Skip to content

zhongkaifu/TensorSharp

BranchesTags

Repository files navigation

TensorSharp

English | 中文

A C# inference engine for running large language models (LLMs) locally using GGUF model files. TensorSharp provides a console application, a web-based chatbot interface, and Ollama/OpenAI-compatible HTTP APIs for programmatic access.

Features

  • Multi-architecture support -- Gemma 4, Gemma 3, Qwen 3, Qwen 3.5
  • Multimodal inference -- image, video, and audio inputs (Gemma 4); images for Gemma 3 / Qwen 3.5
  • Thinking / reasoning mode -- structured chain-of-thought output with <think> / <|channel>thought tags (Qwen 3, Qwen 3.5, Gemma 4)
  • Tool calling / function calling -- models can invoke user-defined tools; multi-turn tool-call conversations supported across all three API styles
  • Quantized model support -- loads GGUF files with Q4_K_M, Q8_0, F16, and other quantization formats; performs native quantized matmul without dequantizing to FP32, including memory-efficient pure C# CPU loading for large GGUFs
  • GPU-accelerated -- GGML Metal on macOS and GGML CUDA on Linux/NVIDIA, with fused whole-model GPU dispatch for Gemma 4 decode on Metal (~2.6x speedup over per-op dispatch)
  • Optimized pure C# CPU backend -- managed GEMM fast paths plus fused SIMD kernels for RMSNorm, RoPE, softmax, fused activations, and other inference hot paths
  • Ollama & OpenAI API compatibility -- drop-in replacement endpoints for existing tooling
  • Configurable sampling -- temperature, top-k, top-p, min-p, repetition/presence/frequency penalties, seed, stop sequences
  • Chat templates -- auto-loaded from GGUF metadata (Jinja2), with hardcoded fallbacks per architecture
  • Request queue -- FIFO inference queue ensures single-request execution for KV cache stability, with real-time position tracking for clients
  • Batch processing -- JSONL input support in the console application
  • Streaming -- token-by-token output via SSE (web) or stdout (console)
  • Mixture of Experts -- Gemma 4 MoE variants (e.g. gemma-4-26B-A4B)
  • Large file uploads -- supports video/audio uploads up to 500 MB in the web interface

Supported Model Architectures

Architecture Example Models Multimodal Thinking Tool Calling
Gemma 4 gemma-4-E4B, gemma-4-31B, gemma-4-26B-A4B (MoE) Image, Video, Audio Yes Yes
Gemma 3 gemma-3-4b Image No No
Qwen 3 Qwen3-4B Text only Yes Yes
Qwen 3.5 Qwen3.5-9B Image Yes Yes

Compute Backends

Backend Flag Description
GGML Metal --backend ggml_metal GPU-accelerated via Apple Metal (macOS). Recommended for Apple Silicon.
GGML CUDA --backend ggml_cuda GPU-accelerated via GGML CUDA on Linux with an NVIDIA GPU.
GGML CPU --backend ggml_cpu CPU inference using native GGML with optimized kernels.
Pure C# CPU --backend cpu Portable CPU inference with no native dependencies.

Project Structure

TensorSharp/
├── TensorSharp/                 # Core tensor library (CPU operations, SIMD)
├── TensorSharp.GGML/            # GGML backend bindings (Metal/CUDA/CPU via native library)
├── TensorSharp.GGML.Native/     # Native C++ bridge to ggml (builds libGgmlOps)
├── AdvUtils/                    # Utility library
├── InferenceEngine/             # Model loading, tokenization, and inference logic
│   ├── Models/
│   │   ├── Gemma3/
│   │   ├── Gemma4/              # Vision encoder, audio encoder, MoE, fused GPU decode
│   │   ├── Qwen3/
│   │   └── Qwen35/
│   ├── GgufReader.cs            # GGUF file parser
│   ├── ModelBase.cs             # Base class for all model architectures
│   ├── ChatTemplate.cs          # Chat template rendering (hardcoded + Jinja2 from GGUF)
│   ├── Jinja2Template.cs        # Jinja2 template renderer
│   ├── OutputParser.cs          # Extracts thinking, content, and tool calls from model output
│   ├── SamplingConfig.cs        # Sampling parameter configuration
│   ├── TokenSampler.cs          # Token sampling (greedy, top-k, top-p, min-p, penalties)
│   └── MediaHelper.cs           # Video frame extraction, audio decoding
├── InferenceConsole/            # CLI application
├── InferenceWeb/                # Web chatbot + API server (ASP.NET Core)
│   ├── ModelService.cs          # Model lifecycle management
│   ├── InferenceQueue.cs        # FIFO request queue with position tracking
│   ├── wwwroot/index.html       # Chat UI
│   ├── testdata/                # Integration test suites (bash + Python)
│   └── API_EXAMPLES.md          # Detailed API documentation
└── ExternalProjects/            # Third-party dependencies (ggml)

