README.md 15.9 KB
Newer Older
Lianmin Zheng's avatar
Lianmin Zheng committed
1
2
3
4
5
6
<div align="center">
<img src="assets/logo.png" alt="logo" width="400"></img>
</div>

--------------------------------------------------------------------------------

7
| [**Blog**](https://lmsys.org/blog/2024-01-17-sglang/) | [**Paper**](https://arxiv.org/abs/2312.07104) |
Lianmin Zheng's avatar
Lianmin Zheng committed
8

Ying Sheng's avatar
Ying Sheng committed
9
10
SGLang is a fast serving framework for large language models and vision language models.
It makes your interaction with models faster and more controllable by co-designing the backend runtime and frontend language.
Lianmin Zheng's avatar
Lianmin Zheng committed
11

12
The core features include:
Ying Sheng's avatar
Ying Sheng committed
13
- **Fast Backend Runtime**: Efficient serving with RadixAttention for prefix caching, jump-forward constrained decoding, continuous batching, token attention (paged attention), tensor parallelism, flashinfer kernels, and quantization (AWQ/FP8/GPTQ/Marlin).
Lianmin Zheng's avatar
Lianmin Zheng committed
14
- **Flexible Frontend Language**: Enables easy programming of LLM applications with chained generation calls, advanced prompting, control flow, multiple modalities, parallelism, and external interactions.
Lianmin Zheng's avatar
Lianmin Zheng committed
15

Ying Sheng's avatar
Ying Sheng committed
16
## News
Ying Sheng's avatar
Ying Sheng committed
17
- [2024/04] 🔥 SGLang is used by the official **LLaVA-NeXT (video)** release ([blog](https://llava-vl.github.io/blog/2024-04-30-llava-next-video/)).
Lianmin Zheng's avatar
Lianmin Zheng committed
18
19
- [2024/02] 🔥 SGLang enables **3x faster JSON decoding** with compressed finite state machine ([blog](https://lmsys.org/blog/2024-02-05-compressed-fsm/)).
- [2024/01] SGLang provides up to **5x faster inference** with RadixAttention ([blog](https://lmsys.org/blog/2024-01-17-sglang/)).
Ying Sheng's avatar
Ying Sheng committed
20

Ying Sheng's avatar
Ying Sheng committed
21
22
23
24
25
26
27
<details>
<summary>More</summary>

- [2024/01] SGLang powers the serving of the official **LLaVA v1.6** release demo ([usage](https://github.com/haotian-liu/LLaVA?tab=readme-ov-file#demo)).

</details>

Lianmin Zheng's avatar
Lianmin Zheng committed
28
29
30
## Contents
- [Install](#install)
- [Backend: SGLang Runtime (SRT)](#backend-sglang-runtime-srt)
Ying Sheng's avatar
Ying Sheng committed
31
- [Frontend: Structured Generation Language (SGLang)](#frontend-structured-generation-language-sglang)
Lianmin Zheng's avatar
Lianmin Zheng committed
32
33
34
35
36
37
- [Benchmark And Performance](#benchmark-and-performance)
- [Roadmap](#roadmap)
- [Citation And Acknowledgment](#citation-and-acknowledgment)

## Install

Lianmin Zheng's avatar
Lianmin Zheng committed
38
39
### Method 1: With pip
```
Lianmin Zheng's avatar
Lianmin Zheng committed
40
pip install --upgrade pip
Lianmin Zheng's avatar
Lianmin Zheng committed
41
pip install "sglang[all]"
Lianmin Zheng's avatar
Lianmin Zheng committed
42

Lianmin Zheng's avatar
Lianmin Zheng committed
43
44
45
# Install FlashInfer CUDA kernels
pip install flashinfer -i https://flashinfer.ai/whl/cu121/torch2.3/
```
46

Lianmin Zheng's avatar
Lianmin Zheng committed
47
### Method 2: From source
Lianmin Zheng's avatar
Lianmin Zheng committed
48
```
49
git clone https://github.com/sgl-project/sglang.git
Lianmin Zheng's avatar
Lianmin Zheng committed
50
51
cd sglang

Lianmin Zheng's avatar
Lianmin Zheng committed
52
pip install --upgrade pip
Lianmin Zheng's avatar
Lianmin Zheng committed
53
54
pip install -e "python[all]"

Lianmin Zheng's avatar
Lianmin Zheng committed
55
56
57
# Install FlashInfer CUDA kernels
pip install flashinfer -i https://flashinfer.ai/whl/cu121/torch2.3/
```
58

