feat:vllm llm support version0

Author: kakack

.gitignore

@@ -186,6 +186,9 @@ dmypy.json
 # Cython debug symbols
 cython_debug/
 
+# auth file
+*_auth.yaml
+
 # PyCharm
 #  JetBrains specific template is maintained in a separate JetBrains.gitignore that can
 #  be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore

docs/getting_started/memory_scheduler_tutorial.md

@@ -0,0 +1,113 @@
+# Mem Scheduler Beginner's Guide
+
+Welcome to the mem_scheduler tutorial!
+
+mem_scheduler is a system for managing memory and conversation scheduling, consisting of:
+- Mem Cube: Stores and manages user memory data
+- Scheduler: Coordinates memory storage and retrieval processes
+- Dispatcher (Router): Trigger different memory reorganization strategies by checking messages from MemOS systems.
+- Message Queue Hub: Central communication backbone
+
+Memory Scheduler is built to optimize memory when MemOS is running. Here are two approaches to initializing Memory Scheduler with MemOS.
+
+An example code can be found in ```examples/mem_scheduler/schedule_w_memos.py```
+
+## Configs for Initialization
+
+Below is an example YAML configuration. You can find this in  ```examples/data/config/mem_scheduler/memos_config_w_scheduler.yaml```
+
+```
+user_id: "root"
+chat_model:
+  backend: "huggingface"
+  config:
+    model_name_or_path: "Qwen/Qwen3-1.7B"
+    temperature: 0.1
+    remove_think_prefix: true
+    max_tokens: 4096
+mem_reader:
+  backend: "simple_struct"
+  config:
+    llm:
+      backend: "ollama"
+      config:
+        model_name_or_path: "qwen3:0.6b"
+        remove_think_prefix: true
+        temperature: 0.8
+        max_tokens: 1024
+        top_p: 0.9
+        top_k: 50
+    embedder:
+      backend: "ollama"
+      config:
+        model_name_or_path: "nomic-embed-text:latest"
+    chunker:
+      backend: "sentence"
+      config:
+        tokenizer_or_token_counter: "gpt2"
+        chunk_size: 512
+        chunk_overlap: 128
+        min_sentences_per_chunk: 1
+mem_scheduler:
+  backend: "general_scheduler"
+  config:
+    top_k: 10
+    top_n: 5
+    act_mem_update_interval: 300
+    context_window_size: 5
+    activation_mem_size: 1000
+    thread_pool_max_workers: 10
+    consume_interval_seconds: 3
+    enable_parallel_dispatch: true
+max_turns_window: 20
+top_k: 5
+enable_textual_memory: true
+enable_activation_memory: true
+enable_parametric_memory: false
+enable_mem_scheduler: true
+```
+## Steps to Initialize with MemOS
+
+### 1. Load config and initialize MOS
+```
+config = parse_yaml("./examples/data/config/mem_scheduler/memos_config_w_scheduler.yaml")
+mos_config = MOSConfig(**config)
+mos = MOS(mos_config)
+```
+### 2. Create a User
+```
+user_id = "user_1"
+mos.create_user(user_id)
+```
+### 3. Create and Register a Memory Cube
+```
+config = GeneralMemCubeConfig.from_yaml_file("mem_cube_config.yaml")
+mem_cube_id = "mem_cube_5"
+mem_cube = GeneralMemCube(config)
+mem_cube.dump(mem_cube_name_or_path)
+mos.register_mem_cube(mem_cube_name_or_path, mem_cube_id, user_id)
+```
+
+## Run with MemOS
+
+### 4. Add Conversations (transformed to memories) to Memory Cube
+```
+mos.add(conversations, user_id=user_id, mem_cube_id=mem_cube_id)
+```
+
+### 5. Scheduler with the chat function of MemOS
+```
+for item in questions:
+    query = item["question"]
+    response = mos.chat(query, user_id=user_id)
+    print(f"Query:\n {query}\n\nAnswer:\n {response}")
+```
+### 6. Display Logs and Stop the Scheduler
+```
+show_web_logs(mos.mem_scheduler)
+mos.mem_scheduler.stop()
+```
+
+## Check the Scheduling Info
+
+This guide provides a foundational understanding of setting up and using mem_scheduler. Explore and modify configurations to suit your needs!

