use self hosted embedding

app.py

@@ -1,5 +1,5 @@
 import gradio as gr
-import llama_cpp
+# import llama_cpp
 import base64
 from Crypto.Cipher import AES
 from Crypto.Util.Padding import unpad
@@ -20,14 +20,21 @@ def decrypt_file(input_path, key):
     return plaintext.decode("utf-8")
 
 
-llm = llama_cpp.Llama.from_pretrained(
-    repo_id="mradermacher/bge-large-zh-v1.5-GGUF",
-    filename="bge-large-zh-v1.5.Q4_K_M.gguf",
-    embedding=True,
-)
+# llm = llama_cpp.Llama.from_pretrained(
+#     repo_id="mradermacher/bge-large-zh-v1.5-GGUF",
+#     filename="bge-large-zh-v1.5.Q4_K_M.gguf",
+#     embedding=True,
+# )
 
 # embedding_1 = llm.create_embedding("Hello, world!")
 # embedding_2 = llm.create_embedding("你好, 世界!") # type(embedding_1['data'][0]['embedding']) list
+from openai import OpenAI
+
+client = OpenAI(
+    api_key=os.getenv("DASHSCOPE_API_KEY"),  # 如果您没有配置环境变量，请在此处用您的API Key进行替换
+    base_url="https://dashscope.aliyuncs.com/compatible-mode/v1"  # 百炼服务的base_url
+)
+
 
 from pymilvus import MilvusClient
 
@@ -92,10 +99,30 @@ for vhjx_index, vhjx_item in enumerate(raw_jsons):
 
 
 def greet(name):
-    embeddings = llm.create_embedding(name)
+    """
+    Search for relevant critical commentary entries based on an input query from the Analects.
+
+    This function parses the input query, performs a fuzzy search in the indexed original text field,
+    and extracts related critiques. 
+
+    Args:
+        query (str): The input text (a line from the Analects, possibly fuzzy or partial) to search.
+
+    Returns:
+        List[dict]: A list of result entries. Each entry contains the original hit and a list of related entries
+                    under the key "extra", retrieved via index references mentioned in the commentary.
+    """
+    # embeddings = llm.create_embedding(name)
+    completion = client.embeddings.create(
+            model="text-embedding-v3",
+            input=name,
+            dimensions=1024,  # 仅 text-embedding-v3 支持
+            encoding_format="float"
+        )
     res = client.search(
         collection_name="collection_1",
-        data=[embeddings["data"][0]["embedding"]],
+        # data=[embeddings["data"][0]["embedding"]],
+        data=[completion.data[0].embedding],
         limit=5,
         output_fields=["index", "text", "annotation", "critique"],
     )

embeddings.json

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:caa8a181528c6921eadc4ee98d21126bbe4a8c081426ae40ef039e31733dd87b
-size 10926548
+oid sha256:e879dc69d30510924d7e26d5ad58ce193e9901003a0c634c94ac025b74655819
+size 11621667

requirements.txt

@@ -2,6 +2,5 @@ whoosh-reloaded==2.7.5
 jieba==0.42.1
 huggingface-hub
 pymilvus
---extra-index-url https://abetlen.github.io/llama-cpp-python/whl/cpu
-llama-cpp-python
+openai
 pycryptodome