diff --git a/README.md b/README.md
index 0325cbecbc40a7f6c1c5e2bc75ea4edf82987a33..c55847d55223d0b9df4ec303edf59c21737f8fc4 100644
--- a/README.md
+++ b/README.md
@@ -20,7 +20,7 @@
<img height="21" src="https://img.shields.io/badge/License-Apache--2.0-ffffff?style=flat-square&labelColor=d4eaf7&color=7d09f1" alt="license">
</a>
</p>
-[RagFlow](http://demo.ragflow.io) is a knowledge management platform built on custom-build document understanding engine and LLM, with reasoned and well-founded answers to your question. Clone this repository, you can deploy your own knowledge management platform to empower your business with AI.
+[RagFlow](https://demo.ragflow.io) is a knowledge management platform built on custom-build document understanding engine and LLM, with reasoned and well-founded answers to your question. Clone this repository, you can deploy your own knowledge management platform to empower your business with AI.
<div align="center" style="margin-top:20px;margin-bottom:20px;">
@@ -56,12 +56,12 @@
Then, you need to check the following command:
```bash
-121:/ragflow# sysctl vm.max_map_count
+$ sysctl vm.max_map_count
vm.max_map_count = 262144
```
If **vm.max_map_count** is not greater than 65535:
```bash
-121:/ragflow# sudo sysctl -w vm.max_map_count=262144
+$ sudo sysctl -w vm.max_map_count=262144
```
Note that this change is reset after a system reboot. To render your change permanent, add or update the following line in **/etc/sysctl.conf**:
@@ -126,6 +126,7 @@ Open your browser, enter the IP address of your server, _**Hallelujah**_ again!
<div align="center" style="margin-top:20px;margin-bottom:20px;">
<img src="https://github.com/infiniflow/ragflow/assets/12318111/d6ac5664-c237-4200-a7c2-a4a00691b485" width="1000"/>
</div>
+
## 🔧 Configurations
If you need to change the default setting of the system when you deploy it. There several ways to configure it.
diff --git a/api/apps/llm_app.py b/api/apps/llm_app.py
index e8b3dcdd2f0b326ddce6e195f5412be6d8ba73dc..25bcca25942fb8769618a2193d7968bfb62e2790 100644
--- a/api/apps/llm_app.py
+++ b/api/apps/llm_app.py
@@ -45,7 +45,7 @@ def set_api_key():
for llm in LLMService.query(fid=factory):
if llm.model_type == LLMType.EMBEDDING.value:
mdl = EmbeddingModel[factory](
- req["api_key"], llm.llm_name)
+ req["api_key"], llm.llm_name, req.get("base_url"))
try:
arr, tc = mdl.encode(["Test if the api key is available"])
if len(arr[0]) == 0 or tc == 0:
@@ -54,7 +54,7 @@ def set_api_key():
msg += f"\nFail to access embedding model({llm.llm_name}) using this api key." + str(e)
elif not chat_passed and llm.model_type == LLMType.CHAT.value:
mdl = ChatModel[factory](
- req["api_key"], llm.llm_name)
+ req["api_key"], llm.llm_name, req.get("base_url"))
try:
m, tc = mdl.chat(None, [{"role": "user", "content": "Hello! How are you doing!"}], {
"temperature": 0.9})
@@ -83,7 +83,9 @@ def set_api_key():
llm_factory=factory,
llm_name=llm.llm_name,
model_type=llm.model_type,
- api_key=req["api_key"])
+ api_key=req["api_key"],
+ api_base=req.get("base_url", "")
+ )
return get_json_result(data=True)
diff --git a/api/db/services/llm_service.py b/api/db/services/llm_service.py
index f4e2a41c1cde9fed8b696e2026bd4636a7f50dd4..83eb82f162331d45b0bdbb5cf183d449dc79c108 100644
--- a/api/db/services/llm_service.py
+++ b/api/db/services/llm_service.py
