klas/t6_mem0
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

User/dyadav/add platform docs (#1491)

Browse Source
This commit is contained in:
Deshraj Yadav
2024-07-17 17:39:57 -07:00
committed by GitHub
parent 1e7618dfa4
commit c9240e7ca6
10 changed files with 669 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

BIN
docs/images/platform/activity.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 180 KiB

BIN
docs/images/platform/api-key.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 169 KiB

11
docs/mint.json
View File

@@ -20,6 +20,10 @@
{
"name": "💡 Examples",
"url": "examples"
},
{
"name": "🖥️ Platform",
"url": "platform"
}
],
"topbarLinks": [
@@ -73,6 +77,13 @@
"examples/customer-support-agent",
"examples/personal-travel-assistant"
]
},
{
"group": "🖥️ Platform",
"pages": [
"platform/overview",
"platform/quickstart"
]
}
],
"footerSocials": {

45
docs/platform/overview.mdx Normal file
View File

@@ -0,0 +1,45 @@
---
title: Introduction
description: 'Empower your AI applications with long-term memory and personalization'
---
## Welcome to Mem0 Platform
Mem0 Platform is a managed service that revolutionizes the way AI applications handle memory. By providing a smart, self-improving memory layer for Large Language Models (LLMs), we enable developers to create personalized AI experiences that evolve with each user interaction.
## Why Choose Mem0 Platform?
1. **Enhanced User Experience**: Deliver tailored interactions that make your AI applications truly stand out.
2. **Simplified Development**: Our API-first approach streamlines integration, allowing you to focus on building great features.
3. **Scalable Solution**: Designed to grow with your application, from prototypes to production-ready systems.
## Key Features
- **Comprehensive Memory Management**: Easily manage long-term, short-term, semantic, and episodic memories for individual users, agents, and sessions through our robust APIs.
- **Self-Improving Memory**: Our adaptive system continuously learns from user interactions, refining its understanding over time.
- **Cross-Platform Consistency**: Ensure a unified user experience across various AI platforms and applications.
- **Centralized Memory Control**: Store, update, and delete memories effortlessly, taking away the hassle of memory management.
## Common Use Cases
- Personalized Learning Assistants
- Customer Support AI Agents
- Healthcare Assistants
- Virtual Companions
- Productivity Tools
- Gaming AI
## Getting Started
Ready to supercharge your AI application with Mem0? Follow these steps:
1. **Sign Up**: Create your Mem0 account at our platform.
2. **API Key**: Generate your API key in the dashboard.
3. **Installation**: Install our Python SDK using pip: `pip install mem0ai`
4. **Quick Implementation**: Check out our [Quickstart Guide](/platform/quickstart) to start using Mem0 quickly.
## Next Steps
- Explore our API Reference for detailed endpoint documentation.
- Join our [slack](https://mem0.ai/slack) or [discord](https://mem0.ai/discord) with other developers and get support.
We're excited to see what you'll build with Mem0 Platform. Let's create smarter, more personalized AI experiences together!

