# datawhalechina/zero-to-sglang

> A Datawhale and RadixArk course on LLM inference: explains KV Cache, prefill/decode and related concepts, then walks through building a mini-sglang from scratch toward reading the real SGLang source and landing a first PR.

- Magnitude: 0.9 out of 10 — Steady
- Stars: 1,379 total · +5 stars measured 2026-10-05, 00:14–23:44 UTC
- Star trust: star growth looks organic
- Category: Learning & lists · Language: Python · Created: 2026-08-27 · Last push: 2026-10-02
- GitHub: https://github.com/datawhalechina/zero-to-sglang · Homepage: https://datawhalechina.github.io/zero-to-sglang/ · Page: https://gitnova.dev/en/r/datawhalechina/zero-to-sglang

## Useful for

- Understand how KV Cache works and how prefill differs from decode
- Build your own mini-sglang step by step following the course
- Prepare for a first SGLang PR by studying its source code

## Why it’s here

- Star-counter measurements on 2026-10-05 (UTC), 00:14–23:44: 1374 → 1379 stars (+5). This is the change over that interval.
- The repository is 39 days old and already has 1,379 stars.

## Star trust

Star growth looks organic.

Star-trust labels are heuristics based on the repository’s behavior, not a check of every stargazer.

## Numbers

- Forks: 106
- Issues and pull requests: 71
- Watchers: 7
- Average over the last week: 8 per day
- Usual pace: 34 per day

## Stars per day, last 30 days (oldest → newest, today is partial)

2026-09-30 … 2026-10-29: 10, 6, 8, 8, 4, 4, None, None, None, None, None, None, None, None, None, None, None, None, None, None, None, None, None, None, None, None, None, None, None, None

## Magnitude by day, last 14 days

09-22 2.0, 09-23 1.7, 09-24 1.7, 09-25 1.6, 09-26 1.3, 09-27 1.3, 09-28 1.3, 09-29 1.5, 09-30 1.3, 10-01 1.1, 10-02 1.1, 10-03 1.1, 10-04 1.1, 10-05 0.9

---

Magnitude (0–10) measures how fast and how unusually interest in a repository is growing right now. It is not a quality score.
Days are UTC. “So far today” is a fact; “expected by the end of the day” is a forecast.
Summaries and use cases are written by an LLM (DeepSeek V4.1 Flash) from the README and may be inaccurate: verify specific claims (benchmarks, speed, hardware) in the repository itself.
Data as of 2026-10-05 23:47 UTC, updated every 30 minutes.