This is an automated email from the ASF dual-hosted git repository.
ming pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/incubator-hugegraph-ai.git
The following commit(s) were added to refs/heads/main by this push:
new ec4b9da feat: added local LLM API, changed the config displayed in
the initial gradio demo, and added openai's apibase configuration. (#41)
ec4b9da is described below
commit ec4b9da8a774e503f5f36191702961273ff77ec4
Author: vichayturen <[email protected]>
AuthorDate: Fri Apr 26 10:05:08 2024 +0800
feat: added local LLM API, changed the config displayed in the initial
gradio demo, and added openai's apibase configuration. (#41)
* add api_bot
* add local qwen api
change initial gradio_demo config shown
add apibase config on openai
* fix some questions
* fix some questions
* fix dependency
* go through ./style/code_format_and_analysis.sh and check some warnings
* 1. fixed dependencies of local llm api
2. provide some usage examples of local llm api in README
* fix the argument's descriptions
* 1. Removed unused function.
2. Added one configurable arguments max_new_tokens to command line
argumnents.
3. Fixed some code style issues.
* fix some style issues
* 1. fix import error when python version > 3.12
---
.github/workflows/pylint.yml | 1 +
hugegraph-llm/llm_api/README.md | 19 ++
hugegraph-llm/llm_api/main.py | 163 ++++++++++++++++
hugegraph-llm/llm_api/requirements.txt | 6 +
hugegraph-llm/src/hugegraph_llm/config/config.ini | 25 ++-
hugegraph-llm/src/hugegraph_llm/llms/api_bot.py | 78 ++++++++
hugegraph-llm/src/hugegraph_llm/llms/init_llm.py | 4 +
hugegraph-llm/src/hugegraph_llm/llms/openai.py | 3 +
.../operators/hugegraph_op/graph_rag_query.py | 1 +
hugegraph-llm/src/hugegraph_llm/utils/config.py | 8 +-
.../src/hugegraph_llm/utils/gradio_demo.py | 215 +++++++++++----------
11 files changed, 413 insertions(+), 110 deletions(-)
diff --git a/.github/workflows/pylint.yml b/.github/workflows/pylint.yml
index 8ccaf8d..ff6da17 100644
--- a/.github/workflows/pylint.yml
+++ b/.github/workflows/pylint.yml
@@ -23,6 +23,7 @@ jobs:
python -m pip install --upgrade pip
pip install pylint pytest
pip install -r ./hugegraph-llm/requirements.txt
+ pip install -r ./hugegraph-llm/llm_api/requirements.txt
pip install -r ./hugegraph-python-client/requirements.txt
- name: Analysing the code with pylint
run: |
diff --git a/hugegraph-llm/llm_api/README.md b/hugegraph-llm/llm_api/README.md
new file mode 100644
index 0000000..4c8bf31
--- /dev/null
+++ b/hugegraph-llm/llm_api/README.md
@@ -0,0 +1,19 @@
+# Local LLM Api
+
+## Usage
+If hugegraph-llm wants to use local LLM, you can configure it as follows.
+
+Run the program:
+```shell
+python main.py \
+ --model_name_or_path "Qwen/Qwen1.5-0.5B-Chat" \
+ --device "cuda" \
+ --port 7999
+```
+
+The LLM Section of [config.ini](../src/hugegraph_llm/config/config.ini) can be
configured as follows:
+```ini
+[LLM]
+type = local_api
+llm_url = http://localhost:7999/v1/chat/completions
+```
diff --git a/hugegraph-llm/llm_api/main.py b/hugegraph-llm/llm_api/main.py
new file mode 100644
index 0000000..ddd2478
--- /dev/null
+++ b/hugegraph-llm/llm_api/main.py
@@ -0,0 +1,163 @@
+# Licensed to the Apache Software Foundation (ASF) under one
+# or more contributor license agreements. See the NOTICE file
+# distributed with this work for additional information
+# regarding copyright ownership. The ASF licenses this file
+# to you under the Apache License, Version 2.0 (the
+# "License"); you may not use this file except in compliance
+# with the License. You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing,
+# software distributed under the License is distributed on an
+# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+# KIND, either express or implied. See the License for the
+# specific language governing permissions and limitations
+# under the License.