@@ -84,19 +84,21 @@ class TenantLLMService(CommonService):
if model_config["llm_factory"] not in EmbeddingModel:
return
return EmbeddingModel[model_config["llm_factory"]](
- model_config["api_key"], model_config["llm_name"])
+ model_config["api_key"], model_config["llm_name"], model_config["api_base"])
if llm_type == LLMType.IMAGE2TEXT.value:
if model_config["llm_factory"] not in CvModel:
return
return CvModel[model_config["llm_factory"]](
- model_config["api_key"], model_config["llm_name"], lang)
+ model_config["api_key"], model_config["llm_name"], lang,
+ base_url=model_config["api_base"]
+ )
if llm_type == LLMType.CHAT.value:
if model_config["llm_factory"] not in ChatModel:
return
return ChatModel[model_config["llm_factory"]](
- model_config["api_key"], model_config["llm_name"])
+ model_config["api_key"], model_config["llm_name"], model_config["api_base"])
@classmethod
@DB.connection_context()
diff --git a/deepdoc/vision/recognizer.py b/deepdoc/vision/recognizer.py
index ad8b9ba24fa1310fac7827d7a746b580dc552587..67e096ef2def315e1b51a1891e741491cbee89d2 100644
--- a/deepdoc/vision/recognizer.py
+++ b/deepdoc/vision/recognizer.py
@@ -43,6 +43,8 @@ class Recognizer(object):
if not os.path.exists(model_file_path):
model_dir = snapshot_download(repo_id="InfiniFlow/deepdoc")
model_file_path = os.path.join(model_dir, task_name + ".onnx")
+ else:
+ model_file_path = os.path.join(model_dir, task_name + ".onnx")
if not os.path.exists(model_file_path):
raise ValueError("not find model file path {}".format(
diff --git a/rag/llm/chat_model.py b/rag/llm/chat_model.py
index e44af5360c7513759f9a69dd756462d42449d291..af3bbac33f76c13aece7f15f9189b3aedea1e241 100644
--- a/rag/llm/chat_model.py
+++ b/rag/llm/chat_model.py
@@ -31,8 +31,9 @@ class Base(ABC):
class GptTurbo(Base):
- def __init__(self, key, model_name="gpt-3.5-turbo"):
- self.client = OpenAI(api_key=key)
+ def __init__(self, key, model_name="gpt-3.5-turbo", base_url="https://api.openai.com/v1"):
+ if not base_url: base_url="https://api.openai.com/v1"
+ self.client = OpenAI(api_key=key, base_url=base_url)
self.model_name = model_name
def chat(self, system, history, gen_conf):
@@ -53,9 +54,10 @@ class GptTurbo(Base):
class MoonshotChat(GptTurbo):
- def __init__(self, key, model_name="moonshot-v1-8k"):
+ def __init__(self, key, model_name="moonshot-v1-8k", base_url="https://api.moonshot.cn/v1"):
+ if not base_url: base_url="https://api.moonshot.cn/v1"
self.client = OpenAI(
- api_key=key, base_url="https://api.moonshot.cn/v1",)
+ api_key=key, base_url=base_url)
self.model_name = model_name
def chat(self, system, history, gen_conf):
@@ -76,7 +78,7 @@ class MoonshotChat(GptTurbo):
class QWenChat(Base):
- def __init__(self, key, model_name=Generation.Models.qwen_turbo):
+ def __init__(self, key, model_name=Generation.Models.qwen_turbo, **kwargs):
import dashscope
dashscope.api_key = key
self.model_name = model_name
@@ -105,7 +107,7 @@ class QWenChat(Base):
class ZhipuChat(Base):
- def __init__(self, key, model_name="glm-3-turbo"):
+ def __init__(self, key, model_name="glm-3-turbo", **kwargs):
self.client = ZhipuAI(api_key=key)
self.model_name = model_name
@@ -154,7 +156,7 @@ class LocalLLM(Base):
return do_rpc
- def __init__(self, key, model_name="glm-3-turbo"):
+ def __init__(self, **kwargs):
self.client = LocalLLM.RPCProxy("127.0.0.1", 7860)