Lianmin Zheng's avatar
Lianmin Zheng committed
59
60
### Method 3: Using docker
The docker images are available on Docker Hub as [lmsysorg/sglang](https://hub.docker.com/r/lmsysorg/sglang/tags).
Ying Sheng's avatar
Ying Sheng committed
61

Liangsheng Yin's avatar
Liangsheng Yin committed
62
63
64
65
66
67
68
69
70
71
```bash
docker run --gpus all \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HUGGING_FACE_HUB_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server --model-path meta-llama/Meta-Llama-3-8B --host 0.0.0.0 --port 30000
```

Lianmin Zheng's avatar
Lianmin Zheng committed
72
### Common Notes
73
74
75
76
77
- If you see errors from the Triton compiler, please install the [Triton Nightly](https://triton-lang.org/main/getting-started/installation.html) by
```
pip uninstall -y triton triton-nightly
pip install -U --index-url https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/Triton-Nightly/pypi/simple/ triton-nightly
```
Lianmin Zheng's avatar
Lianmin Zheng committed
78
79
- If you cannot install FlashInfer, check out its [installation](https://docs.flashinfer.ai/installation.html#) page. If you still cannot install it, you can use the slower Triton kernels by adding `--disable-flashinfer` when launching the server.
- If you only need to use the OpenAI backend, you can avoid installing other dependencies by using `pip install "sglang[openai]"`.
Ying Sheng's avatar
Ying Sheng committed
80

Ying Sheng's avatar
Ying Sheng committed
81
82
83
## Backend: SGLang Runtime (SRT)
The SGLang Runtime (SRT) is an efficient serving engine.

Ying Sheng's avatar
Ying Sheng committed
84
### Quick Start
Ying Sheng's avatar
Ying Sheng committed
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
Launch a server
```
python -m sglang.launch_server --model-path meta-llama/Meta-Llama-3-8B-Instruct --port 30000
```

Send a request
```
curl http://localhost:30000/generate \
  -H "Content-Type: application/json" \
  -d '{
    "text": "Once upon a time,",
    "sampling_params": {
      "max_new_tokens": 16,
      "temperature": 0
    }
  }'
```
Learn more about the argument format [here](docs/sampling_params.md).

### OpenAI Compatible API
In addition, the server supports OpenAI-compatible APIs.

```python
import openai
client = openai.Client(
    base_url="http://127.0.0.1:30000/v1", api_key="EMPTY")

# Text completion
response = client.completions.create(
	model="default",
	prompt="The capital of France is",
	temperature=0,
	max_tokens=32,
)
print(response)

# Chat completion
response = client.chat.completions.create(
    model="default",
    messages=[
        {"role": "system", "content": "You are a helpful AI assistant"},
        {"role": "user", "content": "List 3 countries and their capitals."},
    ],
    temperature=0,
    max_tokens=64,
)
print(response)
```

Ying Sheng's avatar
Ying Sheng committed
134
It supports streaming, vision, and most features of the Chat/Completions/Models endpoints specified by the [OpenAI API Reference](https://platform.openai.com/docs/api-reference/).
Ying Sheng's avatar
Ying Sheng committed
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149

### Additional Server Arguments
- Add `--tp 2` to enable tensor parallelism. If it indicates `peer access is not supported between these two devices`, add `--enable-p2p-check` option.
```
python -m sglang.launch_server --model-path meta-llama/Meta-Llama-3-8B-Instruct --port 30000 --tp 2
```
- Add `--dp 2` to enable data parallelism. It can also be used together with tp. Data parallelism is better for throughput if there is enough memory.
```
python -m sglang.launch_server --model-path meta-llama/Meta-Llama-3-8B-Instruct --port 30000 --dp 2 --tp 2
```
- If you see out-of-memory errors during serving, please try to reduce the memory usage of the KV cache pool by setting a smaller value of `--mem-fraction-static`. The default value is `0.9`
```
python -m sglang.launch_server --model-path meta-llama/Meta-Llama-3-8B-Instruct --port 30000 --mem-fraction-static 0.7
```
- See [hyperparameter_tuning.md](docs/hyperparameter_tuning.md) on tuning hyperparameters for better performance.
150
- Add `--nnodes 2` to run tensor parallelism on multiple nodes. If you have two nodes with two GPUs on each node and want to run TP=4, let `sgl-dev-0` be the hostname of the first node and `50000` be an available port.
Ying Sheng's avatar
Ying Sheng committed
151
152
```
# Node 0
153
python -m sglang.launch_server --model-path meta-llama/Meta-Llama-3-8B-Instruct --tp 4 --nccl-init sgl-dev-0:50000 --nnodes 2 --node-rank 0
Ying Sheng's avatar
Ying Sheng committed
154
155

# Node 1
156
python -m sglang.launch_server --model-path meta-llama/Meta-Llama-3-8B-Instruct --tp 4 --nccl-init sgl-dev-0:50000 --nnodes 2 --node-rank 1
Ying Sheng's avatar
Ying Sheng committed
157
158
```
- If the model does not have a template in the Hugging Face tokenizer, you can specify a [custom chat template](docs/custom_chat_template.md).
159
- To enable fp8 quantization, you can add `--quantization fp8` on a fp16 checkpoint or directly load a fp8 checkpoint without specifying any arguments.
Ying Sheng's avatar
Ying Sheng committed
160
161
162
163
164
165
166
167

