update model_config.py

configs/model_config.py

@@ -27,7 +27,6 @@ EMBEDDING_MODEL = "text2vec"
 # Embedding running device
 EMBEDDING_DEVICE = "cuda" if torch.cuda.is_available() else "mps" if torch.backends.mps.is_available() else "cpu"
 
-
 # supported LLM models
 # llm_model_dict 处理了loader的一些预设行为，如加载位置，模型名称，模型处理器实例
 # 在以下字典中修改属性值，以指定本地 LLM 模型存储位置
@@ -93,13 +92,13 @@ llm_model_dict = {
         "pretrained_model_name": "vicuna-13b-hf",
         "local_model_path": None,
         "provides": "LLamaLLM"
-    },    
+    },
     # 直接调用返回requests.exceptions.ConnectionError错误，需要通过huggingface_hub包里的snapshot_download函数
     # 下载模型，如果snapshot_download还是返回网络错误，多试几次，一般是可以的，
     # 如果仍然不行，则应该是网络加了防火墙(在服务器上这种情况比较常见)，基本只能从别的设备上下载，
     # 然后转移到目标设备了.
-    "bloomz-7b1":{
-        "name" : "bloomz-7b1",
+    "bloomz-7b1": {
+        "name": "bloomz-7b1",
         "pretrained_model_name": "bigscience/bloomz-7b1",
         "local_model_path": None,
         "provides": "MOSSLLM"
@@ -107,21 +106,21 @@ llm_model_dict = {
     },
     # 实测加载bigscience/bloom-3b需要170秒左右，暂不清楚为什么这么慢
     # 应与它要加载专有token有关
-    "bloom-3b":{
-        "name" : "bloom-3b",
+    "bloom-3b": {
+        "name": "bloom-3b",
         "pretrained_model_name": "bigscience/bloom-3b",
         "local_model_path": None,
         "provides": "MOSSLLM"
 
-    },   
-    "baichuan-7b":{
-        "name":"baichuan-7b",
-        "pretrained_model_name":"baichuan-inc/baichuan-7B",
-        "local_model_path":None,
-        "provides":"MOSSLLM"
-    }, 
+    },
+    "baichuan-7b": {
+        "name": "baichuan-7b",
+        "pretrained_model_name": "baichuan-inc/baichuan-7B",
+        "local_model_path": None,
+        "provides": "MOSSLLM"
+    },
     # llama-cpp模型的兼容性问题参考https://github.com/abetlen/llama-cpp-python/issues/204
-    "ggml-vicuna-13b-1.1-q5":{
+    "ggml-vicuna-13b-1.1-q5": {
         "name": "ggml-vicuna-13b-1.1-q5",
         "pretrained_model_name": "lmsys/vicuna-13b-delta-v1.1",
         # 这里需要下载好模型的路径,如果下载模型是默认路径则它会下载到用户工作区的
@@ -131,7 +130,7 @@ llm_model_dict = {
         # 此外不同时期的ggml格式并不兼容，因此不同时期的ggml需要安装不同的llama-cpp-python库，且实测pip install 不好使
         # 需要手动从https://github.com/abetlen/llama-cpp-python/releases/tag/下载对应的wheel安装
         # 实测v0.1.63与本模型的vicuna/ggml-vicuna-13b-1.1/ggml-vic13b-q5_1.bin可以兼容
-        "local_model_path":f'''{"/".join(os.path.abspath(__file__).split("/")[:3])}/.cache/huggingface/hub/models--vicuna--ggml-vicuna-13b-1.1/blobs/''',
+        "local_model_path": f'''{"/".join(os.path.abspath(__file__).split("/")[:3])}/.cache/huggingface/hub/models--vicuna--ggml-vicuna-13b-1.1/blobs/''',
         "provides": "LLamaLLM"
     },
 
@@ -161,10 +160,10 @@ llm_model_dict = {
         "api_base_url": "http://localhost:8000/v1",  # "name"修改为fastchat服务中的"api_base_url"
         "api_key": "EMPTY"
     },
-    "openai-chatgpt-3.5":{
+    "openai-chatgpt-3.5": {
         "name": "gpt-3.5-turbo",
         "pretrained_model_name": "gpt-3.5-turbo",
-        "provides":"FastChatOpenAILLM",
+        "provides": "FastChatOpenAILLM",
         "local_model_path": None,
         "api_base_url": "https://api.openapi.com/v1",
         "api_key": ""