add rerank model (#969)

### What problem does this PR solve?

feat: add rerank models to the project #724 #162

### Type of change

- [x] New Feature (non-breaking change which adds functionality)

api/apps/chunk_app.py

@@ -257,8 +257,15 @@ def retrieval_test():
 
         embd_mdl = TenantLLMService.model_instance(
             kb.tenant_id, LLMType.EMBEDDING.value, llm_name=kb.embd_id)
-        ranks = retrievaler.retrieval(question, embd_mdl, kb.tenant_id, [kb_id], page, size, similarity_threshold,
-                                      vector_similarity_weight, top, doc_ids)
+
+        rerank_mdl = None
+        if req.get("rerank_id"):
+            rerank_mdl = TenantLLMService.model_instance(
+                kb.tenant_id, LLMType.RERANK.value, llm_name=req["rerank_id"])
+
+        ranks = retrievaler.retrieval(question, embd_mdl, kb.tenant_id, [kb_id], page, size,
+                                      similarity_threshold, vector_similarity_weight, top,
+                                      doc_ids, rerank_mdl=rerank_mdl)
         for c in ranks["chunks"]:
             if "vector" in c:
                 del c["vector"]

api/apps/dialog_app.py

@@ -33,6 +33,9 @@ def set_dialog():
     name = req.get("name", "New Dialog")
     description = req.get("description", "A helpful Dialog")
     top_n = req.get("top_n", 6)
+    top_k = req.get("top_k", 1024)
+    rerank_id = req.get("rerank_id", "")
+    if not rerank_id: req["rerank_id"] = ""
     similarity_threshold = req.get("similarity_threshold", 0.1)
     vector_similarity_weight = req.get("vector_similarity_weight", 0.3)
     llm_setting = req.get("llm_setting", {})
@@ -83,6 +86,8 @@ def set_dialog():
                 "llm_setting": llm_setting,
                 "prompt_config": prompt_config,
                 "top_n": top_n,
+                "top_k": top_k,
+                "rerank_id": rerank_id,
                 "similarity_threshold": similarity_threshold,
                 "vector_similarity_weight": vector_similarity_weight
             }

api/apps/llm_app.py

@@ -20,7 +20,7 @@ from api.utils.api_utils import server_error_response, get_data_error_result, va
 from api.db import StatusEnum, LLMType
 from api.db.db_models import TenantLLM
 from api.utils.api_utils import get_json_result
-from rag.llm import EmbeddingModel, ChatModel
+from rag.llm import EmbeddingModel, ChatModel, RerankModel
 
 
 @manager.route('/factories', methods=['GET'])
@@ -28,7 +28,7 @@ from rag.llm import EmbeddingModel, ChatModel
 def factories():
     try:
         fac = LLMFactoriesService.get_all()
-        return get_json_result(data=[f.to_dict() for f in fac if f.name not in ["Youdao", "FastEmbed"]])
+        return get_json_result(data=[f.to_dict() for f in fac if f.name not in ["Youdao", "FastEmbed", "BAAI"]])
     except Exception as e:
         return server_error_response(e)
 
@@ -64,6 +64,16 @@ def set_api_key():
             except Exception as e:
                 msg += f"\nFail to access model({llm.llm_name}) using this api key." + str(
                     e)
+        elif llm.model_type == LLMType.RERANK:
+            mdl = RerankModel[factory](
+                req["api_key"], llm.llm_name, base_url=req.get("base_url"))
+            try:
+                m, tc = mdl.similarity("What's the weather?", ["Is it sunny today?"])
+                if len(arr[0]) == 0 or tc == 0:
+                    raise Exception("Fail")
+            except Exception as e:
+                msg += f"\nFail to access model({llm.llm_name}) using this api key." + str(
+                    e)
 
     if msg:
         return get_data_error_result(retmsg=msg)
@@ -199,7 +209,7 @@ def list_app():
         llms = [m.to_dict()
                 for m in llms if m.status == StatusEnum.VALID.value]
         for m in llms:
-            m["available"] = m["fid"] in facts or m["llm_name"].lower() == "flag-embedding" or m["fid"] in ["Youdao","FastEmbed"]
+            m["available"] = m["fid"] in facts or m["llm_name"].lower() == "flag-embedding" or m["fid"] in ["Youdao","FastEmbed", "BAAI"]
 
         llm_set = set([m["llm_name"] for m in llms])
         for o in objs:

api/apps/user_app.py

@@ -26,8 +26,9 @@ from api.db.services.llm_service import TenantLLMService, LLMService
 from api.utils.api_utils import server_error_response, validate_request
 from api.utils import get_uuid, get_format_time, decrypt, download_img, current_timestamp, datetime_format
 from api.db import UserTenantRole, LLMType, FileType
-from api.settings import RetCode, GITHUB_OAUTH, FEISHU_OAUTH, CHAT_MDL, EMBEDDING_MDL, ASR_MDL, IMAGE2TEXT_MDL, PARSERS, API_KEY, \
-    LLM_FACTORY, LLM_BASE_URL
+from api.settings import RetCode, GITHUB_OAUTH, FEISHU_OAUTH, CHAT_MDL, EMBEDDING_MDL, ASR_MDL, IMAGE2TEXT_MDL, PARSERS, \
+    API_KEY, \
+    LLM_FACTORY, LLM_BASE_URL, RERANK_MDL
 from api.db.services.user_service import UserService, TenantService, UserTenantService
 from api.db.services.file_service import FileService
 from api.settings import stat_logger
@@ -288,7 +289,8 @@ def user_register(user_id, user):
         "embd_id": EMBEDDING_MDL,
         "asr_id": ASR_MDL,
         "parser_ids": PARSERS,
-        "img2txt_id": IMAGE2TEXT_MDL
+        "img2txt_id": IMAGE2TEXT_MDL,
+        "rerank_id": RERANK_MDL
     }
     usr_tenant = {
         "tenant_id": user_id,