+
+
+import gc
+import sys
+from contextlib import asynccontextmanager
+from enum import Enum
+from typing import Literal, List
+
+import torch
+import uvicorn
+from fastapi import FastAPI, HTTPException, status
+from fastapi.middleware.cors import CORSMiddleware
+from pydantic import BaseModel
+from transformers import AutoModelForCausalLM, AutoTokenizer,
Qwen2ForCausalLM, \
+ Qwen2Tokenizer, GenerationConfig
+if sys.version_info >= (3, 12): # >=3.12
+ from typing import TypedDict
+else: # <3.12
+ from typing_extensions import TypedDict
+
+
+class Message(TypedDict):
+ role: Literal["system", "user", "assistant", "tool", "function"]
+ content: str
+
+
+class ChatRequest(BaseModel):
+ messages: List[Message]
+
+
+class ChatResponse(BaseModel):
+ content: str
+
+
+class Role(Enum):
+ SYSTEM = "system"
+ USER = "user"
+ ASSISTANT = "assistant"
+ TOOL = "tool"
+ FUNCTION = "function"
+
+
+class QwenChatModel:
+ def __init__(self, model_name_or_path: str, device: str = "cuda",
max_new_tokens: int = 512,
+ generation_config: GenerationConfig = None):
+ self.model: Qwen2ForCausalLM = AutoModelForCausalLM.from_pretrained(
+ model_name_or_path,
+ torch_dtype="auto",
+ device_map=device
+ )
+ self.tokenizer: Qwen2Tokenizer =
AutoTokenizer.from_pretrained(model_name_or_path)
+ self.device = torch.device(device)
+ self.max_new_tokens = max_new_tokens
+ self.generation_config = generation_config
+
+ @torch.inference_mode()
+ async def achat(self, messages: List[Message]):
+ text = self.tokenizer.apply_chat_template(
+ messages,
+ tokenize=False,
+ add_generation_prompt=True
+ )
+ model_inputs = self.tokenizer([text],
return_tensors="pt").to(self.device)
+ generated_ids = self.model.generate(
+ model_inputs.input_ids,
+ max_new_tokens=self.max_new_tokens,
+ )
+ generated_ids = [
+ output_ids[len(input_ids):]
+ for input_ids, output_ids in zip(model_inputs.input_ids,
generated_ids)
+ ]
+ response = self.tokenizer.batch_decode(generated_ids,
skip_special_tokens=True)[0]
+ return response
+
+
+def torch_gc() -> None:
+ r"""
+ Collects GPU memory.
+ """
+ gc.collect()
+ if torch.cuda.is_available():
+ torch.cuda.empty_cache()
+ torch.cuda.ipc_collect()
+
+
+@asynccontextmanager
+async def lifespan(app: "FastAPI"): # collects GPU memory
+ yield
+ torch_gc()
+
+
+def create_app(chat_model: "QwenChatModel") -> "FastAPI":
+ app = FastAPI(lifespan=lifespan)
+
+ app.add_middleware(
+ CORSMiddleware,
+ allow_origins=["*"],
+ allow_credentials=True,
+ allow_methods=["*"],
+ allow_headers=["*"],
+ )
+
+ @app.post("/v1/chat/completions", response_model=ChatResponse,
status_code=status.HTTP_200_OK)
+ async def create_chat_completion(request: ChatRequest):
+ if len(request.messages) == 0:
+ raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST,
detail="Invalid length")
+
+ if len(request.messages) % 2 == 0:
+ raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST,
+ detail="Only supports u/a/u/a/u...")
+
+ print("* ============= [input] ============= *")
+ print(request.messages[-1]["content"])
+
+ content = await chat_model.achat(
+ messages=request.messages,
+ )
+ print("* ============= [output] ============= *")
+ print(content)
+
+ return ChatResponse(content=content)
+
+ return app
+
+
+def main():
+ import argparse
+ parser = argparse.ArgumentParser(description="Local LLM Api for Hugegraph
LLM.")