Prerequisites

  • .NET 10 SDK
  • macOS (Metal backend): CMake 3.20+ and Xcode command-line tools for building the native GGML library
  • Linux (GGML CPU / CUDA backends): CMake 3.20+; for ggml_cuda, install an NVIDIA driver plus CUDA Toolkit 12.x or another compatible CUDA toolkit
  • GGUF model files (e.g., from Hugging Face)

Building

Build the entire solution

dotnet build TensorSharp.slnx

Build individual applications

# Console application
dotnet build InferenceConsole/InferenceConsole.csproj

# Web application
dotnet build InferenceWeb/InferenceWeb.csproj

Build the native GGML library

The native library is built automatically during the first dotnet build if it doesn't exist. To build it manually:

cd TensorSharp.GGML.Native

macOS:

bash build-macos.sh

Linux (CPU-only):

bash build-linux.sh

Linux (GGML_CUDA enabled):

bash build-linux.sh --cuda

You can also request a CUDA-enabled native build from dotnet build:

TENSORSHARP_GGML_NATIVE_ENABLE_CUDA=ON dotnet build InferenceConsole/InferenceConsole.csproj -c Release

On macOS this compiles libGgmlOps.dylib with Metal GPU support. On Linux, build-linux.sh builds libGgmlOps.so with the GGML CPU backend by default, and build-linux.sh --cuda enables GGML_CUDA support for NVIDIA GPUs. The build output is automatically copied to the application's output directory.

Usage

Console Application

cd InferenceConsole/bin

# Text inference
./InferenceConsole --model <model.gguf> --input prompt.txt --output result.txt \
    --max-tokens 200 --backend ggml_metal

# Text inference on Linux + NVIDIA GPU
./InferenceConsole --model <model.gguf> --input prompt.txt --output result.txt \
    --max-tokens 200 --backend ggml_cuda

# Image inference (Gemma 3/4, Qwen 3.5)
./InferenceConsole --model <model.gguf> --image photo.png --backend ggml_metal

# Video inference (Gemma 4)
./InferenceConsole --model <model.gguf> --video clip.mp4 --backend ggml_metal

# Audio inference (Gemma 4)
./InferenceConsole --model <model.gguf> --audio speech.wav --backend ggml_metal

# Thinking / reasoning mode
./InferenceConsole --model <model.gguf> --input prompt.txt --backend ggml_metal --think

# Tool calling
./InferenceConsole --model <model.gguf> --input prompt.txt --backend ggml_metal \
    --tools tools.json

# With sampling parameters
./InferenceConsole --model <model.gguf> --input prompt.txt --backend ggml_metal \
    --temperature 0.7 --top-p 0.9 --top-k 40 --repeat-penalty 1.2 --seed 42

# Batch processing (JSONL)
./InferenceConsole --model <model.gguf> --input-jsonl requests.jsonl \
    --output results.txt --backend ggml_metal

Command-line options:

Option Description
--model <path> Path to a GGUF model file (required)
--input <path> Text file containing the user prompt
--input-jsonl <path> JSONL file with batch requests (one JSON per line)
--output <path> Write generated text to this file
--image <path> Image file for vision inference
--video <path> Video file for video inference
--audio <path> Audio file (WAV, MP3, OGG) for audio inference
--mmproj <path> Path to the multimodal projector GGUF file
--max-tokens <N> Maximum tokens to generate (default: 100)
--backend <type> Compute backend: cpu, ggml_cpu, ggml_metal, or ggml_cuda
--think Enable thinking/reasoning mode (chain-of-thought)
--tools <path> JSON file with tool/function definitions
--temperature <f> Sampling temperature (0 = greedy)
--top-k <N> Top-K filtering (0 = disabled)
--top-p <f> Nucleus sampling threshold (1.0 = disabled)
--min-p <f> Minimum probability filtering (0 = disabled)
--repeat-penalty <f> Repetition penalty (1.0 = none)
--presence-penalty <f> Presence penalty (0 = disabled)
--frequency-penalty <f> Frequency penalty (0 = disabled)
--seed <N> Random seed (-1 = non-deterministic)
--stop <string> Stop sequence (can be repeated)
--test Run built-in test suite

The multimodal projector file is auto-detected if placed alongside the model file with a recognized name (e.g., gemma-4-mmproj-F16.gguf).