examples/data/config/mem_scheduler/mem_cube_config.yaml

@@ -20,7 +20,7 @@ text_mem:
     graph_db:
       backend: "neo4j"
       config:
-        uri: "bolt://123.57.48.226:7687"
+        uri: "bolt://localhost:7687"
         user: "neo4j"
         password: "12345678"
         db_name: "user11alice"

examples/mem_os/chat_w_generated_cube_explicit_memory.py

@@ -84,7 +84,7 @@
                 "graph_db": {
                     "backend": "neo4j",
                     "config": {
-                        "uri": "bolt://123.57.48.226:7687",
+                        "uri": "bolt://localhost:7687",
                         "user": "neo4j",
                         "password": "12345678",
                         "db_name": "user03alice11",

examples/mem_os/multi_user_memos_example.py

@@ -0,0 +1,125 @@
+"""
+Example demonstrating how to use MOSProduct for multi-user scenarios.
+"""
+
+from memos.configs.mem_cube import GeneralMemCubeConfig
+from memos.configs.mem_os import MOSConfig
+from memos.mem_cube.general import GeneralMemCube
+from memos.mem_os.product import MOSProduct
+
+
+def get_config(user_name):
+    openapi_config = {
+        "model_name_or_path": "gpt-4o-mini",
+        "temperature": 0.8,
+        "max_tokens": 1024,
+        "top_p": 0.9,
+        "top_k": 50,
+        "remove_think_prefix": True,
+        "api_key": "your-api-key-here",
+        "api_base": "https://api.openai.com/v1",
+    }
+    # Create a default configuration
+    default_config = MOSConfig(
+        user_id="root",
+        chat_model={"backend": "openai", "config": openapi_config},
+        mem_reader={
+            "backend": "naive",
+            "config": {
+                "llm": {
+                    "backend": "openai",
+                    "config": openapi_config,
+                },
+                "embedder": {
+                    "backend": "ollama",
+                    "config": {
+                        "model_name_or_path": "nomic-embed-text:latest",
+                    },
+                },
+            },
+        },
+        enable_textual_memory=True,
+        enable_activation_memory=False,
+        top_k=5,
+        max_turns_window=20,
+    )
+    default_cube_config = GeneralMemCubeConfig.model_validate(
+        {
+            "user_id": user_name,
+            "cube_id": f"{user_name}_default_cube",
+            "text_mem": {
+                "backend": "tree_text",
+                "config": {
+                    "extractor_llm": {"backend": "openai", "config": openapi_config},
+                    "dispatcher_llm": {"backend": "openai", "config": openapi_config},
+                    "graph_db": {
+                        "backend": "neo4j",
+                        "config": {
+                            "uri": "bolt://localhost:7687",
+                            "user": "neo4j",
+                            "password": "12345678",
+                            "db_name": user_name,
+                            "auto_create": True,
+                        },
+                    },
+                    "embedder": {
+                        "backend": "ollama",
+                        "config": {
+                            "model_name_or_path": "nomic-embed-text:latest",
+                        },
+                    },
+                },
+            },
+            "act_mem": {},
+            "para_mem": {},
+        }
+    )
+    default_mem_cube = GeneralMemCube(default_cube_config)
+    return default_config, default_mem_cube
+
+
+def main():
+    default_config, default_mem_cube = get_config(user_name="alice")
+    # Initialize MOSProduct with default config
+    mos_product = MOSProduct(default_config=default_config)
+
+    # Register first user with default config
+    result1 = mos_product.user_register(
+        user_id="alice",
+        user_name="alice",
+        interests="I'm interested in machine learning and AI research.",
+        default_mem_cube=default_mem_cube,
+    )
+    print(f"User registration result: {result1}")
+
+    # Chat with Alice
+    print("\n=== Chatting with Alice ===")
+    for response_chunk in mos_product.chat(query="What are my interests?", user_id="alice"):
+        print(response_chunk, end="")
+
+    # Add memory for Alice
+    mos_product.add(
+        user_id="alice",
+        memory_content="I attended a machine learning conference last week.",
+        mem_cube_id=result1["default_cube_id"],
+    )
+
+    # Search memories for Alice
+    search_result = mos_product.search(query="conference", user_id="alice")
+    print(f"\nSearch result for Alice: {search_result}")
+
+    # Search memories for Alice
+    search_result = mos_product.get_all(query="conference", user_id="alice", memory_type="text_mem")
+    print(f"\nSearch result for Alice: {search_result}")
+
+    # List all users
+    users = mos_product.list_users()
+    print(f"\nAll registered users: {users}")
+
+    # Get user info
+    alice_info = mos_product.get_user_info("alice")
+    print(f"\nAlice's info: {alice_info}")
+
+
+if __name__ == "__main__":
+    main()