add support for Tencent Hunyuan

api/apps/llm_app.py

@@ -106,7 +106,7 @@ def set_api_key():
 
 @manager.route('/add_llm', methods=['POST'])
 @login_required
-@validate_request("llm_factory", "llm_name", "model_type")
+@validate_request("llm_factory")
 def add_llm():
     req = request.json
     factory = req["llm_factory"]
@@ -120,6 +120,11 @@ def add_llm():
         api_key = '{' + f'"volc_ak": "{req.get("volc_ak", "")}", ' \
                         f'"volc_sk": "{req.get("volc_sk", "")}", ' \
                         f'"ep_id": "{endpoint_id}", ' + '}'
+    elif factory == "Tencent Hunyuan":
+        api_key = '{' + f'"hunyuan_sid": "{req.get("hunyuan_sid", "")}", ' \
+                        f'"hunyuan_sk": "{req.get("hunyuan_sk", "")}"' + '}'
+        req["api_key"] = api_key
+        return set_api_key()
     elif factory == "Bedrock":
         # For Bedrock, due to its special authentication method
         # Assemble bedrock_ak, bedrock_sk, bedrock_region
@@ -132,7 +137,7 @@ def add_llm():
         api_key = "xxxxxxxxxxxxxxx"
     elif factory == "OpenAI-API-Compatible":
         llm_name = req["llm_name"]+"___OpenAI-API"
-        api_key = req.get("api_key","xxxxxxxxxxxxxxx") 
+        api_key = req.get("api_key","xxxxxxxxxxxxxxx")
     else:
         llm_name = req["llm_name"]
         api_key = req.get("api_key","xxxxxxxxxxxxxxx") 

conf/llm_factories.json

@@ -3120,6 +3120,44 @@
             "tags": "LLM,TEXT EMBEDDING",
             "status": "1",
             "llm": []
+        },
+        {
+            "name": "Tencent Hunyuan",
+            "logo": "",
+            "tags": "LLM,IMAGE2TEXT",
+            "status": "1",
+            "llm": [
+                {
+                    "llm_name": "hunyuan-pro",
+                    "tags": "LLM,CHAT,32k",
+                    "max_tokens": 32768,
+                    "model_type": "chat"
+                },
+                {
+                    "llm_name": "hunyuan-standard",
+                    "tags": "LLM,CHAT,32k",
+                    "max_tokens": 32768,
+                    "model_type": "chat"
+                },
+                {
+                    "llm_name": "hunyuan-standard-256K",
+                    "tags": "LLM,CHAT,256k",
+                    "max_tokens": 262144,
+                    "model_type": "chat"
+                },
+                {
+                    "llm_name": "hunyuan-lite",
+                    "tags": "LLM,CHAT,256k",
+                    "max_tokens": 262144,
+                    "model_type": "chat"
+                },
+                {
+                    "llm_name": "hunyuan-vision",
+                    "tags": "LLM,IMAGE2TEXT,8k",
+                    "max_tokens": 8192,
+                    "model_type": "image2text"
+                }
+            ]
         }
     ]
 }

rag/llm/__init__.py

@@ -63,7 +63,8 @@
     "StepFun":StepFunCV,
     "OpenAI-API-Compatible": OpenAI_APICV,
     "TogetherAI": TogetherAICV,
-    "01.AI": YiCV
+    "01.AI": YiCV,
+    "Tencent Hunyuan": HunyuanCV
 }
 
 
@@ -98,7 +99,8 @@
     "novita.ai": NovitaAIChat,
     "SILICONFLOW": SILICONFLOWChat,
     "01.AI": YiChat,
-    "Replicate": ReplicateChat
+    "Replicate": ReplicateChat,
+    "Tencent Hunyuan": HunyuanChat
 }
 
 

rag/llm/chat_model.py

@@ -1088,3 +1088,83 @@ def chat_streamly(self, system, history, gen_conf):
             yield ans + "\n**ERROR**: " + str(e)
 
