Merge branch 'feat/model-runtime' into deploy/dev

api/core/model_runtime/model_providers/azure_openai/llm/llm.py

@@ -11,14 +11,13 @@
 from core.model_runtime.entities.llm_entities import LLMMode, LLMResult, \
     LLMResultChunk, LLMResultChunkDelta
 from core.model_runtime.entities.message_entities import PromptMessageTool, PromptMessage, AssistantPromptMessage, \
-    PromptMessageFunction, UserPromptMessage, PromptMessageContentType, ImagePromptMessageContent, \
+    UserPromptMessage, PromptMessageContentType, ImagePromptMessageContent, \
     TextPromptMessageContent, SystemPromptMessage, ToolPromptMessage
 from core.model_runtime.entities.model_entities import AIModelEntity, ModelPropertyKey
 from core.model_runtime.errors.validate import CredentialsValidateFailedError
 from core.model_runtime.model_providers.__base.large_language_model import LargeLanguageModel
 from core.model_runtime.model_providers.azure_openai._common import _CommonAzureOpenAI
 from core.model_runtime.model_providers.azure_openai._constant import LLM_BASE_MODELS, AzureBaseModel
-from core.model_runtime.utils import helper
 
 logger = logging.getLogger(__name__)
 

api/core/model_runtime/model_providers/huggingface_hub/huggingface_hub.yaml

@@ -1,6 +1,6 @@
 provider: huggingface_hub
 label:
-  en_US: HuggingFace Hub
+  en_US: Hugging Face Model
 icon_small:
     en_US: icon_s_en.svg
 icon_large:
@@ -25,7 +25,8 @@ model_credential_schema:
   credential_form_schemas:
   - variable: huggingfacehub_api_type
     label:
-      en_US: HuggingFace Hub API Type
+      en_US: Endpoint Type
+      zh_Hans: 端点类型
     type: radio
     required: true
     default: hosted_inference_api
@@ -38,36 +39,43 @@ model_credential_schema:
         en_US: Inference Endpoints
   - variable: huggingfacehub_api_token
     label:
-      en_US: HuggingFace Hub Access Token
+      en_US: API Token
+      zh_Hans: API Token
     type: secret-input
     required: true
     placeholder:
-      en_US: Enter your HuggingFace Hub Access Token
+      en_US: Enter your Hugging Face Hub API Token here
+      zh_Hans: 在此输入您的 Hugging Face Hub API Token
   - variable: huggingface_namespace
     label:
-      en_US: HuggingFace Namespace
+      en_US: 'User Name / Organization Name'
+      zh_Hans: '用户名 / 组织名称'
     type: text-input
     required: true
     placeholder:
-      en_US: Enter your HuggingFace Namespace
+      en_US: 'Enter your User Name / Organization Name here'
+      zh_Hans: '在此输入您的用户名 / 组织名称'
     show_on:
     - variable: __model_type
       value: text-embedding
     - variable: huggingfacehub_api_type
       value: inference_endpoints
   - variable: huggingfacehub_endpoint_url
     label:
-      en_US: HuggingFace Hub Endpoint URL
+      en_US: Endpoint URL
+      zh_Hans: 端点 URL
     type: text-input
     required: true
     placeholder:
-      en_US: Enter your HuggingFace Hub Endpoint URL
+      en_US: Enter your Endpoint URL here
+      zh_Hans: 在此输入您的端点 URL
     show_on:
     - variable: huggingfacehub_api_type
       value: inference_endpoints
   - variable: task_type
     label:
-      en_US: Task Type
+      en_US: Task
+      zh_Hans: Task
     type: select
     options:
     - value: text2text-generation
@@ -79,6 +87,7 @@ model_credential_schema:
     - value: text-generation
       label:
         en_US: Text Generation
+        zh_Hans: 文本生成
       show_on:
       - variable: __model_type
         value: llm

api/core/model_runtime/model_providers/huggingface_hub/llm/llm.py

@@ -136,10 +136,12 @@ def _get_customizable_model_parameter_rules() -> list[ParameterRule]:
             DefaultParameterName.TEMPERATURE).copy()
         temperature_rule_dict['name'] = 'temperature'
         temperature_rule = ParameterRule(**temperature_rule_dict)
+        temperature_rule.default = 0.5
 
         top_p_rule_dict = PARAMETER_RULE_TEMPLATE.get(DefaultParameterName.TOP_P).copy()
         top_p_rule_dict['name'] = 'top_p'
         top_p_rule = ParameterRule(**top_p_rule_dict)
+        top_p_rule.default = 0.5
 
         top_k_rule = ParameterRule(
             name='top_k',

api/core/model_runtime/model_providers/huggingface_hub/text_embedding/text_embedding.py

@@ -5,6 +5,8 @@
 from huggingface_hub import InferenceClient, HfApi
 from huggingface_hub.utils import HfHubHTTPError
 
+from core.model_runtime.entities.common_entities import I18nObject
+from core.model_runtime.entities.model_entities import AIModelEntity, FetchFrom, ModelType
 from core.model_runtime.entities.text_embedding_entities import TextEmbeddingResult, EmbeddingUsage
 from core.model_runtime.errors.invoke import InvokeError, InvokeBadRequestError
 from core.model_runtime.errors.validate import CredentialsValidateFailedError
@@ -62,38 +64,53 @@ def get_num_tokens(self, model: str, credentials: dict, texts: list[str]) -> int
         return 0
 
     def validate_credentials(self, model: str, credentials: dict) -> None:
-        if 'huggingfacehub_api_type' not in credentials:
-            raise CredentialsValidateFailedError('Huggingface Hub Endpoint Type must be provided.')
+        try:
+            if 'huggingfacehub_api_type' not in credentials:
+                raise CredentialsValidateFailedError('Huggingface Hub Endpoint Type must be provided.')
 