### Supported Models

- Llama / Llama 2 / Llama 3
- Mistral / Mixtral
- Gemma / Gemma 2
- Qwen / Qwen 2 / Qwen 2 MoE
- LLaVA 1.5 / 1.6
Ying Sheng's avatar
Ying Sheng committed
168
169
170
  - `python -m sglang.launch_server --model-path liuhaotian/llava-v1.5-7b --tokenizer-path llava-hf/llava-1.5-7b-hf --chat-template vicuna_v1.1 --port 30000`
  - `python -m sglang.launch_server --model-path liuhaotian/llava-v1.6-vicuna-7b --tokenizer-path llava-hf/llava-1.5-7b-hf --chat-template vicuna_v1.1 --port 30000`
  - `python -m sglang.launch_server --model-path liuhaotian/llava-v1.6-34b --tokenizer-path liuhaotian/llava-v1.6-34b-tokenizer --port 30000`
Ying Sheng's avatar
Ying Sheng committed
171
- LLaVA-NeXT-Video
Ying Sheng's avatar
Ying Sheng committed
172
  - see [examples/usage/llava_video](examples/usage/llava_video)
Ying Sheng's avatar
Ying Sheng committed
173
174
175
176
177
178
179
180
181
182
183
- Yi-VL
  - see [srt_example_yi_vl.py](examples/quick_start/srt_example_yi_vl.py).
- StableLM
- Command-R
- DBRX
- Grok
- ChatGLM
- InternLM 2

Instructions for supporting a new model are [here](https://github.com/sgl-project/sglang/blob/main/docs/model_support.md).

Ying Sheng's avatar
Ying Sheng committed
184
185
186
187
188
189
190
191
192
193
194
### Benchmark Performance

- Benchmark a single static batch. Run the following command without launching a server. The arguments are the same as those for `launch_server.py`.
  ```
  python -m sglang.bench_latency --model-path meta-llama/Meta-Llama-3-8B-Instruct --batch 32 --input-len 256 --output-len 32
  ```
- Benchmark online serving. Launch a server first and run the following command.
  ```
  python3 -m sglang.bench_serving --backend sglang --num-prompt 10
  ```

Ying Sheng's avatar
Ying Sheng committed
195
196
197
198
## Frontend: Structured Generation Language (SGLang)
The frontend language can be used with local models or API models.

### Quick Start
Lianmin Zheng's avatar
Lianmin Zheng committed
199
200
The example below shows how to use sglang to answer a mulit-turn question.

Ying Sheng's avatar
Ying Sheng committed
201
#### Using Local Models
202
First, launch a server with
Lianmin Zheng's avatar
Lianmin Zheng committed
203
```
Ying Sheng's avatar
Ying Sheng committed
204
python -m sglang.launch_server --model-path meta-llama/Meta-Llama-3-8B-Instruct --port 30000
Lianmin Zheng's avatar
Lianmin Zheng committed
205
206
```

207
208
Then, connect to the server and answer a multi-turn question.

