Triton Inference Server

LiteLLM supports Embedding Models on Triton Inference Servers

Property	Details
Description	NVIDIA Triton Inference Server
Provider Route on LiteLLM	triton/
Supported Operations	/chat/completion, /completion, /embedding
Supported Triton endpoints	/infer, /generate, /embeddings
Link to Provider Doc	Triton Inference Server ↗

Triton /generate - Chat Completion

SDK

Use the triton/ prefix to route to triton server

from litellm import completion
response = completion(
    model="triton/llama-3-8b-instruct",
    messages=[{"role": "user", "content": "who are u?"}],
    max_tokens=10,
    api_base="http://localhost:8000/generate",
)

PROXY

1. Add models to your config.yaml

model_list:
  - model_name: my-triton-model
    litellm_params:
      model: triton/<your-triton-model>"
      api_base: https://your-triton-api-base/triton/generate

2. Start the proxy

$ litellm --config /path/to/config.yaml --detailed_debug

3. Send Request to LiteLLM Proxy Server

OpenAI Python v1.0.0+

import openai
from openai import OpenAI

# set base_url to your proxy server
# set api_key to send to proxy server
client = OpenAI(api_key="<proxy-api-key>", base_url="http://0.0.0.0:4000")

response = client.chat.completions.create(
    model="my-triton-model",
    messages=[{"role": "user", "content": "who are u?"}],
    max_tokens=10,
)

print(response)

curl

--header is optional, only required if you're using litellm proxy with Virtual Keys

curl --location 'http://0.0.0.0:4000/chat/completions' \
--header 'Content-Type: application/json' \
--header 'Authorization: Bearer sk-1234' \
--data ' {
"model": "my-triton-model",
"messages": [{"role": "user", "content": "who are u?"}]
}'

Triton /infer - Chat Completion

SDK

Use the triton/ prefix to route to triton server

from litellm import completion


response = completion(
    model="triton/llama-3-8b-instruct",
    messages=[{"role": "user", "content": "who are u?"}],
    max_tokens=10,
    api_base="http://localhost:8000/infer",
)

PROXY

1. Add models to your config.yaml

model_list:
  - model_name: my-triton-model
    litellm_params:
      model: triton/<your-triton-model>"
      api_base: https://your-triton-api-base/triton/infer

2. Start the proxy

$ litellm --config /path/to/config.yaml --detailed_debug

3. Send Request to LiteLLM Proxy Server

OpenAI Python v1.0.0+

import openai
from openai import OpenAI

# set base_url to your proxy server
# set api_key to send to proxy server
client = OpenAI(api_key="<proxy-api-key>", base_url="http://0.0.0.0:4000")

response = client.chat.completions.create(
    model="my-triton-model",
    messages=[{"role": "user", "content": "who are u?"}],
    max_tokens=10,
)

print(response)

curl

--header is optional, only required if you're using litellm proxy with Virtual Keys

curl --location 'http://0.0.0.0:4000/chat/completions' \
--header 'Content-Type: application/json' \
--header 'Authorization: Bearer sk-1234' \
--data ' {
"model": "my-triton-model",
"messages": [{"role": "user", "content": "who are u?"}]
}'

Triton /embeddings - Embedding

SDK

Use the triton/ prefix to route to triton server

from litellm import embedding
import os

response = await litellm.aembedding(
    model="triton/<your-triton-model>",                                                       
    api_base="https://your-triton-api-base/triton/embeddings", # /embeddings endpoint you want litellm to call on your server
    input=["good morning from litellm"],
)

PROXY

1. Add models to your config.yaml

model_list:
  - model_name: my-triton-model
    litellm_params:
      model: triton/<your-triton-model>"
      api_base: https://your-triton-api-base/triton/embeddings

2. Start the proxy

$ litellm --config /path/to/config.yaml --detailed_debug

3. Send Request to LiteLLM Proxy Server

OpenAI Python v1.0.0+

import openai
from openai import OpenAI

# set base_url to your proxy server
# set api_key to send to proxy server
client = OpenAI(api_key="<proxy-api-key>", base_url="http://0.0.0.0:4000")

response = client.embeddings.create(
    input=["hello from litellm"],
    model="my-triton-model"
)

print(response)

curl

--header is optional, only required if you're using litellm proxy with Virtual Keys

curl --location 'http://0.0.0.0:4000/embeddings' \
--header 'Content-Type: application/json' \
--header 'Authorization: Bearer sk-1234' \
--data ' {
"model": "my-triton-model",
"input": ["write a litellm poem"]
}'