+ parser.add_argument("--model_name_or_path", type=str, required=True,
help="Model name or path")
+ parser.add_argument("--device", type=str, default="cpu", help="Device to
use")
+ parser.add_argument("--port", type=int, default=7999, help="Port of the
service")
+ parser.add_argument("--max_new_tokens", type=int, default=512,
+ help="The max number of tokens to generate")
+
+ args = parser.parse_args()
+
+ model_path = args.model_name_or_path
+ device = args.device
+ port = args.port
+ max_new_tokens = args.max_new_tokens
+ chat_model = QwenChatModel(model_path, device, max_new_tokens)
+ app = create_app(chat_model)
+ uvicorn.run(app, host="0.0.0.0", port=port, workers=1)
+
+
+if __name__ == '__main__':
+ main()
diff --git a/hugegraph-llm/llm_api/requirements.txt
b/hugegraph-llm/llm_api/requirements.txt
new file mode 100644
index 0000000..d1661f7
--- /dev/null
+++ b/hugegraph-llm/llm_api/requirements.txt
@@ -0,0 +1,6 @@
+torch==2.1.2
+transformers==4.39.3
+fastapi==0.110.1
+accelerate==0.29.2
+charset-normalizer==3.3.2
+uvicorn==0.29.0
diff --git a/hugegraph-llm/src/hugegraph_llm/config/config.ini
b/hugegraph-llm/src/hugegraph_llm/config/config.ini
index 7ff45c4..d3ca7d3 100644
--- a/hugegraph-llm/src/hugegraph_llm/config/config.ini
+++ b/hugegraph-llm/src/hugegraph_llm/config/config.ini
@@ -24,9 +24,30 @@ pwd = admin
graph = hugegraph
[llm]
-type = openai
+## local llm
+# type = local_api
+# llm_url = http://localhost:7999/v1/chat/completions
+#
+## openai
+# type = openai
+# api_key = xxx
+# api_base = xxx
+# model_name = gpt-3.5-turbo-16k
+# max_token = 4000
+#
+## ernie
+# type = ernie
+# api_key = xxx
+# secret_key = xxx
+# llm_url = xxx
+# model_name = ernie
+#
+# type = openai
+type = local_api
api_key = xxx
+api_base = https://api.openai.com/v1
secret_key = xxx
-llm_url =
https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/completions_pro?access_token=
+# llm_url =
https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/completions_pro?access_token=
+llm_url = http://localhost:7999/v1/chat/completions
model_name = gpt-3.5-turbo-16k
max_token = 4000
diff --git a/hugegraph-llm/src/hugegraph_llm/llms/api_bot.py
b/hugegraph-llm/src/hugegraph_llm/llms/api_bot.py
new file mode 100644
index 0000000..d18d792
--- /dev/null
+++ b/hugegraph-llm/src/hugegraph_llm/llms/api_bot.py
@@ -0,0 +1,78 @@
+# Licensed to the Apache Software Foundation (ASF) under one
+# or more contributor license agreements. See the NOTICE file
+# distributed with this work for additional information
+# regarding copyright ownership. The ASF licenses this file
+# to you under the Apache License, Version 2.0 (the
+# "License"); you may not use this file except in compliance
+# with the License. You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing,
+# software distributed under the License is distributed on an
+# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+# KIND, either express or implied. See the License for the
+# specific language governing permissions and limitations
+# under the License.
+
+import json
+from typing import Optional, List, Dict, Any, Callable
+
+import requests
+from retry import retry
+
+from hugegraph_llm.llms.base import BaseLLM
+from hugegraph_llm.utils.config import Config
+from hugegraph_llm.utils.constants import Constants
+
+
+class ApiBotClient(BaseLLM):
+ def __init__(self):
+ self.c = Config(section=Constants.LLM_CONFIG)
+ self.base_url = self.c.get_llm_url()
+
+ @retry(tries=3, delay=1)
+ def generate(
+ self,
+ messages: Optional[List[Dict[str, Any]]] = None,
+ prompt: Optional[str] = None,
+ ) -> str:
+ if messages is None:
+ assert prompt is not None, "Messages or prompt must be provided."
