Skip to content

Commit c22bb86

Browse files
cristipufucristipufu
committed
docs: update llm and ecs
1 parent 4c913d1 commit c22bb86

2 files changed

Lines changed: 162 additions & 164 deletions

File tree

‎docs/context_grounding.md‎

Lines changed: 3 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -69,5 +69,6 @@ response = react_agent.chat("Answer user questions as best as you can using the
6969
```
7070

7171

72-
73-
> **HINT:** Check our [travel-helper-RAG-agent sample](https://github.com/UiPath/uipath-llamaindex-python/tree/main/samples/travel-helper-RAG-agent) to see context grounding query engines in action.
72+
/// tip
73+
Check our [travel-helper-RAG-agent sample](https://github.com/UiPath/uipath-llamaindex-python/tree/main/samples/travel-helper-RAG-agent) to see context grounding query engines in action.
74+
///

‎docs/llms_and_embeddings.md‎

Lines changed: 159 additions & 162 deletions
Original file line numberDiff line numberDiff line change
@@ -1,162 +1,159 @@
1-
# UiPath LLMs and Embeddings
2-
3-
This guide covers the UiPath-integrated Large Language Models (LLMs) and embedding models available in the UiPath LlamaIndex SDK.
4-
5-
## Overview
6-
7-
The UiPath LlamaIndex SDK provides pre-configured LLM and embedding classes that integrate seamlessly with UiPath. These classes handle authentication, routing, and configuration automatically, allowing you to focus on building your agents.
8-
9-
## Prerequisites
10-
11-
Before using these classes, ensure you have:
12-
13-
- Authenticated with UiPath using `uipath auth`
14-
- Set up your environment variables (automatically configured after authentication)
15-
16-
## UiPathOpenAI
17-
18-
The `UiPathOpenAI` class is a pre-configured Azure OpenAI client that routes requests through UiPath.
19-
20-
### Available Models
21-
22-
The following OpenAI models are available through the `OpenAIModel` enum:
23-
24-
- `GPT_4_1_2025_04_14`
25-
- `GPT_4_1_MINI_2025_04_14`
26-
- `GPT_4_1_NANO_2025_04_14`
27-
- `GPT_4O_2024_05_13`
28-
- `GPT_4O_2024_08_06`
29-
- `GPT_4O_2024_11_20`
30-
- `GPT_4O_MINI_2024_07_18` (default)
31-
- `O3_MINI_2025_01_31`
32-
- `TEXT_DAVINCI_003`
33-
34-
### Basic Usage
35-
36-
```python
37-
from uipath_llamaindex.llms import UiPathOpenAI
38-
from llama_index.core.llms import ChatMessage
39-
40-
# Create an LLM instance with default settings
41-
llm = UiPathOpenAI()
42-
43-
# Create chat messages
44-
messages = [
45-
ChatMessage(
46-
role="system", content="You are a pirate with colorful personality."
47-
),
48-
ChatMessage(role="user", content="Hello"),
49-
]
50-
51-
# Generate a response
52-
response = llm.chat(messages)
53-
print(response)
54-
```
55-
56-
### Custom Model Configuration
57-
58-
```python
59-
from uipath_llamaindex.llms import UiPathOpenAI, OpenAIModel
60-
61-
# Use a specific model
62-
llm = UiPathOpenAI(model=OpenAIModel.GPT_4O_2024_11_20)
63-
64-
# Or use a model string directly
65-
llm = UiPathOpenAI(model="gpt-4o-2024-11-20")
66-
```
67-
68-
## UiPathOpenAIEmbedding
69-
70-
The `UiPathOpenAIEmbedding` class provides text embedding capabilities using OpenAI's embedding models through UiPath.
71-
72-
### Available Embedding Models
73-
74-
The following embedding models are available through the `OpenAIEmbeddingModel` enum:
75-
76-
- `TEXT_EMBEDDING_ADA_002` (default)
77-
- `TEXT_EMBEDDING_3_LARGE`
78-
79-
### Basic Usage
80-
81-
```python
82-
from uipath_llamaindex.embeddings import UiPathOpenAIEmbedding
83-
84-
# Create an embedding model instance
85-
