anaksor-ocr
Sokheng · Kit
Reads Khmer and English text from documents, receipts and photos of signs, using a detector for text lines and a recogniser for each line. Suits developers who…
A browser simulator for LLM serving schedulers with an animated Gantt chart that compares continuous batching, chunked prefill, shortest-job-first and disaggregated serving. For engineers and students.
sched-sim is a kit in the Education category, listed by Mindket. It is free. This is version 1, last updated on . Download it on this page, a zip of its files. Nothing in it runs by itself.
pip install -U huggingface_hub hf download prajwalpujari16/sched-sim --repo-type space --local-dir sched-sim cd sched-sim
sched-sim is a browser simulator for how LLM serving schedulers order work. Behind the page is an event-by-event simulation of three ideas: batching that never pauses, prefill split into pieces, and a long prefill that holds up the requests queued behind it. The result plays as an animated Gantt chart. It is a static page, so nothing runs on a server.
It suits people who want to see why scheduling choices change how long requests wait.
| Industry | Who uses it | How they use it | Example |
|---|---|---|---|
| ML infrastructure | Serving engineers | Compare scheduling policies before changing a system | Play the blocking case and the chunked case one after the other |
| Education | Students | Learn how a serving scheduler works | Watch the Gantt player step through requests |
| AI research | Researchers | Reason about latency effects of scheduling | Try the shortest-job-first policy |
If you need to tune a live serving stack with real traffic, you can skip it, since this is a simulation.
You can use it online or open its files on your own computer.
| Use case | Who | What they get back |
|---|---|---|
| See prefill blocking | Students | An animation of one long prompt holding up shorter ones |
| Compare policies | Engineers | Gantt views for chunked prefill, shortest-job-first and disaggregated serving |
| Teach scheduling | Instructors | A visual to walk a class through |
The hosted page needs nothing. To get the files you need Python and pip.
pip install -U huggingface_hub hf download prajwalpujari16/sched-sim --repo-type space --local-dir sched-sim cd sched-sim
The readme also names a PyPI package, which installs with this command.
pip install sched-sim
Open the Space online, or open index.html from your downloaded folder. Then press play on the Gantt player and compare the policies it demonstrates.
| Tool | What it does |
|---|---|
| Head-of-Line prefill blocking | Shows how a long prefill delays the requests behind it |
| Chunked prefill | Splits prefill into chunks of 512 tokens, the Sarathi approach |
| Shortest-job-first | Orders requests so the shortest run first |
| Disaggregated serving | Separates the stages of serving across resources |
You play the Head-of-Line blocking scene and watch short requests queue behind a long prefill. Then you play the chunked prefill scene and compare how the same requests line up on the Gantt chart.
If you are new to this, start with the blocking scene. Prefill is the step where the model reads the prompt, and a long one can hold up everything behind it.
If you know the field, the page runs a discrete-event simulation and compares Head-of-Line blocking against Sarathi-style chunked prefill with a 512-token chunk, shortest-job-first and disaggregated serving. It names a serving traces dataset.
This is a simulation of scheduling, not a live serving engine.
A browser simulator for LLM serving schedulers with an animated Gantt chart that compares continuous batching, chunked prefill, shortest-job-first and disaggregated serving. For engineers and students.
Made by prajwalpujari16 on Hugging Face. These notes are Mindket's, written from its Space page and README; the app's own files and license are on Hugging Face.
What it is: sched-sim is a browser simulator for how LLM serving schedulers order work. Behind the page is an event-by-event simulation of three ideas: batching that never pauses, prefill split into pieces, and a long prefill that holds up the requests queued behind it. The result plays as an animated Gantt chart. It is a static page, so nothing runs on a server.
Who it suits: It suits people who want to see why scheduling choices change how long requests wait.