+ messages = [{"role": "user", "content": prompt}]
+ url = self.base_url
+
+ payload = json.dumps({
+ "messages": messages,
+ })
+ headers = {"Content-Type": "application/json"}
+ response = requests.request("POST", url, headers=headers,
data=payload, timeout=30)
+ if response.status_code != 200:
+ raise Exception(
+ f"Request failed with code {response.status_code}, message:
{response.text}"
+ )
+ response_json = json.loads(response.text)
+ return response_json["content"]
+
+ def generate_streaming(
+ self,
+ messages: Optional[List[Dict[str, Any]]] = None,
+ prompt: Optional[str] = None,
+ on_token_callback: Callable = None,
+ ) -> str:
+ return self.generate(messages, prompt)
+
+ def num_tokens_from_string(self, string: str) -> int:
+ return len(string)
+
+ def max_allowed_token_length(self) -> int:
+ return 4096
+
+ def get_llm_type(self) -> str:
+ return "local_api"
+
+
+if __name__ == "__main__":
+ client = ApiBotClient()
+ print(client.generate(prompt="What is the capital of China?"))
+ print(client.generate(messages=[{"role": "user", "content": "What is the
capital of China?"}]))
diff --git a/hugegraph-llm/src/hugegraph_llm/llms/init_llm.py
b/hugegraph-llm/src/hugegraph_llm/llms/init_llm.py
index f8e7138..1fb8859 100644
--- a/hugegraph-llm/src/hugegraph_llm/llms/init_llm.py
+++ b/hugegraph-llm/src/hugegraph_llm/llms/init_llm.py
@@ -17,6 +17,7 @@
from hugegraph_llm.llms.openai import OpenAIChat
from hugegraph_llm.llms.ernie_bot import ErnieBotClient
+from hugegraph_llm.llms.api_bot import ApiBotClient
from hugegraph_llm.utils.config import Config
from hugegraph_llm.utils.constants import Constants
@@ -32,9 +33,12 @@ class LLMs:
if self.config.get_llm_type() == "openai":
return OpenAIChat(
api_key=self.config.get_llm_api_key(),
+ api_base=self.config.get_llm_api_base(),
model_name=self.config.get_llm_model_name(),
max_tokens=self.config.get_llm_max_token(),
)
+ if self.config.get_llm_type() == "local_api":
+ return ApiBotClient()
raise Exception("llm type is not supported !")
diff --git a/hugegraph-llm/src/hugegraph_llm/llms/openai.py
b/hugegraph-llm/src/hugegraph_llm/llms/openai.py
index c9b753c..58af40c 100644
--- a/hugegraph-llm/src/hugegraph_llm/llms/openai.py
+++ b/hugegraph-llm/src/hugegraph_llm/llms/openai.py
@@ -31,11 +31,14 @@ class OpenAIChat(BaseLLM):
def __init__(
self,
api_key: Optional[str] = None,
+ api_base: Optional[str] = None,
model_name: str = "gpt-3.5-turbo",
max_tokens: int = 1000,
temperature: float = 0.0,
) -> None:
openai.api_key = api_key or os.getenv("OPENAI_API_KEY")
+ if api_base is not None:
+ openai.api_base = api_base or os.getenv("OPENAI_API_BASE")
self.model = model_name
self.max_tokens = max_tokens
self.temperature = temperature
diff --git
a/hugegraph-llm/src/hugegraph_llm/operators/hugegraph_op/graph_rag_query.py
b/hugegraph-llm/src/hugegraph_llm/operators/hugegraph_op/graph_rag_query.py
index 6d5ede6..3a0905f 100644
--- a/hugegraph-llm/src/hugegraph_llm/operators/hugegraph_op/graph_rag_query.py
+++ b/hugegraph-llm/src/hugegraph_llm/operators/hugegraph_op/graph_rag_query.py
@@ -84,6 +84,7 @@ class GraphRAGQuery:
self._prop_to_match = prop_to_match
self._schema = ""
+
def run(self, context: Dict[str, Any]) -> Dict[str, Any]:
if self._client is None:
if isinstance(context.get("graph_client"), PyHugeClient):
diff --git a/hugegraph-llm/src/hugegraph_llm/utils/config.py
b/hugegraph-llm/src/hugegraph_llm/utils/config.py