358
docs/platform/quickstart.mdx Normal file
View File

@@ -0,0 +1,358 @@
---
title: Quickstart
description: 'Get started with Mem0 Platform in minutes'
---
## 1. Installation
Install the Mem0 Python package:
```bash
pip install mem0ai
```
## 2. API Key Setup
1. Sign in to [Mem0 Platform](https://app.mem0.ai/dashboard/api-keys)
2. Copy your API Key from the dashboard
![Get API Key from Mem0 Platform](/images/platform/api-key.png)
## 3. Instantiate Client
```python
from mem0 import MemoryClient
client = MemoryClient(api_key="your-api-key")
```
## 4. Memory Operations
We provide a simple yet customizable interface for performing CRUD operations on memory. Here is how you can create and get memories:
### 4.1 Create Memories
For users (long-term memory):
```python
# create long-term memory for users
client.add("Remember my name is Deshraj Yadav.", user_id="deshraj")
client.add("I like to eat pizza and go out on weekends.", user_id="deshraj")
client.add("Oh I am actually allergic to cheese to cannot eat pizza anymore.", user_id="deshraj")
```
Output:
```python
{'message': 'Memory added successfully!'}
```
You can see all the memory operations happening on the platform itself.
![Mem0 Platform Activity](/images/platform/activity.png)
You can also add memories for a particular session or for an AI agent that you are building:
- For user sessions (short-term memory):
```python
client.add("Deshraj is building Gmail AI agent", user_id="deshraj", session_id="session-1")
```
- For agents (long-term memory):
```python
client.add("Return short responses when responding to emails", agent_id="gmail-agent")
```
### 4.2 Retrieve Memories
<CodeGroup>
```python Code
client.get_all(user_id="deshraj")
```
```python Output
[
{
'id': 'dbce6e06-6adf-40b8-9187-3d30bd13b741',
'agent': None,
'consumer': {
'id': 8,
'user_id': 'deshraj',
'metadata': None,
'created_at': '2024-07-17T16:47:23.899900-07:00',
'updated_at': '2024-07-17T16:47:23.899918-07:00'
},
'app': None,
'run': None,
'hash': '57288ac8a87c4ac8d3ac7f2075d264ca',
'input': 'Remember my name is Deshraj Yadav.',
'text': 'My name is Deshraj Yadav.',
'metadata': None,
'created_at': '2024-07-17T16:47:25.670180-07:00',
'updated_at': '2024-07-17T16:47:25.670197-07:00'
},
{
'id': 'f6dec5d1-b5db-45f5-a2fb-3979a0f27d30',
'agent': None,
'consumer': {
'id': 8,
'user_id': 'deshraj',
# ... other consumer fields ...
},
# ... other fields ...
'text': 'I am allergic to cheese so I cannot eat pizza anymore.',
# ... remaining fields ...
},
# ... additional memory entries ...
]
```
</CodeGroup>
Similarly, you can get all memories for an agent:
```python
agent_memories = client.get_all(agent_id="gmail-agent")
```
Get specific memory:
<CodeGroup>
```python Code
memory = client.get(memory_id="dbce6e06-6adf-40b8-9187-3d30bd13b741")
```
```python Output
{
'id': 'dbce6e06-6adf-40b8-9187-3d30bd13b741',
'agent': None,
'consumer': {
'id': 8,
'user_id': 'deshraj',
'metadata': None,
'created_at': '2024-07-17T16:47:23.899900-07:00',
'updated_at': '2024-07-17T16:47:23.899918-07:00'
},
'app': None,
'run': None,
'hash': '57288ac8a87c4ac8d3ac7f2075d264ca',
'input': 'Remember my name is Deshraj Yadav.',
'text': 'My name is Deshraj Yadav.',
'metadata': None,
'created_at': '2024-07-17T16:47:25.670180-07:00',
'updated_at': '2024-07-17T16:47:25.670197-07:00'
}
```
</CodeGroup>
### 4.3 Update Memory
You can also update specific memory by using the following method:
<CodeGroup>
```python Code
client.update(memory_id, data="Updated name is Deshraj Kumar")
```
```python Output
{
'id': 'dbce6e06-6adf-40b8-9187-3d30bd13b741',
'agent': None,
'consumer': {
'id': 8,
'user_id': 'deshraj',
'metadata': None,
'created_at': '2024-07-17T16:47:23.899900-07:00',
'updated_at': '2024-07-17T16:47:23.899918-07:00'
},