Title: Who might use this simulator
Industry | Who uses it | How they use it | Example
ML infrastructure | Serving engineers | Compare scheduling policies before changing a system | Play the blocking case and the chunked case one after the other
Education | Students | Learn how a serving scheduler works | Watch the Gantt player step through requests
AI research | Researchers | Reason about latency effects of scheduling | Try the shortest-job-first policy
If you need to tune a live serving stack with real traffic, you can skip it, since this is a simulation.
Where to use it: You can use it online or open its files on your own computer.
Title: What the simulator is good for
Use case | Who | What they get back
See prefill blocking | Students | An animation of one long prompt holding up shorter ones
Compare policies | Engineers | Gantt views for chunked prefill, shortest-job-first and disaggregated serving
Teach scheduling | Instructors | A visual to walk a class through
How to set it up: The hosted page needs nothing. To get the files you need Python and pip.
pip install -U huggingface_hub
hf download prajwalpujari16/sched-sim --repo-type space --local-dir sched-sim
cd sched-sim
The readme also names a PyPI package, which installs with this command.
pip install sched-sim
How to use it: Open the Space online, or open index.html from your downloaded folder. Then press play on the Gantt player and compare the policies it demonstrates.
Title: What the Gantt player demonstrates
Tool | What it does
Head-of-Line prefill blocking | Shows how a long prefill delays the requests behind it
Chunked prefill | Splits prefill into chunks of 512 tokens, the Sarathi approach
Shortest-job-first | Orders requests so the shortest run first
Disaggregated serving | Separates the stages of serving across resources
Example: You play the Head-of-Line blocking scene and watch short requests queue behind a long prefill. Then you play the chunked prefill scene and compare how the same requests line up on the Gantt chart.
For newcomers and experts: If you are new to this, start with the blocking scene. Prefill is the step where the model reads the prompt, and a long one can hold up everything behind it.
If you know the field, the page runs a discrete-event simulation and compares Head-of-Line blocking against Sarathi-style chunked prefill with a 512-token chunk, shortest-job-first and disaggregated serving. It names a serving traces dataset.
Open the simulator
Pick a scheduling policy
The simulation runs the requests
The Gantt player animates the result
Good to know: This is a simulation of scheduling, not a live serving engine.
Open the Space in your browser, or copy it to your own Hugging Face account from its page. Nothing in this kit runs by itself.
README.md 5 KB
Listed Hugging Face Spaces in the same category that share the most tags with this one.
| Product | Likes | Built with | License | Last change |
|---|---|---|---|---|
| sched-sim This page | 1 | static | apache-2.0 | |
| The Ultra-Scale PlaybookShares static | 4,131 | static | apache-2.0 | |
| LLM训练终极指南 | The Ultra-Scale PlaybookShares static | 281 | static | apache-2.0 | |
| Attention VisualizationShares static | 200 | static | apache-2.0 | |
| Number Tokenization BlogShares static | 127 | static | mit | |
| Train the agent slidesShares static | 20 | static | apache-2.0 |
Sokheng · Kit
Reads Khmer and English text from documents, receipts and photos of signs, using a detector for text lines and a recogniser for each line. Suits developers who…
Qwen · Kit
Qwen2.5-7B-Instruct as GGUF files in eight quantization sizes, to chat with locally through llama.cpp. For people who want a capable 7.6B model on their own co…
ACE-Step · Kit
Generates music from text with a 4B diffusion decoder, 50 steps and guidance control, aiming for higher audio quality than the base. For creators with a GPU of…
prathoshap · Kit
Reads classical Sanskrit verses aloud in a chant style, using one trained voice and a reference clip that sets the rhythm. For people who study, recite or need…
paradigma-inc · Kit
Solves hard math problems with a 1B-parameter model trained from scratch for fast, high-volume reasoning. It suits developers who run math workloads on an NVID…
murai-labs · Kit
Chats in Tamil script, romanized Tamil and English, and translates between Tamil and English. A 2.5B model built from scratch by Murai Labs, for Tamil speakers…