index aa73fbb..5a205de 100644
--- a/hugegraph-llm/src/hugegraph_llm/utils/config.py
+++ b/hugegraph-llm/src/hugegraph_llm/utils/config.py
@@ -17,6 +17,7 @@
import configparser
import os
+from .constants import Constants
class Config:
@@ -35,8 +36,8 @@ class Config:
if not os.path.exists(config_file):
config = configparser.ConfigParser()
- config.add_section("llm")
- config.add_section("hugegraph")
+ config.add_section(Constants.HUGEGRAPH_CONFIG)
+ config.add_section(Constants.LLM_CONFIG)
with open(config_file, "w", encoding="utf-8") as file:
config.write(file)
return config_file
@@ -68,6 +69,9 @@ class Config:
def get_llm_api_key(self):
return self.config.get(self.section, "api_key")
+ def get_llm_api_base(self):
+ return self.config.get(self.section, "api_base")
+
def get_llm_secret_key(self):
return self.config.get(self.section, "secret_key")
diff --git a/hugegraph-llm/src/hugegraph_llm/utils/gradio_demo.py
b/hugegraph-llm/src/hugegraph_llm/utils/gradio_demo.py
index de97074..a5df0d7 100644
--- a/hugegraph-llm/src/hugegraph_llm/utils/gradio_demo.py
+++ b/hugegraph-llm/src/hugegraph_llm/utils/gradio_demo.py
@@ -22,13 +22,13 @@ import os
import gradio as gr
import uvicorn
from fastapi import FastAPI
+from pyhugegraph.client import PyHugeClient
from hugegraph_llm.llms.init_llm import LLMs
from hugegraph_llm.operators.graph_rag_task import GraphRAG
from hugegraph_llm.operators.kg_construction_task import KgBuilder
from hugegraph_llm.utils.config import Config
from hugegraph_llm.utils.constants import Constants
-from pyhugegraph.client import PyHugeClient
def init_hg_test_data():
@@ -122,12 +122,12 @@ def get_hg_client():
def init_config(
- ip, port, user, pwd, graph, type, api_key, secret_key, llm_url,
model_name, max_token
+ ip, port, user, pwd, graph, type, api_key, secret_key, llm_url,
model_name, max_token
):
root_dir =
os.path.dirname(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
config_file = os.path.join(root_dir, "hugegraph_llm", "config",
"config.ini")
- config = Config(config_file=config_file, section="hugegraph")
+ config = Config(config_file=config_file,
section=Constants.HUGEGRAPH_CONFIG)
config.update_config({"ip": ip, "port": port, "user": user, "pwd": pwd,
"graph": graph})
config = Config(config_file=config_file, section="llm")
@@ -146,110 +146,113 @@ def init_config(
return content
-with gr.Blocks() as hugegraph_llm:
- gr.Markdown(
- """# HugeGraph LLM Demo
- 1. Set up the HugeGraph server."""
- )
- with gr.Row():
- inp = [
- gr.Textbox(value="127.0.0.1", label="ip"),
- gr.Textbox(value="8080", label="port"),
- gr.Textbox(value="admin", label="user"),
- gr.Textbox(value="admin", label="pwd"),
- gr.Textbox(value="hugegraph", label="graph"),
- ]
- gr.Markdown("2. Set up the LLM.")
- with gr.Row():
- inp2 = [
- gr.Textbox(value="ernie", label="type"),
- gr.Textbox(value="", label="api_key"),
- gr.Textbox(value="", label="secret_key"),
- gr.Textbox(
-
value="https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/";
- "chat/completions_pro?access_token=",
- label="llm_url",
- ),
- gr.Textbox(value="wenxin", label="model_name"),
- gr.Textbox(value="4000", label="max_token"),
- ]
- with gr.Row():
- out = gr.Textbox(label="Output")
- btn = gr.Button("Initialize configs")
- btn.click(fn=init_config, inputs=inp + inp2, outputs=out) # pylint:
disable=no-member
-
- gr.Markdown(
- """## 1. build knowledge graph
- - Text: The input text.
