Text Generation Inference (TGI): Hugging Face's Production Server

Text Generation Inference (TGI): Hugging Face's Production Server

Hugging Face's Text Generation Inference (TGI) is the production-grade LLM server behind Hugging Face's Inference Endpoints and used by major companies. For local deployment, it offers continuous batching, flash attention, tensor parallelism, and quantization support in a single Docker container. Here's when and how to use it.

When to use TGI: TGI shines for multi-user serving where throughput matters more than single-user latency. For personal use, Ollama or llama.cpp server is simpler. For GPU hardware suitable for TGI serving, see LLMHardware.io's server GPU guide.

Quick Start with Docker

TGI runs as a Docker container: docker run --gpus all -p 8080:80 -v $PWD/models:/data ghcr.io/huggingface/text-generation-inference --model-id meta-llama/Llama-3.1-8B-Instruct --huggingface-hub-token YOUR_TOKEN. The container downloads the model automatically. TGI exposes an OpenAI-compatible endpoint and its own REST API with streaming support. The --quantize bitsandbytes-nf4 flag enables 4-bit quantization without separate model preparation.

Key TGI Features

Continuous batching (Orca scheduling) dynamically groups requests for maximum GPU utilization — idle GPU cycles between user requests are eliminated. PagedAttention manages KV cache memory like OS virtual memory, allowing more concurrent users per GPU. Guided generation (structured output) enforces JSON schemas or regex patterns on model outputs, making it excellent for tool calling and structured data extraction pipelines.

TGI vs vLLM vs Ollama

TGI and vLLM have comparable throughput for multi-user serving. TGI has slightly better HuggingFace model compatibility and simpler Docker setup. vLLM has more community momentum and better OpenAI API compatibility. Ollama is significantly easier to operate but has lower multi-user throughput. For a single developer or small team, Ollama wins on simplicity. For 10+ concurrent users, TGI or vLLM is the right choice.

Comments

Post a Comment

Popular posts from this blog

AMD RX 9070 XT for Local LLMs: ROCm Performance Guide (2026)

The AMD RX 9070 XT launched in early 2026 at $599 with 16GB GDDR6 and 648 GB/s memory bandwidth. For local LLM users who want a mid-range GPU without NVIDIA prices, this changes the calculus significantly. Here is what it actually delivers. RX 9070 XT Specs vs Competition GPU VRAM Bandwidth Price LLM Score RX 9070 XT 16 GB 648 GB/s $599 ~60 tok/s RTX 4070 Ti Super 16 GB 672 GB/s $799 ~62 tok/s RTX 4080 Super 16 GB 736 GB/s $999 ~68 tok/s RTX 4090 24 GB 1,008 GB/s ~$1,100 ~90 tok/s At $599 with 16GB VRAM, the RX 9070 XT is the best value GPU for local LLMs in the mid-range tier in 2026. See the full comparison at LLMHardware.io's AMD GPU guide . ROCm Setup on Linux (2026) # Ubuntu 22.04/24.04 sudo apt install rocm-hip-libraries hip-base # or use the AMD installer: wget https://repo.radeon.com/amdgpu-install/6.3/ubuntu/jammy/amdgpu-install_6.3.60300-1_all.deb sudo dpkg -i amdgpu-install*.deb sudo amdgpu-install --usecase=rocm,lrt,openclsdk # Verify ROCm sees your GPU r...
Read more

llama.cpp flash attention: how to enable and benchmark

Image
Flash attention computes attention with far less memory traffic. The math is identical; the memory movement is not. In llama.cpp it matters most in two situations: long context windows and tight video memory (VRAM). With flash attention enabled, the key-value cache shrinks substantially. A 32k-token context window becomes realistic on a 16GB card, and the prompt-processing phase speeds up, so long documents get ingested faster. The catch is compatibility: not every model and quantization combination supports it, and on some setups the speedup is modest. If you run short chats on a strong GPU, you may barely notice. Where it shines is retrieval-augmented generation (RAG) and long-document work on mid-range cards. Enable it, compare your tokens per second, and keep it if your workload benefits. For the common case of pasting a long PDF and asking questions, it is close to free performance. See how context length changes VRAM needs .
Read more

SillyTavern with Local LLMs: Complete Setup for AI Companions

Image
SillyTavern is the most feature-rich frontend for character-based AI conversations. What began as an AI roleplay tool has evolved into a powerful platform with persona management, memory systems, extensive prompt engineering tools, and connections to every major local inference backend. Here's how to set it up with local models. Quick connect: SillyTavern connects to Ollama, KoboldCpp, llama.cpp server, and TabbyAPI (ExLlamaV2) out of the box. Any of the hardware setups at LLMHardware.io provides a solid backend for SillyTavern. Installation Install via git: git clone https://github.com/SillyTavern/SillyTavern && cd SillyTavern && npm install then node server.js . Access at localhost:8000. For persistent installation, use the included start.sh script. SillyTavern uses Node.js and runs on any platform. The companion app (STMP) enables mobile access to your SillyTavern instance via Tailscale or local network. Character Cards and the TavernAI Format Chara...
Read more