'app': None,
'run': None,
'hash': '57288ac8a87c4ac8d3ac7f2075d264ca',
'input': 'Updated name is Deshraj Kumar.',
'text': 'Name is Deshraj Kumar.',
'metadata': None,
'created_at': '2024-07-17T16:47:25.670180-07:00',
'updated_at': '2024-07-17T16:47:25.670197-07:00'
}
```
</CodeGroup>
### 4.4 Memory History
Get history of how a memory has changed over time
<CodeGroup>
```python Code
history = client.history(memory_id)
```
```python Output
[
{
'id': '51193804-2ee6-4f81-b4e7-497e98b70858',
'memory': {
'id': 'dbce6e06-6adf-40b8-9187-3d30bd13b741',
'agent': None,
'consumer': {
'id': 8,
'user_id': 'deshraj',
'metadata': None,
'created_at': '2024-07-17T16:47:23.899900-07:00',
'updated_at': '2024-07-17T16:47:23.899918-07:00'
},
'app': None,
'run': None,
'hash': '57288ac8a87c4ac8d3ac7f2075d264ca',
'input': 'Remember my name is Deshraj Yadav.',
'text': 'My name is Deshraj Yadav.',
'metadata': None,
'created_at': '2024-07-17T16:47:25.670180-07:00',
'updated_at': '2024-07-17T16:47:25.670197-07:00'
},
'hash': '57288ac8a87c4ac8d3ac7f2075d264ca',
'event': 'ADD',
'input': 'Remember my name is Deshraj Yadav.',
'previous_text': None,
'text': 'My name is Deshraj Yadav.',
'metadata': None,
'created_at': '2024-07-17T16:47:25.686899-07:00',
'updated_at': '2024-07-17T16:47:25.670197-07:00',
'change_description': 'Memory ADD event'
}
]
```
</CodeGroup>
### 4.5 Search for relevant memories
<CodeGroup>
```python Code
client.search("What does Deshraj like to eat?", user_id="deshraj", limit=3)
```
```python Output
[
{
"id": "dbce6e06-6adf-40b8-9187-3d30bd13b741",
"agent": null,
"consumer": {
"id": 8,
"user_id": "deshraj",
"metadata": null,
"created_at": "...",
"updated_at": "..."
},
"app": null,
"run": null,
"hash": "57288ac8a87c4ac8d3ac7f2075d264ca",
"input": "Remember my name is Deshraj Yadav.",
"text": "My name is Deshraj Yadav.",
"metadata": null,
"created_at": "2024-07-17T16:47:25.670180-07:00",
"updated_at": "..."
},
{
"id": "091dbed6-74b4-4e15-b765-81be2abe0d6b",
"agent": null,
"consumer": {
"id": 8,
"user_id": "deshraj",
"metadata": null,
"created_at": "...",
"updated_at": "..."
},
"app": null,
"run": null,
"hash": "622a5a24d5ac54136414a22ec12f9520",
"input": "Oh I am actually allergic to cheese to cannot eat pizza anymore.",
"text": "I like to eat pizza and go out on weekends.",
"metadata": null,
"created_at": "2024-07-17T16:49:24.276695-07:00",
"updated_at": "..."
},
{
"id": "5fb8f85d-3383-4bad-9d46-f171272478a4",
"agent": null,
"consumer": {
"id": 8,
"user_id": "deshraj",
"metadata": null,
"created_at": "...",
"updated_at": "..."
},
"app": null,
"run": {
"id": 1,
"run_id": "session-1",
"name": "",
"metadata": null,
"created_at": "...",
"updated_at": "..."
},
"hash": "179ced9649ac2b85350ece4946b1ee9b",
"input": "Deshraj is building Gmail AI agent",
"text": "Deshraj is building Gmail AI agent",
"metadata": null,
"created_at": "2024-07-17T16:52:41.278920-07:00",
"updated_at": "..."
},
{
"id": "f6dec5d1-b5db-45f5-a2fb-3979a0f27d30",
"agent": null,
"consumer": {
"id": 8,
"user_id": "deshraj",
"metadata": null,
"created_at": "...",
"updated_at": "..."
},
"app": null,
"run": null,
"hash": "19248f0766044b5973fc0ef1bf3955ef",
"input": "Oh I am actually allergic to cheese to cannot eat pizza anymore.",
"text": "I am allergic to cheese so I cannot eat pizza anymore.",
"metadata": null,
"created_at": "2024-07-17T16:49:38.622084-07:00",
"updated_at": "..."
}
]
```
</CodeGroup>
### 4.6 Delete Memory
Delete specific memory:
<CodeGroup>
```python Code
client.delete(memory_id)
```
```python Output
{'message': 'Memory deleted successfully!'}
```
</CodeGroup>
Delete all memories of a user:
<CodeGroup>
```python Code
client.delete_all(user_id="alex")
```
```python Output
{'message': 'Memories deleted successfully!'}
```
</CodeGroup>

