 
     steps:
       - uses: actions/checkout@v3
@@ -78,3 +78,48 @@ jobs:
       - name: Test with pytest
         run: |
           python3 -m pytest
+
+  build-linux-opencl:
+    runs-on: ubuntu-latest
+
+    steps:
+      - uses: actions/checkout@v3
+        with:
+          submodules: "true"
+
+      - name: Dependencies
+        run: |
+          wget -O- https://apt.repos.intel.com/intel-gpg-keys/GPG-PUB-KEY-INTEL-SW-PRODUCTS.PUB | gpg --dearmor | sudo tee /usr/share/keyrings/oneapi-archive-keyring.gpg > /dev/null
+          echo "deb [signed-by=/usr/share/keyrings/oneapi-archive-keyring.gpg] https://apt.repos.intel.com/oneapi all main" | sudo tee /etc/apt/sources.list.d/oneAPI.list  
+          sudo apt-get update
+          sudo apt-get install -y --no-install-recommends llvm intel-oneapi-runtime-opencl intel-oneapi-runtime-compilers libclblast-dev
+          sudo apt-get install -y python3-dev python3-pip make
+      
+      - name: Setup
+        run: |
+          python3 -m pip install pytest
+          make build.clblast
+
+      - name: Test with pytest
+        run: |
+          make test
+
+  build-macos-metal:
+
+    runs-on: macos-latest
+
+    steps:
+      - uses: actions/checkout@v3
+        with:
+          submodules: "true"
+      - name: Set up Python
+        uses: actions/setup-python@v4
+        with:
+          python-version: "3.8"
+      - name: Install dependencies
+        run: |
+          python3 -m pip install --upgrade pip pytest cmake scikit-build setuptools
+          CMAKE_ARGS="-DGGML_METAL=On" python3 -m pip install --verbose --editable .
+      - name: Test with pytest
+        run: |
+          python3 -m pytest

{ggml_python-0.0.20 → ggml_python-0.0.22}/CMakeLists.txt

@@ -1,4 +1,4 @@
-cmake_minimum_required(VERSION 3.4...3.22)
+cmake_minimum_required(VERSION 3.21)
 
 project(
     ${SKBUILD_PROJECT_NAME}

{ggml_python-0.0.20 → ggml_python-0.0.22}/Makefile

@@ -8,37 +8,37 @@ ${submodules}:
 update-pip:
 	python3 -m pip install --upgrade pip
 
-build: ${submodules} update-pip
+build: ${submodules} update-pip ## Build ggml-python with cpu support
 	python3 -m pip install --verbose --editable .
 
-build.debug: ${submodules} update-pip
+build.debug: ${submodules} update-pip ## Build ggml-python with cpu support and debug symbols
 	CMAKE_ARGS="-DCMAKE_BUILD_TYPE=Debug" python3 -m pip install --verbose --config-settings=cmake.verbose=true --config-settings=logging.level=INFO --config-settings=install.strip=false  --editable .
 
-build.openblas: ${submodules} update-pip
+build.openblas: ${submodules} update-pip ## Build ggml-python with openblas support
 	CMAKE_ARGS="-DGGML_OPENBLAS=On" python3 -m pip install --verbose --editable .
 
-build.cublas: ${submodules} update-pip
+build.cublas: ${submodules} update-pip ## Build ggml-python with cublas / cuda support
 	CMAKE_ARGS="-DGGML_CUBLAS=On" python3 -m pip install --verbose --editable .
 
-build.clblast: ${submodules} update-pip
+build.clblast: ${submodules} update-pip ## Build ggml-python with clblast / opencl support
 	CMAKE_ARGS="-DGGML_CLBLAST=On" python3 -m pip install --verbose --editable .
 
-sdist:
+sdist: ## Build source distribution
 	python3 -m build --sdist
 
-deploy:
+deploy: ## Deploy to pypi
 	twine upload dist/*
 
-test:
+test: ## Run tests
 	python3 -m pytest
 
-test.gdb:
+test.gdb: ## Run tests with gdb
 	gdb -ex r -ex "thread apply all bt" --args python -m pytest -s -vvvv
 
-docs:
+docs: ## Build documentation using mkdocs and serve it
 	mkdocs serve
 
-clean:
+clean: ## Clean build artifacts
 	- rm -rf build
 	- rm -rf dist
 	- rm ggml/*.so
@@ -49,4 +49,20 @@ clean:
 	- rm ${submodules}/*.dylib
 	- cd ${submodules} && make clean
 
-.PHONY: all update-pip build build.openblas build.cuda sdist deploy test test.gdb docs clean
+help: ## Prints help menu
+	@grep -E '^[\.a-zA-Z_-]+:.*?## .*$$' Makefile | sort | awk 'BEGIN {FS = ":.*?## "}; {printf "\033[36m%-30s\033[0m %s\n", $$1, $$2}'
+
+.PHONY: \
+	all \
+	build \
+	build.debug \
+	build.openblas \
+	build.cublas \
+	build.clblast \
+	sdist \
+	deploy \
+	test \
+	test.gdb \
+	docs \
+	clean \
+	help