         yield num_tokens_from_string(ans)
+
+
+class HunyuanChat(Base):
+    def __init__(self, key, model_name, base_url=None):
+        from tencentcloud.common import credential
+        from tencentcloud.hunyuan.v20230901 import hunyuan_client
+
+        key = json.loads(key)
+        sid = key.get("hunyuan_sid", "")
+        sk = key.get("hunyuan_sk", "")
+        cred = credential.Credential(sid, sk)
+        self.model_name = model_name
+        self.client = hunyuan_client.HunyuanClient(cred, "")
+
+    def chat(self, system, history, gen_conf):
+        from tencentcloud.hunyuan.v20230901 import models
+        from tencentcloud.common.exception.tencent_cloud_sdk_exception import (
+            TencentCloudSDKException,
+        )
+
+        _gen_conf = {}
+        _history = [{k.capitalize(): v for k, v in item.items() } for item in history]
+        if system:
+            _history.insert(0, {"Role": "system", "Content": system})
+        if "temperature" in gen_conf:
+            _gen_conf["Temperature"] = gen_conf["temperature"]
+        if "top_p" in gen_conf:
+            _gen_conf["TopP"] = gen_conf["top_p"]
+
+        req = models.ChatCompletionsRequest()
+        params = {"Model": self.model_name, "Messages": _history, **_gen_conf}
+        req.from_json_string(json.dumps(params))
+        ans = ""
+        try:
+            response = self.client.ChatCompletions(req)
+            ans = response.Choices[0].Message.Content
+            return ans, response.Usage.TotalTokens
+        except TencentCloudSDKException as e:
+            return ans + "\n**ERROR**: " + str(e), 0
+
+    def chat_streamly(self, system, history, gen_conf):
+        from tencentcloud.hunyuan.v20230901 import models
+        from tencentcloud.common.exception.tencent_cloud_sdk_exception import (
+            TencentCloudSDKException,
+        )
+
+        _gen_conf = {}
+        _history = [{k.capitalize(): v for k, v in item.items() } for item in history]
+        if system:
+            _history.insert(0, {"Role": "system", "Content": system})
+
+        if "temperature" in gen_conf:
+            _gen_conf["Temperature"] = gen_conf["temperature"]
+        if "top_p" in gen_conf:
+            _gen_conf["TopP"] = gen_conf["top_p"]
+        req = models.ChatCompletionsRequest()
+        params = {
+            "Model": self.model_name,
+            "Messages": _history,
+            "Stream": True,
+            **_gen_conf,
+        }
+        req.from_json_string(json.dumps(params))
+        ans = ""
+        total_tokens = 0
+        try:
+            response = self.client.ChatCompletions(req)
+            for resp in response:
+                resp = json.loads(resp["data"])
+                if not resp["Choices"] or not resp["Choices"][0]["Delta"]["Content"]:
+                    continue
+                ans += resp["Choices"][0]["Delta"]["Content"]
+                total_tokens += 1
+
+                yield ans
+
+        except TencentCloudSDKException as e:
+            yield ans + "\n**ERROR**: " + str(e)
+
+        yield total_tokens

rag/llm/cv_model.py

@@ -664,4 +664,56 @@ class YiCV(GptV4):
     def __init__(self, key, model_name, lang="Chinese",base_url="https://api.lingyiwanwu.com/v1",):
         if not base_url:
             base_url = "https://api.lingyiwanwu.com/v1"
-        super().__init__(key, model_name,lang,base_url)
+        super().__init__(key, model_name,lang,base_url)
+
+
+class HunyuanCV(Base):
+    def __init__(self, key, model_name, lang="Chinese",base_url=None):
+        from tencentcloud.common import credential
+        from tencentcloud.hunyuan.v20230901 import hunyuan_client
+
+        key = json.loads(key)
+        sid = key.get("hunyuan_sid", "")
+        sk = key.get("hunyuan_sk", "")
+        cred = credential.Credential(sid, sk)
+        self.model_name = model_name
+        self.client = hunyuan_client.HunyuanClient(cred, "")
+        self.lang = lang
+
+    def describe(self, image, max_tokens=4096):
+        from tencentcloud.hunyuan.v20230901 import models
+        from tencentcloud.common.exception.tencent_cloud_sdk_exception import (
+            TencentCloudSDKException,
+        )
+
+        b64 = self.image2base64(image)
+        req = models.ChatCompletionsRequest()
+        params = {"Model": self.model_name, "Messages": self.prompt(b64)}
+        req.from_json_string(json.dumps(params))
+        ans = ""
+        try:
+            response = self.client.ChatCompletions(req)
+            ans = response.Choices[0].Message.Content
+            return ans, response.Usage.TotalTokens
+        except TencentCloudSDKException as e:
+            return ans + "\n**ERROR**: " + str(e), 0
+
+    def prompt(self, b64):
+        return [
+            {
+                "Role": "user",
+                "Contents": [
+                    {
+                        "Type": "image_url",
+                        "ImageUrl": {
+                            "Url": f"data:image/jpeg;base64,{b64}"
+                        },
+                    },
+                    {
+                        "Type": "text",
+                        "Text": "请用中文详细描述一下图中的内容，比如时间，地点，人物，事情，人物心情等，如果有数据请提取出数据。" if self.lang.lower() == "chinese" else
+                        "Please describe the content of this picture, like where, when, who, what happen. If it has number data, please extract them out.",
+                    },
+                ],
+            }
+        ]