1
mem0/__init__.py
View File

@@ -3,3 +3,4 @@ import importlib.metadata
__version__ = importlib.metadata.version("mem0ai")
from mem0.memory.main import Memory # noqa
from mem0.client.main import MemoryClient # noqa

0
mem0/client/__init__.py Normal file
View File

243
mem0/client/main.py Normal file
View File

@@ -0,0 +1,243 @@
import httpx
import os
import logging
import warnings
from typing import Optional, Dict, Any
from mem0.memory.setup import setup_config
from mem0.memory.telemetry import capture_client_event
logger = logging.getLogger(__name__)
# Setup user config
setup_config()
class MemoryClient:
def __init__(self, api_key: Optional[str] = None, host: Optional[str] = None):
"""
Initialize the Mem0 client.
Args:
api_key (Optional[str]): API Key from Mem0 Platform. Defaults to environment variable 'MEM0_API_KEY' if not provided.
host (Optional[str]): API host URL. Defaults to 'https://api.mem0.ai/v1'.
"""
self.api_key = api_key or os.getenv("MEM0_API_KEY")
self.host = host or "https://api.mem0.ai/v1"
self.client = httpx.Client(
base_url=self.host,
headers={"Authorization": f"Token {self.api_key}"},
)
self._validate_api_key()
capture_client_event("client.init", self)
def _validate_api_key(self):
if not self.api_key:
warnings.warn("API Key not provided. Please provide an API Key.")
response = self.client.get("/memories/", params={"user_id": "test"})
if response.status_code != 200:
raise ValueError(
"Invalid API Key. Please get a valid API Key from https://app.mem0.ai"
)
def add(
self,
data: str,
user_id: Optional[str] = None,
agent_id: Optional[str] = None,
session_id: Optional[str] = None,
metadata: Optional[Dict[str, Any]] = None,
filters: Optional[Dict[str, Any]] = None,
) -> Dict[str, Any]:
"""
Create a new memory.
Args:
data (str): The data to be stored in the memory.
user_id (Optional[str]): User ID to save the memory specific to a user. Defaults to None.
agent_id (Optional[str]): Agent ID for agent-specific memory. Defaults to None.
session_id (Optional[str]): Run ID to save memory for a specific session. Defaults to None.
metadata (Optional[Dict[str, Any]]): Metadata to be saved with the memory. Defaults to None.
filters (Optional[Dict[str, Any]]): Filters to apply to the memory. Defaults to None.
Returns:
Dict[str, Any]: The response from the server.
"""
capture_client_event("client.add", self)
payload = {"text": data}
if metadata:
payload["metadata"] = metadata
if filters:
payload["filters"] = filters
if user_id:
payload["user_id"] = user_id
if agent_id:
payload["agent_id"] = agent_id
if session_id:
payload["run_id"] = session_id
response = self.client.post("/memories/", json=payload, timeout=60)
if response.status_code != 200:
logger.error(response.json())
raise ValueError(f"Failed to add memory. Response: {response.json()}")
return response.json()
def get(self, memory_id: str) -> Dict[str, Any]:
"""
Get a memory by ID.
Args:
memory_id (str): Memory ID.
Returns:
Dict[str, Any]: The memory data.
"""
capture_client_event("client.get", self)
response = self.client.get(f"/memories/{memory_id}/")
return response.json()
def get_all(
self,
user_id: Optional[str] = None,
agent_id: Optional[str] = None,
session_id: Optional[str] = None,
limit: int = 100,
) -> Dict[str, Any]:
"""
Get all memories.
Args:
user_id (Optional[str]): User ID to filter memories. Defaults to None.
agent_id (Optional[str]): Agent ID to filter memories. Defaults to None.
session_id (Optional[str]): Run ID to filter memories. Defaults to None.
limit (int): Number of memories to return. Defaults to 100.
Returns:
Dict[str, Any]: The list of memories.
"""
params = {
"user_id": user_id,
"agent_id": agent_id,
"run_id": session_id,
"limit": limit,
}
response = self.client.get(
"/memories/", params={k: v for k, v in params.items() if v is not None}
