support gpt-4o (#773)

### What problem does this PR solve?
#771 

### Type of change

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

api/apps/llm_app.py

 
         res = {}
         for m in llms:
-            if model_type and m["model_type"] != model_type:
+            if model_type and m["model_type"].find(model_type)<0:
                 continue
             if m["fid"] not in res:
                 res[m["fid"]] = []

api/db/init_data.py

     llm_infos = [
         # ---------------------- OpenAI ------------------------
         {
+            "fid": factory_infos[0]["name"],
+            "llm_name": "gpt-4o",
+            "tags": "LLM,CHAT,128K",
+            "max_tokens": 128000,
+            "model_type": LLMType.CHAT.value + "," + LLMType.IMAGE2TEXT.value
+        }, {
             "fid": factory_infos[0]["name"],
             "llm_name": "gpt-3.5-turbo",
             "tags": "LLM,CHAT,4K",

api/db/services/llm_service.py

         if not model_config:
             if llm_type == LLMType.EMBEDDING.value:
                 llm = LLMService.query(llm_name=llm_name)
-                if llm and llm[0].fid in ["Youdao", "FastEmbed"]:
+                if llm and llm[0].fid in ["Youdao", "FastEmbed", "DeepSeek"]:
                     model_config = {"llm_factory": llm[0].fid, "api_key":"", "llm_name": llm_name, "api_base": ""}
             if not model_config:
                 if llm_name == "flag-embedding":

api/settings.py

         "embedding_model": "",
         "image2text_model": "",
         "asr_model": "",
+    },
+    "DeepSeek": {
+        "chat_model": "deepseek-chat",
+        "embedding_model": "BAAI/bge-large-zh-v1.5",
+        "image2text_model": "",
+        "asr_model": "",
     }
 }
 LLM = get_base_config("user_default_llm", {})

rag/llm/__init__.py

     "Tongyi-Qianwen": DefaultEmbedding, #QWenEmbed,
     "ZHIPU-AI": ZhipuEmbed,
     "FastEmbed": FastEmbed,
-    "Youdao": YoudaoEmbed
+    "Youdao": YoudaoEmbed,
+    "DeepSeek": DefaultEmbedding
 }
 
 

rag/svr/task_executor.py

 
         st = timer()
         cks = build(r)
-        cron_logger.info("Build chunks({}): {}".format(r["name"], timer()-st))
+        cron_logger.info("Build chunks({}): {:.2f}".format(r["name"], timer()-st))
         if cks is None:
             continue
         if not cks:
             callback(-1, "Embedding error:{}".format(str(e)))
             cron_logger.error(str(e))
             tk_count = 0
-        cron_logger.info("Embedding elapsed({}): {}".format(r["name"], timer()-st))
+        cron_logger.info("Embedding elapsed({:.2f}): {}".format(r["name"], timer()-st))
 
         callback(msg="Finished embedding({:.2f})! Start to build index!".format(timer()-st))
         init_kb(r)
             if b % 128 == 0:
                 callback(prog=0.8 + 0.1 * (b + 1) / len(cks), msg="")
 
-        cron_logger.info("Indexing elapsed({}): {}".format(r["name"], timer()-st))
+        cron_logger.info("Indexing elapsed({}): {:.2f}".format(r["name"], timer()-st))
         if es_r:
             callback(-1, "Index failure!")
             ELASTICSEARCH.deleteByQuery(
             DocumentService.increment_chunk_num(
                 r["doc_id"], r["kb_id"], tk_count, chunk_count, 0)
             cron_logger.info(
-                "Chunk doc({}), token({}), chunks({}), elapsed:{}".format(
+                "Chunk doc({}), token({}), chunks({}), elapsed:{:.2f}".format(
                     r["id"], tk_count, len(cks), timer()-st))