Lianmin Zheng's avatar
Lianmin Zheng committed
209
```python
210
from sglang import function, system, user, assistant, gen, set_default_backend, RuntimeEndpoint
Lianmin Zheng's avatar
Lianmin Zheng committed
211
212
213
214
215
216
217
218
219

@function
def multi_turn_question(s, question_1, question_2):
    s += system("You are a helpful assistant.")
    s += user(question_1)
    s += assistant(gen("answer_1", max_tokens=256))
    s += user(question_2)
    s += assistant(gen("answer_2", max_tokens=256))

220
set_default_backend(RuntimeEndpoint("http://localhost:30000"))
Lianmin Zheng's avatar
Lianmin Zheng committed
221
222
223
224
225
226
227
228

state = multi_turn_question.run(
    question_1="What is the capital of the United States?",
    question_2="List two local attractions.",
)

for m in state.messages():
    print(m["role"], ":", m["content"])
229
230

print(state["answer_1"])
Lianmin Zheng's avatar
Lianmin Zheng committed
231
232
```

Ying Sheng's avatar
Ying Sheng committed
233
#### Using OpenAI Models
234
Set the OpenAI API Key
Lianmin Zheng's avatar
Lianmin Zheng committed
235
```
236
export OPENAI_API_KEY=sk-******
Lianmin Zheng's avatar
Lianmin Zheng committed
237
238
```

239
Then, answer a multi-turn question.
Lianmin Zheng's avatar
Lianmin Zheng committed
240
```python
241
from sglang import function, system, user, assistant, gen, set_default_backend, OpenAI
Lianmin Zheng's avatar
Lianmin Zheng committed
242
243
244
245
246
247
248
249
250

@function
def multi_turn_question(s, question_1, question_2):
    s += system("You are a helpful assistant.")
    s += user(question_1)
    s += assistant(gen("answer_1", max_tokens=256))
    s += user(question_2)
    s += assistant(gen("answer_2", max_tokens=256))

251
set_default_backend(OpenAI("gpt-3.5-turbo"))
Lianmin Zheng's avatar
Lianmin Zheng committed
252
253
254
255
256
257
258
259

state = multi_turn_question.run(
    question_1="What is the capital of the United States?",
    question_2="List two local attractions.",
)

for m in state.messages():
    print(m["role"], ":", m["content"])
260
261

print(state["answer_1"])
Lianmin Zheng's avatar
Lianmin Zheng committed
262
263
```

Ying Sheng's avatar
Ying Sheng committed
264
#### More Examples
Lianmin Zheng's avatar
Lianmin Zheng committed
265

266
Anthropic and VertexAI (Gemini) models are also supported.
Lianmin Zheng's avatar
Lianmin Zheng committed
267
268
You can find more examples at [examples/quick_start](examples/quick_start).

Ying Sheng's avatar
Ying Sheng committed
269
### Language Feature
Lianmin Zheng's avatar
Lianmin Zheng committed
270
271
272
273
274
To begin with, import sglang.
```python
import sglang as sgl
```

Lianmin Zheng's avatar
Lianmin Zheng committed
275
`sglang` provides some simple primitives such as `gen`, `select`, `fork`, `image`.
Lianmin Zheng's avatar
Lianmin Zheng committed
276
277
You can implement your prompt flow in a function decorated by `sgl.function`.
You can then invoke the function with `run` or `run_batch`.
278
The system will manage the state, chat template, parallelism and batching for you.
Lianmin Zheng's avatar
Lianmin Zheng committed
279

280
281
The complete code for the examples below can be found at [readme_examples.py](examples/usage/readme_examples.py)

Ying Sheng's avatar
Ying Sheng committed
282
#### Control Flow
Lianmin Zheng's avatar
Lianmin Zheng committed
283
284
You can use any Python code within the function body, including control flow, nested function calls, and external libraries.

Lianmin Zheng's avatar
Lianmin Zheng committed
285
286
```python
@sgl.function
287
288
289
def tool_use(s, question):
    s += "To answer this question: " + question + ". "
    s += "I need to use a " + sgl.gen("tool", choices=["calculator", "search engine"]) + ". "
Lianmin Zheng's avatar
Lianmin Zheng committed
290
291
292

    if s["tool"] == "calculator":
        s += "The math expression is" + sgl.gen("expression")
293
294
    elif s["tool"] == "search engine":
        s += "The key word to search is" + sgl.gen("word")
Lianmin Zheng's avatar
Lianmin Zheng committed
295
```
Lianmin Zheng's avatar
Lianmin Zheng committed
296

Ying Sheng's avatar
Ying Sheng committed
297
#### Parallelism
Lianmin Zheng's avatar
Lianmin Zheng committed
298
299
300
Use `fork` to launch parallel prompts.
Because `sgl.gen` is non-blocking, the for loop below issues two generation calls in parallel.

