resin README

README проекта с github.com/ekzhang/resin. Одним файлом markdown: README.md.

Resin

Resin is not an inference engine. It's a library of building blocks for LLM inference.

At its core is the headless Engine: a fast Rust scheduler, KV cache, and metrics server. You then write the forward pass as model-specific kernels, optimized for your hardware.

token IDs ──► Rust Engine ──► frozen BatchPlan ──► Python Runner ───► token IDs
                  ▲                                    │
                  └──────── ownership & commit ────────┘

Resin also has optional HTTP serving and Prometheus metrics.

[!WARNING] This is not production-quality software. If you're looking for an actual inference engine, please use SGLang or vLLM. Read on if you want to hack on LLM inference systems, new models or hardware.

With the right kernels, this makes it possible to autonomously "vibe code" even the most complex models. In 12 hours, we used Resin to build standalone engines for DeepSeek-V4.1-Flash and GLM-5.3-Flash on Blackwell GPUs that rival vLLM and SGLang.

Prefill Decode
DeepSeek V4.1 Flash prefill DeepSeek V4.1 Flash interactivity-throughput curve
Prefill and decode for DeepSeek V4.1 Flash, within 5-10% of DeepInfer.

Features

Start with the two-page architecture overview.

Getting started

Either fork the repository or import resin and resin_models_cuda as standalone packages. Then use the following skills:

Importing resin does not import PyTorch or initialize CUDA.

Installation

Install the core library directly from GitHub with Python 3.11+ and Rust installed:

pip install "git+https://github.com/ekzhang/resin.git"

Pip builds the Rust extension locally. The import remains resin. Use @<commit> after .git to pin a revision.

Model runners are separate packages: resin-models-cuda for optimized CUDA implementations and resin-models-torch for reference implementations. The repository's development workspace installs both; the examples below assume that workspace or a separately installed runner package.

Development

Resin requires Python 3.11+ and Rust.

uv run scripts/rust_build_and_types.py

uv run syncs the project dependencies automatically. The build command then syncs type-library sources and builds the Rust extension with generated stubs.

The Modal harnesses are launched from the repository root. Install the development dependencies first, then authenticate with Modal once if this is a new machine:

uv run modal setup

modal setup is interactive and only needs to be repeated when the local Modal credentials change.

All models share one benchmark CLI and one parameterized serving app. The model registry in deploy/models.py supplies checkpoint revisions, hardware, parallelism, and model-specific runner construction.

# Populate the pinned checkpoint without allocating a GPU.
uv run scripts/bench.py --backend modal --model gpt-oss-20b --download-only

# Sweep prefill and decode, save traces, and plot the measured results.
uv run scripts/bench.py --backend modal --model gpt-oss-20b \
  --context-lengths 4096,8192 --batch-sizes 1,2,4,8 \
  --output bench/gpt-oss-20b.jsonl --trace-dir bench/traces --plot

For local GPUs, use --backend local --checkpoint /path/to/checkpoint; models with a separate draft checkpoint also accept --assistant-checkpoint. The local backend is the default. Modal submits the repository's source into the shared CUDA image, so the client machine does not need CUDA installed.

Prefill charts compare complete-context latency with input tokens/s/GPU; decode charts compare per-user output tokens/s with output tokens/s/GPU. Draft acceptance uses normal verification; the plotter's --accept-rate projects other acceptance rates. Model routing is the default; --balanced-experts enables synthetic balanced routing. See benchmarking for the timing protocol and trace scope.

For an already running endpoint, scripts/check_server.py URL --output REPORT checks serving correctness, prefix-cache reuse, context-size TTFT, and optional image/reasoning/stress cases. It can also evaluate a small local MMLU CSV or JSONL subset with --eval-file (defaults: --eval-limit 100, --eval-concurrency 32); see server checks.

Minimal use

from resin import Engine, SamplingParams
from resin_models_torch.lfm2 import Lfm2TorchRunner

runner = Lfm2TorchRunner(device="cpu")

with Engine(runner, context_length=32768) as engine:
    request = engine.submit(
        [1, 6, 24131],
        sampling_params=SamplingParams(
            max_output_tokens=8,
            stop_token_ids=[runner.eos_token_id],
            temperature=1.0,
            top_p=0.95,
        ),
    )
    while engine.has_work():
        for event in engine.step():
            print(event)
    generated_ids = request.take_result()

SamplingParams is the frozen request policy for generation limits, stop token IDs, sampling, and logprob options. Use the recipe's stop_token_ids for chat requests, or supply the model's stop tokens explicitly for raw token requests. Engine has no default EOS token.

A custom synchronous runner only needs one method:

import numpy as np

from resin import BatchPlan, ForwardOutput


class Runner:
    def forward(self, plan: BatchPlan) -> ForwardOutput:
        return ForwardOutput(
            token_ids=np.zeros(plan.num_sequences, dtype=np.uint32),
            token_indptr=np.arange(plan.num_sequences + 1, dtype=np.uint32),
        )

ForwardOutput carries flat token IDs and row offsets, with optional logprob columns. The same result type covers ordinary and speculative decoding.