{ggml_python-0.0.20 → ggml_python-0.0.22}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ggml_python
-Version: 0.0.20
+Version: 0.0.22
 Summary: Python bindings for ggml
 Author-Email: Andrei Betlen <abetlen@gmail.com>
 License: MIT
@@ -9,6 +9,7 @@ Classifier: Programming Language :: Python :: 3.8
 Classifier: Programming Language :: Python :: 3.9
 Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
 Project-URL: Homepage, https://github.com/abetlen/ggml-python
 Project-URL: Documentation, https://ggml-python.readthedocs.io/en/latest/
 Project-URL: Issues, https://github.com/abetlen/ggml-python/issues

ggml_python-0.0.22/ggml/__init__.py

@@ -0,0 +1,3 @@
+from .ggml import *
+
+__version__ = "0.0.22"

{ggml_python-0.0.20 → ggml_python-0.0.22}/ggml/experimental.py

@@ -531,21 +531,21 @@ class Tensor:
         return Tensor.with_buffer(tensor=op, ctx=ctx, src=[a, b])
 
     @staticmethod
-    def norm(a: Tensor, ctx: Optional[Context] = None):
+    def norm(a: Tensor, eps: float, ctx: Optional[Context] = None):
         ctx = ctx or Context.with_tensor_overhead()
-        op = ggml.ggml_norm(ctx.context, a.tensor)
+        op = ggml.ggml_norm(ctx.context, a.tensor, eps)
         return Tensor.with_buffer(tensor=op, ctx=ctx, src=[a])
 
     @staticmethod
-    def rms_norm(a: Tensor, ctx: Optional[Context] = None):
+    def rms_norm(a: Tensor,  eps: float, ctx: Optional[Context] = None):
         ctx = ctx or Context.with_tensor_overhead()
-        op = ggml.ggml_rms_norm(ctx.context, a.tensor)
+        op = ggml.ggml_rms_norm(ctx.context, a.tensor, eps)
         return Tensor.with_buffer(tensor=op, ctx=ctx, src=[a])
 
     @staticmethod
-    def rms_norm_back(a: Tensor, b: Tensor, ctx: Optional[Context] = None):
+    def rms_norm_back(a: Tensor, b: Tensor, eps: float, ctx: Optional[Context] = None):
         ctx = ctx or Context.with_tensor_overhead()
-        op = ggml.ggml_rms_norm_back(ctx.context, a.tensor, b.tensor)
+        op = ggml.ggml_rms_norm_back(ctx.context, a.tensor, b.tensor, eps)
         return Tensor.with_buffer(tensor=op, ctx=ctx, src=[a, b])
 
     @staticmethod
@@ -890,7 +890,7 @@ class Tensor:
     @staticmethod
     def rope(
         a: Tensor,
-        n_past: int,
+        b: Tensor,
         n_dims: int,
         mode: int,
         n_ctx: int,
@@ -900,7 +900,7 @@ class Tensor:
         op = ggml.ggml_rope(
             ctx.context,
             a.tensor,
-            n_past,
+            b.tensor,
             n_dims,
             mode,
             n_ctx,
@@ -910,31 +910,41 @@ class Tensor:
     @staticmethod
     def rope_inplace(
         a: Tensor,
-        n_past: int,
+        b: Tensor,
         n_dims: int,
         mode: int,
         n_ctx: int,
         ctx: Optional[Context] = None,
     ):
         ctx = ctx or Context.with_tensor_overhead()
-        op = ggml.ggml_rope_inplace(ctx.context, a.tensor, n_past, n_dims, mode, n_ctx)
+        op = ggml.ggml_rope_inplace(ctx.context, a.tensor, b.tensor, n_dims, mode, n_ctx)
         return Tensor.with_buffer(tensor=op, ctx=ctx, src=[a])
 
     @staticmethod
     def rope_back(
         a: Tensor,
-        n_past: int,
+        b: Tensor,
         n_dims: int,
         mode: int,
+        n_ctx: int,
+        freq_base: float,
+        freq_scale: float,
+        xpos_base: float,
+        xpos_down: bool,
         ctx: Optional[Context] = None,
     ):
         ctx = ctx or Context.with_tensor_overhead()
         op = ggml.ggml_rope_back(
             ctx.context,
             a.tensor,
-            n_past,
+            b.tensor,
             n_dims,
             mode,
+            n_ctx,
+            freq_base,
+            freq_scale,
+            xpos_base,
+            xpos_down,
         )
         return Tensor.with_buffer(tensor=op, ctx=ctx, src=[a])
 
@@ -1118,11 +1128,3 @@ class CGraph:
         obj = CGraph(cgraph=ggml.ggml_build_forward(tensor.tensor))
         obj._tensors.append(tensor)
         return obj
-
-    @classmethod
-    def graph_import(cls, fname: bytes, ctx: Context) -> CGraph:
-        return CGraph(
-            cgraph=ggml.ggml_graph_import(
-                fname, ctypes.pointer(ctx.context), ctypes.pointer(ctx.context)
-            ),
-        )