JSONL input format:

Each line is a JSON object with messages, optional prompt, and optional sampling parameters:

{"id": "q1", "messages": [{"role": "user", "content": "What is 2+3?"}], "max_tokens": 50}
{"id": "q2", "messages": [{"role": "user", "content": "Write a haiku."}], "max_tokens": 100, "temperature": 0.8}

Web Application

cd InferenceWeb/bin

# Set environment variables and run
MODEL_DIR=./models BACKEND=ggml_metal ./InferenceWeb

# Linux + NVIDIA GPU
MODEL_DIR=./models BACKEND=ggml_cuda ./InferenceWeb

Open http://localhost:5000 in your browser. The web interface supports:

  • Multi-turn chat conversations
  • Model selection from available GGUF files in MODEL_DIR
  • Image, video, and audio uploads for multimodal inference (up to 500 MB)
  • Thinking/reasoning mode toggle
  • Tool calling with function definitions
  • Streaming token generation via Server-Sent Events
  • Request queue with real-time position feedback

Environment variables:

Variable Description
MODEL_DIR Directory containing GGUF model files
BACKEND Compute backend: cpu, ggml_cpu, ggml_metal, or ggml_cuda (default: ggml_metal on macOS, ggml_cpu elsewhere)
PORT HTTP port (default: 5000)

HTTP APIs

InferenceWeb exposes three API styles. See API_EXAMPLES.md for full documentation with curl and Python examples.

Ollama-compatible API:

# List models
curl http://localhost:5000/api/tags

# Generate text
curl -X POST http://localhost:5000/api/generate \
  -H "Content-Type: application/json" \
  -d '{"model": "Qwen3-4B-Q8_0.gguf", "prompt": "Hello!", "stream": false}'

# Chat
curl -X POST http://localhost:5000/api/chat/ollama \
  -H "Content-Type: application/json" \
  -d '{"model": "Qwen3-4B-Q8_0.gguf", "messages": [{"role": "user", "content": "Hi"}], "stream": false}'

# Chat with thinking mode
curl -X POST http://localhost:5000/api/chat/ollama \
  -H "Content-Type: application/json" \
  -d '{"model": "Qwen3-4B-Q8_0.gguf", "messages": [{"role": "user", "content": "Solve 17*23"}], "think": true, "stream": false}'

# Chat with tool calling
curl -X POST http://localhost:5000/api/chat/ollama \
  -H "Content-Type: application/json" \
  -d '{"model": "Qwen3-4B-Q8_0.gguf", "messages": [{"role": "user", "content": "What is the weather?"}], "tools": [{"function": {"name": "get_weather", "description": "Get current weather", "parameters": {"properties": {"city": {"type": "string"}}, "required": ["city"]}}}], "stream": false}'

OpenAI-compatible API:

# Chat completions
curl -X POST http://localhost:5000/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{"model": "Qwen3-4B-Q8_0.gguf", "messages": [{"role": "user", "content": "Hi"}], "max_tokens": 50}'

OpenAI Python SDK:

from openai import OpenAI

client = OpenAI(base_url="http://localhost:5000/v1", api_key="not-needed")
response = client.chat.completions.create(
    model="Qwen3-4B-Q8_0.gguf",
    messages=[{"role": "user", "content": "What is 2+3?"}],
    max_tokens=50
)
print(response.choices[0].message.content)

Queue status:

curl http://localhost:5000/api/queue/status
# {"busy":false,"pending_requests":0,"total_processed":42}

Thinking / Reasoning Mode

Models that support thinking mode (Qwen 3, Qwen 3.5, Gemma 4) can produce structured chain-of-thought reasoning before generating the final answer. The thinking content is separated from the main response and can be displayed or hidden by the client.

  • Qwen 3 / Qwen 3.5: uses <think>...</think> tags
  • Gemma 4: uses <|channel>thought\n...<channel|> tags

Enable via --think (console), "think": true (Ollama API), or the thinking toggle in the web UI.