-        if 'huggingfacehub_api_token' not in credentials:
-            raise CredentialsValidateFailedError('Huggingface Hub Access Token must be provided.')
+            if 'huggingfacehub_api_token' not in credentials:
+                raise CredentialsValidateFailedError('Huggingface Hub API Token must be provided.')
 
-        if credentials['huggingfacehub_api_type'] == 'inference_endpoints':
-            if 'huggingface_namespace' not in credentials:
-                raise CredentialsValidateFailedError('Huggingface Hub Namespace must be provided.')
+            if credentials['huggingfacehub_api_type'] == 'inference_endpoints':
+                if 'huggingface_namespace' not in credentials:
+                    raise CredentialsValidateFailedError('Huggingface Hub User Name / Organization Name must be provided.')
 
-            if 'huggingfacehub_endpoint_url' not in credentials:
-                raise CredentialsValidateFailedError('Huggingface Hub Endpoint URL must be provided.')
+                if 'huggingfacehub_endpoint_url' not in credentials:
+                    raise CredentialsValidateFailedError('Huggingface Hub Endpoint URL must be provided.')
 
-            if 'task_type' not in credentials:
-                raise CredentialsValidateFailedError('Huggingface Hub Task Type must be provided.')
+                if 'task_type' not in credentials:
+                    raise CredentialsValidateFailedError('Huggingface Hub Task Type must be provided.')
 
-            if credentials['task_type'] != 'feature-extraction':
-                raise CredentialsValidateFailedError('Huggingface Hub Task Type is invalid.')
+                if credentials['task_type'] != 'feature-extraction':
+                    raise CredentialsValidateFailedError('Huggingface Hub Task Type is invalid.')
 
-            model = credentials['huggingfacehub_endpoint_url']
-        elif credentials['huggingfacehub_api_type'] == 'hosted_inference_api':
-            self._check_hosted_model_task_type(credentials['huggingfacehub_api_token'],
-                                               model)
-        else:
-            raise CredentialsValidateFailedError('Huggingface Hub Endpoint Type is invalid.')
+                model = credentials['huggingfacehub_endpoint_url']
+            elif credentials['huggingfacehub_api_type'] == 'hosted_inference_api':
+                self._check_hosted_model_task_type(credentials['huggingfacehub_api_token'],
+                                                   model)
+            else:
+                raise CredentialsValidateFailedError('Huggingface Hub Endpoint Type is invalid.')
 
-        client = InferenceClient(token=credentials['huggingfacehub_api_token'])
-        try:
+            client = InferenceClient(token=credentials['huggingfacehub_api_token'])
             client.feature_extraction(text='hello world', model=model)
         except Exception as ex:
             raise CredentialsValidateFailedError(str(ex))
 
+    def get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
+        entity = AIModelEntity(
+            model=model,
+            label=I18nObject(
+                en_US=model
+            ),
+            fetch_from=FetchFrom.CUSTOMIZABLE_MODEL,
+            model_type=ModelType.TEXT_EMBEDDING,
+            model_properties={
+                'context_size': 10000,
+                'max_chunks': 1
+            }
+        )
+        return entity
+
     @property
     def _invoke_error_mapping(self) -> dict[type[InvokeError], list[type[Exception]]]:
         return {

api/core/model_runtime/model_providers/replicate/text_embedding/text_embedding.py

@@ -4,6 +4,8 @@
 from replicate import Client as ReplicateClient
 from replicate.exceptions import ReplicateError, ModelError
 
+from core.model_runtime.entities.common_entities import I18nObject
+from core.model_runtime.entities.model_entities import AIModelEntity, FetchFrom, ModelType
 from core.model_runtime.entities.text_embedding_entities import TextEmbeddingResult, EmbeddingUsage
 from core.model_runtime.errors.invoke import InvokeError, InvokeBadRequestError
 from core.model_runtime.errors.validate import CredentialsValidateFailedError
@@ -66,6 +68,21 @@ def validate_credentials(self, model: str, credentials: dict) -> None:
         except Exception as e:
             raise CredentialsValidateFailedError(str(e))
 
+    def get_customizable_model_schema(self, model: str, credentials: dict) -> Optional[AIModelEntity]:
+        entity = AIModelEntity(
+            model=model,
+            label=I18nObject(
+                en_US=model
+            ),
+            fetch_from=FetchFrom.CUSTOMIZABLE_MODEL,
+            model_type=ModelType.TEXT_EMBEDDING,
+            model_properties={
+                'context_size': 4096,
+                'max_chunks': 1
+            }
+        )
+        return entity
+
     @property
     def _invoke_error_mapping(self) -> dict[type[InvokeError], list[type[Exception]]]:
         return {