Lianmin Zheng's avatar
Lianmin Zheng committed
301
302
303
304
305
306
307
308
```python
@sgl.function
def tip_suggestion(s):
    s += (
        "Here are two tips for staying healthy: "
        "1. Balanced Diet. 2. Regular Exercise.\n\n"
    )

Lianmin Zheng's avatar
Lianmin Zheng committed
309
    forks = s.fork(2)
Lianmin Zheng's avatar
Lianmin Zheng committed
310
311
312
313
314
315
316
317
    for i, f in enumerate(forks):
        f += f"Now, expand tip {i+1} into a paragraph:\n"
        f += sgl.gen(f"detailed_tip", max_tokens=256, stop="\n\n")

    s += "Tip 1:" + forks[0]["detailed_tip"] + "\n"
    s += "Tip 2:" + forks[1]["detailed_tip"] + "\n"
    s += "In summary" + sgl.gen("summary")
```
Lianmin Zheng's avatar
Lianmin Zheng committed
318

Ying Sheng's avatar
Ying Sheng committed
319
#### Multi Modality
Lianmin Zheng's avatar
Lianmin Zheng committed
320
321
Use `sgl.image` to pass an image as input.

Lianmin Zheng's avatar
Lianmin Zheng committed
322
323
```python
@sgl.function
Lianmin Zheng's avatar
Lianmin Zheng committed
324
def image_qa(s, image_file, question):
Lianmin Zheng's avatar
Lianmin Zheng committed
325
    s += sgl.user(sgl.image(image_file) + question)
Lianmin Zheng's avatar
Lianmin Zheng committed
326
    s += sgl.assistant(sgl.gen("answer", max_tokens=256)
Lianmin Zheng's avatar
Lianmin Zheng committed
327
328
```

329
330
See also [srt_example_llava.py](examples/quick_start/srt_example_llava.py).

Ying Sheng's avatar
Ying Sheng committed
331
#### Constrained Decoding
332
333
Use `regex` to specify a regular expression as a decoding constraint.
This is only supported for local models.
Lianmin Zheng's avatar
Lianmin Zheng committed
334

Lianmin Zheng's avatar
Lianmin Zheng committed
335
```python
Lianmin Zheng's avatar
Lianmin Zheng committed
336
@sgl.function
Lianmin Zheng's avatar
Lianmin Zheng committed
337
338
def regular_expression_gen(s):
    s += "Q: What is the IP address of the Google DNS servers?\n"
Lianmin Zheng's avatar
Lianmin Zheng committed
339
    s += "A: " + sgl.gen(
Lianmin Zheng's avatar
Lianmin Zheng committed
340
341
342
343
344
        "answer",
        temperature=0,
        regex=r"((25[0-5]|2[0-4]\d|[01]?\d\d?).){3}(25[0-5]|2[0-4]\d|[01]?\d\d?)",
    )
```
Lianmin Zheng's avatar
Lianmin Zheng committed
345

Ying Sheng's avatar
Ying Sheng committed
346
#### JSON Decoding
Lianmin Zheng's avatar
Lianmin Zheng committed
347
Use `regex` to specify a JSON schema with a regular expression.
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368

```python
character_regex = (
    r"""\{\n"""
    + r"""    "name": "[\w\d\s]{1,16}",\n"""
    + r"""    "house": "(Gryffindor|Slytherin|Ravenclaw|Hufflepuff)",\n"""
    + r"""    "blood status": "(Pure-blood|Half-blood|Muggle-born)",\n"""
    + r"""    "occupation": "(student|teacher|auror|ministry of magic|death eater|order of the phoenix)",\n"""
    + r"""    "wand": \{\n"""
    + r"""        "wood": "[\w\d\s]{1,16}",\n"""
    + r"""        "core": "[\w\d\s]{1,16}",\n"""
    + r"""        "length": [0-9]{1,2}\.[0-9]{0,2}\n"""
    + r"""    \},\n"""
    + r"""    "alive": "(Alive|Deceased)",\n"""
    + r"""    "patronus": "[\w\d\s]{1,16}",\n"""
    + r"""    "bogart": "[\w\d\s]{1,16}"\n"""
    + r"""\}"""
)

@sgl.function
def character_gen(s, name):
Lianmin Zheng's avatar
Lianmin Zheng committed
369
    s += name + " is a character in Harry Potter. Please fill in the following information about this character.\n"
370
371
372
    s += sgl.gen("json_output", max_tokens=256, regex=character_regex)
```

Lianmin Zheng's avatar
Lianmin Zheng committed
373
See also [json_decode.py](examples/usage/json_decode.py) for an additional example on specifying formats with Pydantic models.
374

Ying Sheng's avatar
Ying Sheng committed
375
#### Batching
Lianmin Zheng's avatar
Lianmin Zheng committed
376
377
Use `run_batch` to run a batch of requests with continuous batching.

