add support for PerfXCloud (#1883)

### What problem does this PR solve?

#1853 add support for PerfXCloud

### Type of change


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

Co-authored-by: Zhedong Cen <[email protected]>

conf/llm_factories.json

@@ -2442,6 +2442,158 @@
                     "model_type": "chat"
                 }
             ]
-        }
+        },
+        {
+            "name": "PerfXCloud",
+            "logo": "",
+            "tags": "LLM,TEXT EMBEDDING",
+            "status": "1",
+            "llm": [
+                {
+                    "llm_name": "deepseek-v2-chat",
+                    "tags": "LLM,CHAT,4k",
+                    "max_tokens": 4096,
+                    "model_type": "chat"
+                },
+                {
+                    "llm_name": "llama3.1:405b",
+                    "tags": "LLM,CHAT,128k",
+                    "max_tokens": 131072,
+                    "model_type": "chat"
+                },
+                {
+                    "llm_name": "Qwen2-72B-Instruct",
+                    "tags": "LLM,CHAT,128k",
+                    "max_tokens": 131072,
+                    "model_type": "chat"
+                },
+                {
+                    "llm_name": "Qwen2-72B-Instruct-GPTQ-Int4",
+                    "tags": "LLM,CHAT,2k",
+                    "max_tokens": 2048,
+                    "model_type": "chat"
+                },
+                {
+                    "llm_name": "Qwen2-72B-Instruct-awq-int4",
+                    "tags": "LLM,CHAT,32k",
+                    "max_tokens": 32768,
+                    "model_type": "chat"
+                },
+                {
+                    "llm_name": "Llama3-Chinese_v2",
+                    "tags": "LLM,CHAT,8k",
+                    "max_tokens": 8192,
+                    "model_type": "chat"
+                },
+                {
+                    "llm_name": "Yi-1_5-9B-Chat-16K",
+                    "tags": "LLM,CHAT,16k",
+                    "max_tokens": 16384,
+                    "model_type": "chat"
+                },
+                {
+                    "llm_name": "Qwen1.5-72B-Chat-GPTQ-Int4",
+                    "tags": "LLM,CHAT,2k",
+                    "max_tokens": 2048,
+                    "model_type": "chat"
+                },
+                {
+                    "llm_name": "Meta-Llama-3.1-8B-Instruct",
+                    "tags": "LLM,CHAT,4k",
+                    "max_tokens": 4096,
+                    "model_type": "chat"
+                },
+                {
+                    "llm_name": "Qwen2-7B-Instruct",
+                    "tags": "LLM,CHAT,32k",
+                    "max_tokens": 32768,
+                    "model_type": "chat"
+                },
+                {
+                    "llm_name": "deepseek-v2-lite-chat",
+                    "tags": "LLM,CHAT,2k",
+                    "max_tokens": 2048,
+                    "model_type": "chat"
+                },
+                {
+                    "llm_name": "Qwen2-7B",
+                    "tags": "LLM,CHAT,128k",
+                    "max_tokens": 131072,
+                    "model_type": "chat"
+                },
+                {
+                    "llm_name": "chatglm3-6b",
+                    "tags": "LLM,CHAT,8k",
+                    "max_tokens": 8192,
+                    "model_type": "chat"
+                },
+                {
+                    "llm_name": "Meta-Llama-3-70B-Instruct-GPTQ-Int4",
+                    "tags": "LLM,CHAT,1k",
+                    "max_tokens": 1024,
+                    "model_type": "chat"
+                },
+                {
+                    "llm_name": "Meta-Llama-3-8B-Instruct",
+                    "tags": "LLM,CHAT,8k",
+                    "max_tokens": 8192,
+                    "model_type": "chat"
+                },
+                {
+                    "llm_name": "Mistral-7B-Instruct",
+                    "tags": "LLM,CHAT,32k",
+                    "max_tokens": 32768,
+                    "model_type": "chat"
+                },
+                {
+                    "llm_name": "MindChat-Qwen-7B-v2",
+                    "tags": "LLM,CHAT,2k",
+                    "max_tokens": 2048,
+                    "model_type": "chat"
+                },
+                {
+                    "llm_name": "phi-2",
+                    "tags": "LLM,CHAT,2k",
+                    "max_tokens": 2048,
+                    "model_type": "chat"
+                },
+                {
+                    "llm_name": "SOLAR-10_7B-Instruct",
+                    "tags": "LLM,CHAT,4k",
+                    "max_tokens": 4096,
+                    "model_type": "chat"
+                },
+                {
+                    "llm_name": "Mixtral-8x7B-Instruct-v0.1-GPTQ",
+                    "tags": "LLM,CHAT,32k",
+                    "max_tokens": 32768,
+                    "model_type": "chat"
+                },
+                {
+                    "llm_name": "Qwen1.5-7B",
+                    "tags": "LLM,CHAT,32k",
+                    "max_tokens": 32768,
+                    "model_type": "chat"
+                },
+                {
+                    "llm_name": "BAAI/bge-large-en-v1.5",
+                    "tags": "TEXT EMBEDDING",
+                    "max_tokens": 512,
+                    "model_type": "embedding"
+                },
+                {
+                    "llm_name": "BAAI/bge-large-zh-v1.5",
+                    "tags": "TEXT EMBEDDING",
+                    "max_tokens": 1024,
+                    "model_type": "embedding"
+                },
+                {
+                    "llm_name": "BAAI/bge-m3",
+                    "tags": "TEXT EMBEDDING",
+                    "max_tokens": 8192,
+                    "model_type": "embedding"
+                }
+            ]
+        }  
     ]
 }