embed_model = UiPathOpenAIEmbedding()
86-
87-
# Get embeddings for a single text
88-
result = embed_model.get_text_embedding("the quick brown fox jumps over the lazy dog")
89-
print(f"Embedding dimension: {len(result)}")
90-
```
91-
92-
### Batch Embeddings
93-
94-
```python
95-
from uipath_llamaindex.embeddings import UiPathOpenAIEmbedding
96-
97-
embed_model = UiPathOpenAIEmbedding()
98-
99-
# Get embeddings for multiple texts
100-
texts = [
101-
"Hello world",
102-
"How are you?",
103-
"This is a test"
104-
]
105-
106-
embeddings = embed_model.get_text_embedding_batch(texts)
107-
print(f"Number of embeddings: {len(embeddings)}")
108-
```
109-
110-
111-
## Integration with LlamaIndex
112-
113-
Both classes integrate seamlessly with LlamaIndex components:
114-
115-
### Using with Agents
116-
117-
```python
118-
from llama_index.core.agent import ReActAgent
119-
from llama_index.core.tools import FunctionTool
120-
from uipath_llamaindex.llms import UiPathOpenAI
121-
122-
def multiply(a: int, b: int) -> int:
123-
"""Multiply two integers and returns the result."""
124-
return a * b
125-
126-
multiply_tool = FunctionTool.from_defaults(fn=multiply)
127-
128-
# Create agent with UiPath LLM
129-
agent = ReActAgent.from_tools(
130-
[multiply_tool],
131-
llm=UiPathOpenAI(model=OpenAIModel.GPT_4O_2024_11_20)
132-
)
133-
134-
response = agent.chat("What is 21 multiplied by 2?")
135-
```
136-
137-
### Using with VectorStoreIndex
138-
139-
```python
140-
from llama_index.core import VectorStoreIndex, Document
141-
from uipath_llamaindex.llms import UiPathOpenAI
142-
from uipath_llamaindex.embeddings import UiPathOpenAIEmbedding
143-
144-
# Create documents
145-
documents = [
146-
Document(text="This is a sample document about artificial intelligence."),
147-
Document(text="Machine learning is a subset of AI that focuses on algorithms."),
148-
]
149-
150-
# Create index with UiPath models
151-
index = VectorStoreIndex.from_documents(
152-
documents,
153-
embed_model=UiPathOpenAIEmbedding()
154-
)
155-
156-
# Create query engine with UiPath LLM
157-
query_engine = index.as_query_engine(
158-
llm=UiPathOpenAI(model=OpenAIModel.GPT_4O_2024_11_20)
159-
)
160-
161-
response = query_engine.query("What is machine learning?")
162-
```
1+
# LLMs and Embeddings
2+
3+
UiPath provides pre-configured LLM and embedding classes that handle authentication, routing, and configuration automatically, allowing you to focus on building your agents.
4+
You do not need to add tokens from OpenAI, usage of these models will consume `Agent Units` on your account.
5+
6+
## UiPathOpenAI
7+
8+
The `UiPathOpenAI` class is a pre-configured Azure OpenAI client that routes requests through UiPath.
9+
10+
### Available Models
11+
12+
The following OpenAI models are available through the `OpenAIModel` enum:
13+
14+
- `GPT_4_1_2025_04_14`
15+
- `GPT_4_1_MINI_2025_04_14`
16+
- `GPT_4_1_NANO_2025_04_14`
17+
- `GPT_4O_2024_05_13`
18+
- `GPT_4O_2024_08_06`
19+
- `GPT_4O_2024_11_20`
20+
- `GPT_4O_MINI_2024_07_18` (default)
21+
- `O3_MINI_2025_01_31`
22+
- `TEXT_DAVINCI_003`
23+
24+
### Basic Usage
25+
26+
```python
27+
from uipath_llamaindex.llms import UiPathOpenAI
28+
from llama_index.core.llms import ChatMessage
29+
30+
# Create an LLM instance with default settings
31+
llm = UiPathOpenAI()
32+
33+
# Create chat messages
34+
messages = [
35+
ChatMessage(
36+
role="system", content="You are a pirate with colorful personality."
37+
),
38+
ChatMessage(role="user", content="Hello"),
