Datasets:

Modalities:
Text
ArXiv:
File size: 3,877 Bytes
f9dff7a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
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
134
135
136
137
138
139
140
141
142
143
144
145
# Serve and Deploy LLMs

This document shows how you can serve a LitGPT for deployment.


 
## Serve an LLM with LitServe

This section illustrates how we can set up an inference server for a phi-2 LLM using `litgpt serve` that is minimal and highly scalable.


 
### Step 1: Start the inference server


```bash
# 1) Download a pretrained model (alternatively, use your own finetuned model)
litgpt download microsoft/phi-2

# 2) Start the server
litgpt serve microsoft/phi-2
```

> [!TIP]
> Use `litgpt serve --help` to display additional options, including the port, devices, LLM temperature setting, and more.


 
### Step 2: Query the inference server

You can now send requests to the inference server you started in step 2. For example, in a new Python session, we can send requests to the inference server as follows:


```python
import requests, json

response = requests.post(
    "http://127.0.0.1:8000/predict",
    json={"prompt": "Fix typos in the following sentence: Example input"}
)

print(response.json()["output"])
```

Executing the code above prints the following output:

```
Example input.
```

 
### Optional: Use the streaming mode

The 2-step procedure described above returns the complete response all at once. If you want to stream the response on a token-by-token basis, start the server with the streaming option enabled:

```bash
litgpt serve microsoft/phi-2 --stream true
```

Then, use the following updated code to query the inference server:

```python
import requests, json

response = requests.post(
    "http://127.0.0.1:8000/predict",
    json={"prompt": "Fix typos in the following sentence: Example input"},
    stream=True
)

# stream the response
for line in response.iter_lines(decode_unicode=True):
    if line:
        print(json.loads(line)["output"], end="")
```

```
Sure, here is the corrected sentence:

Example input
```

 
## Serve an LLM with OpenAI-compatible API

LitGPT provides OpenAI-compatible endpoints that allow you to use the OpenAI SDK or any OpenAI-compatible client to interact with your models. This is useful for integrating LitGPT into existing applications that use the OpenAI API.

 
### Step 1: Start the server with OpenAI specification

```bash
# 1) Download a pretrained model (alternatively, use your own finetuned model)
litgpt download HuggingFaceTB/SmolLM2-135M-Instruct

# 2) Start the server with OpenAI-compatible endpoints
litgpt serve HuggingFaceTB/SmolLM2-135M-Instruct --openai_spec true
```

> [!TIP]
> The `--openai_spec true` flag enables OpenAI-compatible endpoints at `/v1/chat/completions` instead of the default `/predict` endpoint.

 
### Step 2: Query using OpenAI-compatible endpoints

You can now send requests to the OpenAI-compatible endpoint using curl:

```bash
curl -X POST http://127.0.0.1:8000/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "SmolLM2-135M-Instruct",
    "messages": [{"role": "user", "content": "Hello! How are you?"}]
  }'
```

Or use the OpenAI Python SDK:

```python
from openai import OpenAI

# Configure the client to use your local LitGPT server
client = OpenAI(
    base_url="http://127.0.0.1:8000/v1",
    api_key="not-needed"  # LitGPT doesn't require authentication by default
)

response = client.chat.completions.create(
    model="SmolLM2-135M-Instruct",
    messages=[
        {"role": "user", "content": "Hello! How are you?"}
    ]
)

print(response.choices[0].message.content)
```

 
## Serve an LLM UI with Chainlit

If you are interested in developing a simple ChatGPT-like UI prototype, see the Chainlit tutorial in the following Studio:

<a target="_blank" href="https://lightning.ai/lightning-ai/studios/chatgpt-like-llm-uis-via-chainlit">
  <img src="https://pl-bolts-doc-images.s3.us-east-2.amazonaws.com/app-2/studio-badge.svg" alt="Open In Studio"/>
</a>