- - Schema: Accepts two types of text as below:
- - User-defined JSON format Schema.
- - Specify the name of the HugeGraph graph instance, and it will
- automatically extract the schema of the graph.
- - Disambiguate word sense: Whether to perform word sense disambiguation.
- - Commit to hugegraph: Whether to commit the constructed knowledge graph
to the
- HugeGraph server.
- """
- )
- TEXT = (
- "Meet Sarah, a 30-year-old attorney, and her roommate, James, whom
she's shared a home with"
- " since 2010. James, in his professional life, works as a journalist.
Additionally, Sarah"
- " is the proud owner of the website www.sarahsplace.com, while James
manages his own"
- " webpage, though the specific URL is not mentioned here. These two
individuals, Sarah and"
- " James, have not only forged a strong personal bond as roommates but
have also carved out"
- " their distinctive digital presence through their respective
webpages, showcasing their"
- " varied interests and experiences."
- )
-
- SCHEMA = """{
- "vertices": [
- {"vertex_label": "person", "properties": ["name", "age",
"occupation"]},
- {"vertex_label": "webpage", "properties": ["name", "url"]}
- ],
- "edges": [
- {
- "edge_label": "roommate",
- "source_vertex_label": "person",
- "target_vertex_label": "person",
- "properties": {}
- }
- ]
- }
- """
-
- with gr.Row():
- inp = [
- gr.Textbox(value=TEXT, label="Text"),
- gr.Textbox(value=SCHEMA, label="Schema"),
- gr.Textbox(value="false", label="Disambiguate word sense"),
- gr.Textbox(value="false", label="Commit to hugegraph"),
- ]
- with gr.Row():
- out = gr.Textbox(label="Output")
- btn = gr.Button("Build knowledge graph")
- btn.click(fn=build_kg, inputs=inp, outputs=out) # pylint:
disable=no-member
-
- gr.Markdown("""## 2. Retrieval augmented generation by hugegraph""")
- with gr.Row():
- inp = gr.Textbox(value="Tell me about Al Pacino.", label="Question")
- with gr.Row():
- out = gr.Textbox(label="Answer")
- btn = gr.Button("Retrieval augmented generation")
- btn.click(fn=graph_rag, inputs=inp, outputs=out) # pylint:
disable=no-member
-
- gr.Markdown("""## 3. Others """)
- with gr.Row():
- inp = []
- out = gr.Textbox(label="Output")
- btn = gr.Button("Initialize HugeGraph test data")
- btn.click(fn=init_hg_test_data, inputs=inp, outputs=out) # pylint:
disable=no-member
-
- with gr.Row():
- inp = gr.Textbox(value="g.V().limit(10)", label="Gremlin query")
- out = gr.Textbox(label="Output")
- btn = gr.Button("Run gremlin query on HugeGraph")
- btn.click(fn=run_gremlin_query, inputs=inp, outputs=out) # pylint:
disable=no-member
-
if __name__ == "__main__":
app = FastAPI()
+ initial_config = Config(section=Constants.HUGEGRAPH_CONFIG)
+ with gr.Blocks() as hugegraph_llm:
+ gr.Markdown(
+ """# HugeGraph LLM Demo
+ 1. Set up the HugeGraph server."""
+ )
+ with gr.Row():
+ inp = [
+ gr.Textbox(value=initial_config.get_graph_ip(), label="ip"),
+ gr.Textbox(value=initial_config.get_graph_port(),
label="port"),
+ gr.Textbox(value=initial_config.get_graph_user(),
label="user"),
+ gr.Textbox(value=initial_config.get_graph_pwd(), label="pwd"),
+ gr.Textbox(value=initial_config.get_graph_name(),
label="graph"),
+ ]
+ gr.Markdown("2. Set up the LLM.")