Lianmin Zheng's avatar
Lianmin Zheng committed
378
379
380
381
382
383
384
385
386
387
388
389
```python
@sgl.function
def text_qa(s, question):
    s += "Q: " + question + "\n"
    s += "A:" + sgl.gen("answer", stop="\n")

states = text_qa.run_batch(
    [
        {"question": "What is the capital of the United Kingdom?"},
        {"question": "What is the capital of France?"},
        {"question": "What is the capital of Japan?"},
    ],
Lianmin Zheng's avatar
Lianmin Zheng committed
390
    progress_bar=True
Lianmin Zheng's avatar
Lianmin Zheng committed
391
392
)
```
Lianmin Zheng's avatar
Lianmin Zheng committed
393

Ying Sheng's avatar
Ying Sheng committed
394
#### Streaming
Lianmin Zheng's avatar
Lianmin Zheng committed
395
396
Add `stream=True` to enable streaming.

Lianmin Zheng's avatar
Lianmin Zheng committed
397
398
399
400
401
402
```python
@sgl.function
def text_qa(s, question):
    s += "Q: " + question + "\n"
    s += "A:" + sgl.gen("answer", stop="\n")

403
state = text_qa.run(
Lianmin Zheng's avatar
Lianmin Zheng committed
404
    question="What is the capital of France?",
Lianmin Zheng's avatar
Lianmin Zheng committed
405
406
407
    temperature=0.1,
    stream=True
)
Lianmin Zheng's avatar
Lianmin Zheng committed
408

Lianmin Zheng's avatar
Lianmin Zheng committed
409
410
411
for out in state.text_iter():
    print(out, end="", flush=True)
```
Lianmin Zheng's avatar
Lianmin Zheng committed
412

Ying Sheng's avatar
Ying Sheng committed
413
#### Tips and Implementation Details
414
415
- The `choices` argument in `sgl.gen` is implemented by computing the [token-length normalized log probabilities](https://blog.eleuther.ai/multiple-choice-normalization/) of all choices and selecting the one with the highest probability.
- The `regex` argument in `sgl.gen` is implemented through autoregressive decoding with logit bias masking, according to the constraints set by the regex. It is compatible with `temperature=0` and `temperature != 0`.
Lianmin Zheng's avatar
Lianmin Zheng committed
416

Lianmin Zheng's avatar
Lianmin Zheng committed
417
## Benchmark And Performance
Lianmin Zheng's avatar
Lianmin Zheng committed
418
419
420
421
422
423
- Llama-7B on NVIDIA A10G, FP16, Tensor Parallelism=1
![llama_7b](assets/llama_7b.jpg)

- Mixtral-8x7B on NVIDIA A10G, FP16, Tensor Parallelism=8
![mixtral_8x7b](assets/mixtral_8x7b.jpg)

Lianmin Zheng's avatar
Lianmin Zheng committed
424
425
- Learn more about the above [results](docs/benchmark_results.md).
- Synthetic latency and throughput benchmark [scripts](https://github.com/sgl-project/sglang/tree/main/benchmark/latency_throughput).
Lianmin Zheng's avatar
Lianmin Zheng committed
426

Lianmin Zheng's avatar
Lianmin Zheng committed
427
## Roadmap
Ying Sheng's avatar
Ying Sheng committed
428
[Development Roadmap (2024 Q3)](https://github.com/sgl-project/sglang/issues/634)
Lianmin Zheng's avatar
Lianmin Zheng committed
429
430

## Citation And Acknowledgment
Ying Sheng's avatar
Ying Sheng committed
431
432
Please cite our paper, [SGLang: Efficient Execution of Structured Language Model Programs](https://arxiv.org/abs/2312.07104), if you find the project useful.
We also learned from the design and reused code from the following projects: [Guidance](https://github.com/guidance-ai/guidance), [vLLM](https://github.com/vllm-project/vllm), [LightLLM](https://github.com/ModelTC/lightllm), [FlashInfer](https://github.com/flashinfer-ai/flashinfer), [Outlines](https://github.com/outlines-dev/outlines), and [LMQL](https://github.com/eth-sri/lmql).