Managed history and recurrent state are cache groups declared through a frozen RunnerConfig. The Engine binds their one physical arena during construction. Pass prefix_cache=True to reuse aligned prefixes across requests.

Local HTTP server

The built-in native server exposes one Engine through OpenAI Chat Completions and Responses, Anthropic Messages, and a small demo chat page. Model-specific prompt rendering and incremental response parsing live in the independent resin-recipe Rust crate.

# Run one LFM2 prompt in the terminal.
uv run python scripts/lfm2_chat.py

# Serve LFM2 locally over HTTP.
uv run python scripts/lfm2_serve.py

Then open http://127.0.0.1:8000. The page reports prompt and generated token counts, elapsed generation time, and output tokens per second. The same process serves /v1/chat/completions, /v1/responses, /v1/messages, and /v1/models. Both examples explicitly use the CPU Torch runner.

Modal HTTP endpoints

deploy/modal_serve.py serves every registered model. Select RESIN_MODEL at launch; each selection has a separate app and endpoint, with shared download, warmup, process management, and HTTP startup code. Populate its pinned checkpoint before deploying:

RESIN_MODEL=gpt-oss-20b uv run modal run deploy/modal_serve.py
RESIN_MODEL=gpt-oss-20b uv run modal deploy deploy/modal_serve.py

RESIN_MODEL=deepseek-v41 uv run modal run deploy/modal_serve.py
RESIN_MODEL=deepseek-v41 uv run modal deploy deploy/modal_serve.py

Use uv run scripts/bench.py --help to see the current model keys accepted by RESIN_MODEL and the benchmark CLI. The shared MODEL_SPECS registry defines each selection's checkpoint, hardware, and serving limits.

MLX runner (Apple Silicon)

Model runners are not tied to CUDA. resin_models_mlx runs dense and MoE Qwen3.5 models on Apple Silicon with MLX, on the same core engine. For a small local model:

QWEN_CHECKPOINT="$(uv run hf download mlx-community/Qwen3.5-0.8B-8bit \
  --revision 87e768fbfa03994095f3d14527c80c5ae70c5758 --quiet)"
uv run scripts/mlx_serve.py --model qwen3.5-0.8b-mlx --checkpoint "$QWEN_CHECKPOINT"

The download uses Hugging Face's normal cache and returns its snapshot path.

Checks

cargo test --all-targets
uv run pytest -q
uv run ruff format --check .
uv run ruff check .
uv run ty check
uv run python scripts/cloc.py

GPU validation is model- and machine-specific; local tests use fakes or skip unavailable accelerator libraries.

The code-size report uses Tokei to count production Rust/Python by component, excluding comments, blank lines, tests, deployment tools, and generated stubs. Use the current report instead of a checked-in line-count snapshot.

For resident prefill/decode sweeps, optional fixed speculative acceptance and balanced routing, Perfetto traces, and throughput/latency charts, see benchmarking.

Acknowledgements

Resin was based on and inspired by systems from SGLang, vLLM, TokenSpeed, and DeepInfer.

resin (ekzhang/resin)

Ищем в базе…

437 ·GitHub
Что это такое?Resin — экспериментальная библиотека строительных блоков для создания собственных движков инференса LLM, а не готовый движок инференса.
Это готовый движок инференса или то, что я сам собираю?Это библиотека без интерфейса: основной Engine предоставляет планировщик на Rust, KV-кэш и сервер метрик, а ядра прямого прохода под конкретную модель вы пишете сами.
Нужны ли PyTorch или настройка CUDA, чтобы просто импортировать библиотеку?Нет, импорт resin сам по себе не подтягивает PyTorch и не инициализирует CUDA; CUDA нужна только при установке и использовании необязательных пакетов с runner'ами моделей.
Это годится для продакшена?Нет, авторы прямо предупреждают, что качество не продакшенное, и для настоящего движка инференса советуют SGLang или vLLM.
Что уже построили с помощью Resin?За 12 часов команда с помощью Resin создала автономные движки для DeepSeek-V4.1-Flash и GLM-5.3-Flash на GPU Blackwell; по производительности они сопоставимы с vLLM и SGLang и отстают от DeepInfer не более чем на 5–10%.
Что входит в основной движок из коробки?Планировщик пакетов на Rust, единый пул кэша на основе слэбов (KV/SWA/Recurrent), спекулятивное декодирование (MTP, DSpark), CUDA-графы для prefill и decode, а также параллелизм TP/EP/CP/DP.
Можно ли раздавать модели по HTTP?Да, есть необязательный встроенный нативный сервер, который предоставляет OpenAI Chat Completions/Responses, Anthropic Messages и эндпоинт /metrics для Prometheus.
Что такое навыки add-resin-model и optimize-resin-performance?Это пошаговые сценарии: один помогает создать движок инференса для новой модели с нуля, а другой — итеративно профилировать и улучшать его производительность.
Автор
ekzhang
Размер
7.9 MB
Звёзды
437
Обновлён
КоммитNov 17
Возраст
3 мес
Коммиты · 90 дней