def chat(self, system, history, gen_conf):
diff --git a/rag/llm/cv_model.py b/rag/llm/cv_model.py
index cb89509acbbb8759cbbe38b53c7f7679036a0978..61b942cdef72aec227d36f2fdad22889d081f5a6 100644
--- a/rag/llm/cv_model.py
+++ b/rag/llm/cv_model.py
@@ -67,8 +67,9 @@ class Base(ABC):
class GptV4(Base):
- def __init__(self, key, model_name="gpt-4-vision-preview", lang="Chinese"):
- self.client = OpenAI(api_key=key)
+ def __init__(self, key, model_name="gpt-4-vision-preview", lang="Chinese", base_url="https://api.openai.com/v1"):
+ if not base_url: base_url="https://api.openai.com/v1"
+ self.client = OpenAI(api_key=key, base_url=base_url)
self.model_name = model_name
self.lang = lang
@@ -84,7 +85,7 @@ class GptV4(Base):
class QWenCV(Base):
- def __init__(self, key, model_name="qwen-vl-chat-v1", lang="Chinese"):
+ def __init__(self, key, model_name="qwen-vl-chat-v1", lang="Chinese", **kwargs):
import dashscope
dashscope.api_key = key
self.model_name = model_name
@@ -123,7 +124,7 @@ class QWenCV(Base):
class Zhipu4V(Base):
- def __init__(self, key, model_name="glm-4v", lang="Chinese"):
+ def __init__(self, key, model_name="glm-4v", lang="Chinese", **kwargs):
self.client = ZhipuAI(api_key=key)
self.model_name = model_name
self.lang = lang
@@ -140,7 +141,7 @@ class Zhipu4V(Base):
class LocalCV(Base):
- def __init__(self, key, model_name="glm-4v", lang="Chinese"):
+ def __init__(self, key, model_name="glm-4v", lang="Chinese", **kwargs):
pass
def describe(self, image, max_tokens=1024):
diff --git a/rag/llm/embedding_model.py b/rag/llm/embedding_model.py
index 169a4df836f8328eedab6cc72c743030d0985798..06a573d5a47e0f707b4cabb6852bb770656be385 100644
--- a/rag/llm/embedding_model.py
+++ b/rag/llm/embedding_model.py
@@ -51,7 +51,7 @@ class Base(ABC):
class HuEmbedding(Base):
- def __init__(self, key="", model_name=""):
+ def __init__(self, **kwargs):
"""
If you have trouble downloading HuggingFace models, -_^ this might help!!
@@ -81,8 +81,9 @@ class HuEmbedding(Base):
class OpenAIEmbed(Base):
- def __init__(self, key, model_name="text-embedding-ada-002"):
- self.client = OpenAI(api_key=key)
+ def __init__(self, key, model_name="text-embedding-ada-002", base_url="https://api.openai.com/v1"):
+ if not base_url: base_url="https://api.openai.com/v1"
+ self.client = OpenAI(api_key=key, base_url=base_url)
self.model_name = model_name
def encode(self, texts: list, batch_size=32):
@@ -98,7 +99,7 @@ class OpenAIEmbed(Base):
class QWenEmbed(Base):
- def __init__(self, key, model_name="text_embedding_v2"):
+ def __init__(self, key, model_name="text_embedding_v2", **kwargs):
dashscope.api_key = key
self.model_name = model_name
@@ -131,7 +132,7 @@ class QWenEmbed(Base):
class ZhipuEmbed(Base):
- def __init__(self, key, model_name="embedding-2"):
+ def __init__(self, key, model_name="embedding-2", **kwargs):
self.client = ZhipuAI(api_key=key)
self.model_name = model_name
diff --git a/rag/svr/task_executor.py b/rag/svr/task_executor.py
index 517d8a2fe39a74ef5d256a0ed4c3334150650973..79186487d76efa0dc7e4fe00ecd163d7ea29bb52 100644
--- a/rag/svr/task_executor.py
+++ b/rag/svr/task_executor.py
@@ -280,4 +280,5 @@ if __name__ == "__main__":
from mpi4py import MPI
comm = MPI.COMM_WORLD
- main(int(sys.argv[2]), int(sys.argv[1]))
+ while True:
+ main(int(sys.argv[2]), int(sys.argv[1]))