Tool Calling / Function Calling

Models can invoke user-defined tools and participate in multi-turn tool-call conversations. Define tools as JSON and pass them via --tools (console) or the tools parameter in the API.

Each architecture uses its own wire format for tool calls:

  • Qwen 3 / Qwen 3.5: <tool_call>{"name": "...", "arguments": {...}}</tool_call>
  • Gemma 4: <|tool_call>call:function_name{args}<tool_call|>

The output parser (OutputParser.cs) automatically extracts tool calls from the model's raw output regardless of architecture.

Multimodal Support

Gemma 4

Gemma 4 models support image, video, and audio inputs. Place the multimodal projector (gemma-4-mmproj-F16.gguf) in the same directory as the model file for automatic loading.

  • Images: PNG, JPEG
  • Video: MP4 (extracts up to 8 frames at 1 fps using OpenCV)
  • Audio: WAV (16kHz mono), MP3, OGG Vorbis

Gemma 3 / Qwen 3.5

These models support image inputs with their respective multimodal projector files.

Architecture

TensorSharp is structured as a layered system:

  1. TensorSharp provides the core Tensor type, storage abstraction, and an extensible operation registry (Ops). CPU implementations use System.Numerics.Vectors for SIMD acceleration.

  2. TensorSharp.GGML registers accelerated implementations of the same operations via a native C++ bridge (libGgmlOps) that links against ggml. On macOS this provides Metal GPU compute, and on Linux it can expose GGML CUDA for NVIDIA GPUs. Operations include native quantized matmul (Q4_K_M, Q8_0, etc.) without dequantizing to FP32.

  3. InferenceEngine implements model-specific logic: GGUF parsing, tokenization (SentencePiece BPE), chat template rendering (Jinja2 from GGUF metadata with hardcoded fallbacks), configurable token sampling, output parsing (thinking extraction, tool-call extraction), and the forward pass for each architecture. Models are loaded via ModelBase.Create() which auto-detects the architecture from GGUF metadata.

  4. InferenceConsole and InferenceWeb are application layers that handle I/O and user interaction. InferenceWeb provides Ollama-compatible and OpenAI-compatible REST APIs alongside a browser-based chat UI, with a FIFO inference queue to serialize concurrent requests.

Performance Optimizations

  • Fused GPU decode (Gemma 4): all transformer layers are executed in a single GGML compute graph dispatch on Metal, reducing CPU-GPU round-trips from hundreds per token to one. This achieves ~2.6x speedup over per-operation dispatch.
  • Fused weight projections: Q/K/V projections are fused into a single QKV matmul; gate and up projections are fused into a single gate_up matmul.
  • Native quantized compute: quantized weights (Q4_K_M, Q6_K, Q8_0, etc.) are used directly in matmul without expanding to FP32, saving memory and bandwidth.
  • Optimized pure C# CPU path: managed GEMM fast paths and contiguous float32 kernels accelerate decode, softmax, RMSNorm, RoPE, fused activations, and other hot paths while keeping quantized GGUF weights compressed during CPU loading.
  • Circular KV cache: sliding-window attention layers use a fixed-size circular buffer, bounding memory usage regardless of sequence length.
  • Memory-efficient model loading: large tensors are streamed directly to native memory without intermediate managed allocations.

Testing

Integration tests for InferenceWeb are in InferenceWeb/testdata/. They cover all three API styles (Web UI SSE, Ollama, OpenAI), multi-turn conversations, thinking mode, tool calling, queue behavior, concurrent requests, and abort support.

# Start InferenceWeb, then run:
python3 InferenceWeb/testdata/test_multiturn.py
# or
bash InferenceWeb/testdata/test_multiturn.sh

See InferenceWeb/testdata/README.md for the full test matrix.

Author

Zhongkai Fu

License

See LICENSE for details.

About

A C# inference engine for running large language models (LLMs) locally using GGUF model files. TensorSharp provides a console application, a web-based chatbot interface, and Ollama/OpenAI-compatible HTTP APIs for programmatic access.

Topics

metal transformers inference gemma inference-engine llm qwen gguf gemma3 qwen3 qwen35 gemma4

Resources

Readme

License

BSD-3-Clause license
Activity

Stars

2 stars

Watchers

1 watching

Forks

1 fork
Report repository

Releases

No releases published

Packages

 
 
 

Contributors

Languages