+ with gr.Row():
+ inp2 = [
+ gr.Textbox(value="ernie", label="type"),
+ gr.Textbox(value="", label="api_key"),
+ gr.Textbox(value="", label="secret_key"),
+ gr.Textbox(
+
value="https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/";
+ "chat/completions_pro?access_token=",
+ label="llm_url",
+ ),
+ gr.Textbox(value="wenxin", label="model_name"),
+ gr.Textbox(value="4000", label="max_token"),
+ ]
+ with gr.Row():
+ out = gr.Textbox(label="Output")
+ btn = gr.Button("Initialize configs")
+ btn.click(fn=init_config, inputs=inp + inp2, outputs=out) # pylint:
disable=no-member
+
+ gr.Markdown(
+ """## 1. build knowledge graph
+ - Text: The input text.
+ - Schema: Accepts two types of text as below:
+ - User-defined JSON format Schema.
+ - Specify the name of the HugeGraph graph instance, and it will
+ automatically extract the schema of the graph.
+ - Disambiguate word sense: Whether to perform word sense
disambiguation.
+ - Commit to hugegraph: Whether to commit the constructed knowledge
graph to the
+ HugeGraph server.
+ """
+ )
+ TEXT = (
+ "Meet Sarah, a 30-year-old attorney, and her roommate,"
+ " James, whom she's shared a home with since 2010. James,"
+ " in his professional life, works as a journalist. Additionally,"
+ " Sarah is the proud owner of the website www.sarahsplace.com,"
+ " while James manages his own webpage, though the specific URL"
+ " is not mentioned here. These two individuals, Sarah and James,"
+ " have not only forged a strong personal bond as roommates but"
+ " have also carved out their distinctive digital presence through"
+ " their respective webpages, showcasing their varied interests and"
+ " experiences."
+ )
+
+ SCHEMA = """{
+ "vertices": [
+ {"vertex_label": "person", "properties": ["name", "age",
"occupation"]},
+ {"vertex_label": "webpage", "properties": ["name", "url"]}
+ ],
+ "edges": [
+ {
+ "edge_label": "roommate",
+ "source_vertex_label": "person",
+ "target_vertex_label": "person",
+ "properties": {}
+ }
+ ]
+ }
+ """
+
+ with gr.Row():
+ inp = [
+ gr.Textbox(value=TEXT, label="Text"),
+ gr.Textbox(value=SCHEMA, label="Schema"),
+ gr.Textbox(value="false", label="Disambiguate word sense"),
+ gr.Textbox(value="false", label="Commit to hugegraph"),
+ ]
+ with gr.Row():
+ out = gr.Textbox(label="Output")
+ btn = gr.Button("Build knowledge graph")
+ btn.click(fn=build_kg, inputs=inp, outputs=out) # pylint:
disable=no-member
+
+ gr.Markdown("""## 2. Retrieval augmented generation by hugegraph""")
+ with gr.Row():
+ inp = gr.Textbox(value="Tell me about Al Pacino.",
label="Question")
+ with gr.Row():
+ out = gr.Textbox(label="Answer")
+ btn = gr.Button("Retrieval augmented generation")
+ btn.click(fn=graph_rag, inputs=inp, outputs=out) # pylint:
disable=no-member
+
+ gr.Markdown("""## 3. Others """)
+ with gr.Row():
+ inp = []
+ out = gr.Textbox(label="Output")
+ btn = gr.Button("Initialize HugeGraph test data")
+ btn.click(fn=init_hg_test_data, inputs=inp, outputs=out) # pylint:
disable=no-member
+
+ with gr.Row():
+ inp = gr.Textbox(value="g.V().limit(10)", label="Gremlin query")
+ out = gr.Textbox(label="Output")
+ btn = gr.Button("Run gremlin query on HugeGraph")
+ btn.click(fn=run_gremlin_query, inputs=inp, outputs=out) # pylint:
disable=no-member
app = gr.mount_gradio_app(app, hugegraph_llm, path="/")
uvicorn.run(app, host="0.0.0.0", port=8001)