39+
]
40+
41+
# Generate a response
42+
response = llm.chat(messages)
43+
print(response)
44+
```
45+
46+
### Custom Model Configuration
47+
48+
```python
49+
from uipath_llamaindex.llms import UiPathOpenAI, OpenAIModel
50+
51+
# Use a specific model
52+
llm = UiPathOpenAI(model=OpenAIModel.GPT_4O_2024_11_20)
53+
54+
# Or use a model string directly
55+
llm = UiPathOpenAI(model="gpt-4o-2024-11-20")
56+
```
57+
58+
## UiPathOpenAIEmbedding
59+
60+
The `UiPathOpenAIEmbedding` class provides text embedding capabilities using OpenAI's embedding models through UiPath.
61+
62+
### Available Embedding Models
63+
64+
The following embedding models are available through the `OpenAIEmbeddingModel` enum:
65+
66+
- `TEXT_EMBEDDING_ADA_002` (default)
67+
- `TEXT_EMBEDDING_3_LARGE`
68+
69+
### Basic Usage
70+
71+
```python
72+
from uipath_llamaindex.embeddings import UiPathOpenAIEmbedding
73+
74+
# Create an embedding model instance
75+
embed_model = UiPathOpenAIEmbedding()
76+
77+
# Get embeddings for a single text
78+
result = embed_model.get_text_embedding("the quick brown fox jumps over the lazy dog")
79+
print(f"Embedding dimension: {len(result)}")
80+
```
81+
82+
### Batch Embeddings
83+
84+
```python
85+
from uipath_llamaindex.embeddings import UiPathOpenAIEmbedding
86+
87+
embed_model = UiPathOpenAIEmbedding()
88+
89+
# Get embeddings for multiple texts
90+
texts = [
91+
"Hello world",
92+
"How are you?",
93+
"This is a test"
94+
]
95+
96+
embeddings = embed_model.get_text_embedding_batch(texts)
97+
print(f"Number of embeddings: {len(embeddings)}")
98+
```
99+
100+
101+
## Integration with LlamaIndex
102+
103+
Both classes integrate seamlessly with LlamaIndex components:
104+
105+
### Using with Agents
106+
107+
```python
108+
from llama_index.core.agent import ReActAgent
109+
from llama_index.core.tools import FunctionTool
110+
from uipath_llamaindex.llms import UiPathOpenAI
111+
112+
def multiply(a: int, b: int) -> int:
113+
"""Multiply two integers and returns the result."""
114+
return a * b
115+
116+
multiply_tool = FunctionTool.from_defaults(fn=multiply)
117+
118+
# Create agent with UiPath LLM
119+
agent = ReActAgent.from_tools(
120+
[multiply_tool],
121+
llm=UiPathOpenAI(model=OpenAIModel.GPT_4O_2024_11_20)
122+
)
123+
124+
response = agent.chat("What is 21 multiplied by 2?")
125+
```
126+
127+
### Using with VectorStoreIndex
128+
129+
```python
130+
from llama_index.core import VectorStoreIndex, Document
131+
from uipath_llamaindex.llms import UiPathOpenAI
132+
from uipath_llamaindex.embeddings import UiPathOpenAIEmbedding
133+
134+
# Create documents
135+
documents = [
136+
Document(text="This is a sample document about artificial intelligence."),
137+
Document(text="Machine learning is a subset of AI that focuses on algorithms."),
138+
]
139+
140+
# Create index with UiPath models
141+
index = VectorStoreIndex.from_documents(
142+
documents,
143+
embed_model=UiPathOpenAIEmbedding()
144+
)
145+
146+
# Create query engine with UiPath LLM
147+
query_engine = index.as_query_engine(
148+
llm=UiPathOpenAI(model=OpenAIModel.GPT_4O_2024_11_20)
149+
)
150+
151+
response = query_engine.query("What is machine learning?")
152+
```
153+
154+
/// warning
155+
Please note that that you may get errors related to data residency, as some models are not available on all regions.
156+
157+
Example: `[Enforced Region] No model configuration found for product uipath-python-sdk in EU`.
158+
159+
///

0 commit comments

Comments
 (0)