)
capture_client_event(
"client.get_all", self, {"filters": len(params), "limit": limit}
)
return response.json()
def search(
self,
query: str,
user_id: Optional[str] = None,
agent_id: Optional[str] = None,
session_id: Optional[str] = None,
limit: int = 100,
filters: Optional[Dict[str, Any]] = None,
) -> Dict[str, Any]:
"""
Search memories.
Args:
query (str): Query to search for in the memories.
user_id (Optional[str]): User ID to filter memories. Defaults to None.
agent_id (Optional[str]): Agent ID to filter memories. Defaults to None.
session_id (Optional[str]): Run ID to filter memories. Defaults to None.
limit (int): Number of memories to return. Defaults to 100.
filters (Optional[Dict[str, Any]]): Filters to apply to the search. Defaults to None.
Returns:
Dict[str, Any]: The search results.
"""
payload = {
"text": query,
"limit": limit,
"filters": filters,
"user_id": user_id,
"agent_id": agent_id,
"run_id": session_id,
}
response = self.client.post("/memories/search/", json=payload)
capture_client_event("client.search", self, {"limit": limit})
return response.json()
def update(self, memory_id: str, data: str) -> Dict[str, Any]:
"""
Update a memory by ID.
Args:
memory_id (str): Memory ID.
data (str): Data to update in the memory.
Returns:
Dict[str, Any]: The response from the server.
"""
capture_client_event("client.update", self)
response = self.client.put(f"/memories/{memory_id}/", json={"text": data})
return response.json()
def delete(self, memory_id: str) -> Dict[str, Any]:
"""
Delete a memory by ID.
Args:
memory_id (str): Memory ID.
Returns:
Dict[str, Any]: The response from the server.
"""
capture_client_event("client.delete", self)
response = self.client.delete(f"/memories/{memory_id}/")
return response.json()
def delete_all(
self,
user_id: Optional[str] = None,
agent_id: Optional[str] = None,
session_id: Optional[str] = None,
) -> Dict[str, Any]:
"""
Delete all memories.
Args:
user_id (Optional[str]): User ID to filter memories. Defaults to None.
agent_id (Optional[str]): Agent ID to filter memories. Defaults to None.
session_id (Optional[str]): Run ID to filter memories. Defaults to None.
Returns:
Dict[str, Any]: The response from the server.
"""
params = {"user_id": user_id, "agent_id": agent_id, "run_id": session_id}
response = self.client.delete(
"/memories/", params={k: v for k, v in params.items() if v is not None}
)
capture_client_event("client.delete_all", self, {"params": len(params)})
return response.json()
def history(self, memory_id: str) -> Dict[str, Any]:
"""
Get history of a memory by ID.
Args:
memory_id (str): Memory ID.
Returns:
Dict[str, Any]: The memory history.
"""
response = self.client.get(f"/memories/{memory_id}/history/")
capture_client_event("client.history", self)
return response.json()
def reset(self):
"""
Reset the client. (Not implemented yet)
"""
raise NotImplementedError("Reset is not implemented yet")
def chat(self):
"""
Start a chat with the Mem0 AI. (Not implemented yet)
"""
raise NotImplementedError("Chat is not implemented yet")

10
mem0/memory/telemetry.py
View File

@@ -59,3 +59,13 @@ def capture_event(event_name, memory_instance, additional_data=None):
event_data.update(additional_data)
telemetry.capture_event(event_name, event_data)
def capture_client_event(event_name, instance, additional_data=None):
event_data = {
"function": f"{instance.__class__.__module__}.{instance.__class__.__name__}",
}
if additional_data:
event_data.update(additional_data)
telemetry.capture_event(event_name, event_data)

2
pyproject.toml
View File

@@ -1,6 +1,6 @@
[tool.poetry]
name = "mem0ai"
version = "0.0.4"
version = "0.0.5"
description = "Long-term memory for AI Agents"
authors = ["Deshraj Yadav <deshraj@mem0.ai>", "Taranjeet Singh <taranjeet@mem0.ai>"]
exclude = [