requirements.txt

@@ -76,6 +76,7 @@ Shapely==2.0.5
 six==1.16.0
 StrEnum==0.4.15
 tabulate==0.9.0
+tencentcloud-sdk-python==3.0.1215
 tika==2.6.0
 tiktoken==0.6.0
 torch==2.3.0

requirements_arm.txt

@@ -115,6 +115,7 @@ six==1.16.0
 sniffio==1.3.1
 StrEnum==0.4.15
 sympy==1.12
+tencentcloud-sdk-python==3.0.1215
 threadpoolctl==3.3.0
 tika==2.6.0
 tiktoken==0.6.0

web/src/locales/en.ts

@@ -521,6 +521,10 @@ The above is the content you need to summarize.`,
       'eu-central-1': 'Europe (Frankfurt)',
       'us-gov-west-1': 'AWS GovCloud (US-West)',
       'ap-southeast-2': 'Asia Pacific (Sydney)',
+      addHunyuanSID: 'Hunyuan Secret ID',
+      HunyuanSIDMessage: 'Please input your Secret ID',
+      addHunyuanSK: 'Hunyuan Secret Key',
+      HunyuanSKMessage: 'Please input your Secret Key',
     },
     message: {
       registered: 'Registered!',

web/src/locales/zh-traditional.ts

@@ -484,6 +484,10 @@ export default {
       'eu-central-1': '歐洲 (法蘭克福)',
       'us-gov-west-1': 'AWS GovCloud (US-West)',
       'ap-southeast-2': '亞太地區 (雪梨)',
+      addHunyuanSID: '混元 Secret ID',
+      HunyuanSIDMessage: '請輸入 Secret ID',
+      addHunyuanSK: '混元 Secret Key',
+      HunyuanSKMessage: '請輸入 Secret Key',
     },
     message: {
       registered: '註冊成功',

web/src/locales/zh.ts

@@ -501,6 +501,10 @@ export default {
       'eu-central-1': '欧洲 (法兰克福)',
       'us-gov-west-1': 'AWS GovCloud (US-West)',
       'ap-southeast-2': '亚太地区 (悉尼)',
+      addHunyuanSID: '混元 Secret ID',
+      HunyuanSIDMessage: '请输入 Secret ID',
+      addHunyuanSK: '混元 Secret Key',
+      HunyuanSKMessage: '请输入 Secret Key',
     },
     message: {
       registered: '注册成功',

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

@@ -30,8 +30,9 @@ export const IconMap = {
   Upstage: 'upstage',
   'novita.ai': 'novita-ai',
   SILICONFLOW: 'siliconflow',
-  "01.AI": 'yi',
-  "Replicate": 'replicate'
+  '01.AI': 'yi',
+  Replicate: 'replicate',
+  'Tencent Hunyuan': 'hunyuan',
 };
 
 export const BedrockRegionList = [

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

@@ -163,6 +163,33 @@ export const useSubmitVolcEngine = () => {
   };
 };
 
+export const useSubmitHunyuan = () => {
+  const { addLlm, loading } = useAddLlm();
+  const {
+    visible: HunyuanAddingVisible,
+    hideModal: hideHunyuanAddingModal,
+    showModal: showHunyuanAddingModal,
+  } = useSetModalState();
+
+  const onHunyuanAddingOk = useCallback(
+    async (payload: IAddLlmRequestBody) => {
+      const ret = await addLlm(payload);
+      if (ret === 0) {
+        hideHunyuanAddingModal();
+      }
+    },
+    [hideHunyuanAddingModal, addLlm],
+  );
+
+  return {
+    HunyuanAddingLoading: loading,
+    onHunyuanAddingOk,
+    HunyuanAddingVisible,
+    hideHunyuanAddingModal,
+    showHunyuanAddingModal,
+  };
+};
+
 export const useSubmitBedrock = () => {
   const { addLlm, loading } = useAddLlm();
   const {