rag/llm/__init__.py

@@ -38,7 +38,8 @@ EmbeddingModel = {
     "NVIDIA": NvidiaEmbed,
     "LM-Studio": LmStudioEmbed,
     "OpenAI-API-Compatible": OpenAI_APIEmbed,
-    "cohere": CoHereEmbed
+    "cohere": CoHereEmbed,
+    "PerfXCloud": PerfXCloudEmbed,
 }
 
 
@@ -84,7 +85,8 @@ ChatModel = {
     "LM-Studio": LmStudioChat,
     "OpenAI-API-Compatible": OpenAI_APIChat,
     "cohere": CoHereChat,
-    "LeptonAI": LeptonAIChat
+    "LeptonAI": LeptonAIChat,
+    "PerfXCloud": PerfXCloudChat
 }
 
 

rag/llm/chat_model.py

@@ -987,4 +987,11 @@ class LeptonAIChat(Base):
     def __init__(self, key, model_name, base_url=None):
         if not base_url:
             base_url = os.path.join("https://"+model_name+".lepton.run","api","v1")
-        super().__init__(key, model_name, base_url)
+        super().__init__(key, model_name, base_url)
+
+
+class PerfXCloudChat(Base):
+    def __init__(self, key, model_name, base_url="https://cloud.perfxlab.cn/v1"):
+        if not base_url:
+            base_url = "https://cloud.perfxlab.cn/v1"
+        super().__init__(key, model_name, base_url)

rag/llm/embedding_model.py

@@ -553,3 +553,10 @@ class CoHereEmbed(Base):
         return np.array([d for d in res.embeddings.float]), int(
             res.meta.billed_units.input_tokens
         )
+
+
+class PerfXCloudEmbed(OpenAIEmbed):
+    def __init__(self, key, model_name, base_url="https://cloud.perfxlab.cn/v1"):
+        if not base_url:
+            base_url = "https://cloud.perfxlab.cn/v1"
+        super().__init__(key, model_name, base_url)

web/src/pages/user-setting/setting-model/constant.ts

@@ -25,6 +25,7 @@ export const IconMap = {
   'OpenAI-API-Compatible': 'openai-api',
   cohere: 'cohere',
   Lepton: 'lepton',
+  PerfXCould: 'perfx-could'
 };
 
 export const BedrockRegionList = [