AI/dify
1
0
Fork 0
mirror of https://git.mirrors.martin98.com/https://github.com/langgenius/dify.git synced 2025-08-15 21:15:55 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix: visioin model always with low quality (#5253)

Browse Source
This commit is contained in:
rerorero 2024-06-16 10:46:17 +09:00 committed by GitHub
parent 37c87164dc
commit 3a423e8ce7
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 10 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
api/core/workflow/nodes/llm/llm_node.py
View File

@ -12,7 +12,11 @@ from core.file.file_obj import FileVar
from core.memory.token_buffer_memory import TokenBufferMemory from core.memory.token_buffer_memory import TokenBufferMemory
from core.model_manager import ModelInstance, ModelManager from core.model_manager import ModelInstance, ModelManager
from core.model_runtime.entities.llm_entities import LLMUsage from core.model_runtime.entities.llm_entities import LLMUsage
from core.model_runtime.entities.message_entities import PromptMessage, PromptMessageContentType from core.model_runtime.entities.message_entities import (
ImagePromptMessageContent,
PromptMessage,
PromptMessageContentType,
)
from core.model_runtime.entities.model_entities import ModelType from core.model_runtime.entities.model_entities import ModelType
from core.model_runtime.model_providers.__base.large_language_model import LargeLanguageModel from core.model_runtime.model_providers.__base.large_language_model import LargeLanguageModel
from core.model_runtime.utils.encoders import jsonable_encoder from core.model_runtime.utils.encoders import jsonable_encoder
@ -548,6 +552,7 @@ class LLMNode(BaseNode):
stop = model_config.stop stop = model_config.stop
vision_enabled = node_data.vision.enabled vision_enabled = node_data.vision.enabled
vision_detail = node_data.vision.configs.detail if node_data.vision.configs else None
filtered_prompt_messages = [] filtered_prompt_messages = []
for prompt_message in prompt_messages: for prompt_message in prompt_messages:
if prompt_message.is_empty(): if prompt_message.is_empty():
@ -556,7 +561,10 @@ class LLMNode(BaseNode):
if not isinstance(prompt_message.content, str): if not isinstance(prompt_message.content, str):
prompt_message_content = [] prompt_message_content = []
for content_item in prompt_message.content: for content_item in prompt_message.content:
if vision_enabled and content_item.type == PromptMessageContentType.IMAGE: if vision_enabled and content_item.type == PromptMessageContentType.IMAGE and isinstance(content_item, ImagePromptMessageContent):
# Override vision config if LLM node has vision config
if vision_detail:
content_item.detail = ImagePromptMessageContent.DETAIL(vision_detail)
prompt_message_content.append(content_item) prompt_message_content.append(content_item)
elif content_item.type == PromptMessageContentType.TEXT: elif content_item.type == PromptMessageContentType.TEXT:
prompt_message_content.append(content_item) prompt_message_content.append(content_item)