langgenius
Abnormalities in multi-round LLM conversations
Self Checks
- [x] This is only for bug report, if you would like to ask a question, please head to Discussions.
- [x] I have searched for existing issues search for existing issues, including closed ones.
- [x] I confirm that I am using English to submit this report (我已阅读并同意 Language Policy).
- [x] [FOR CHINESE USERS] 请务必使用英文提交 Issue，否则会被关闭。谢谢！:）
- [x] Please do not modify this template :) and fill in all the required fields.
Dify version
v0.15.3
Cloud or Self Hosted
Self Hosted (Docker)
Steps to reproduce
Using vllm to launch the Deepseek-7b model,“Model Provider” - > "ADD MORE MODEL PROVIDER" -> localAI ,Then configure the workflow to upload the file, and then have multiple rounds of conversations，But use "ADD MORE MODEL PROVIDER" -> ollama，Normal operation。
✔️ Expected Behavior
Normal operation
❌ Actual Behavior
“User message content must be str”
DSL：
app:
description: ''
icon: 🤖
icon_background: '#FFEAD5'
mode: advanced-chat
name: 上传文件读取
use_icon_as_answer_icon: false
kind: app
version: 0.1.5
workflow:
conversation_variables: []
environment_variables: []
features:
file_upload:
allowed_file_extensions: []
allowed_file_types:
- document
allowed_file_upload_methods:
- local_file
enabled: true
fileUploadConfig:
audio_file_size_limit: 50
batch_count_limit: 5
file_size_limit: 15
image_file_size_limit: 10
video_file_size_limit: 100
workflow_file_upload_limit: 10
image:
enabled: false
number_limits: 3
transfer_methods:
- local_file
- remote_url
number_limits: 1
opening_statement: ''
retriever_resource:
enabled: true
sensitive_word_avoidance:
enabled: false
speech_to_text:
enabled: false
suggested_questions: []
suggested_questions_after_answer:
enabled: false
text_to_speech:
enabled: false
language: ''
voice: ''
graph:
edges:
- data:
sourceType: llm
targetType: answer
id: llm-answer
source: llm
sourceHandle: source
target: answer
targetHandle: target
type: custom
- data:
isInIteration: false
sourceType: start
targetType: document-extractor
id: 1738899393364-source-1738922032713-target
source: '1738899393364'
sourceHandle: source
target: '1738922032713'
targetHandle: target
type: custom
zIndex: 0
- data:
isInIteration: false
sourceType: document-extractor
targetType: llm
id: 1738922032713-source-llm-target
source: '1738922032713'
sourceHandle: source
target: llm
targetHandle: target
type: custom
zIndex: 0
nodes:
- data:
desc: ''
selected: false
title: 开始
type: start
variables: []
height: 53
id: '1738899393364'
position:
x: 30
y: 251
positionAbsolute:
x: 30
y: 251
selected: false
sourcePosition: right
targetPosition: left
type: custom
width: 244
- data:
context:
enabled: true
variable_selector:
- '1738922032713'
- text
desc: ''
memory:
query_prompt_template: '{{#sys.query#}}'
role_prefix:
assistant: ''
user: ''
window:
enabled: false
size: 10
model:
completion_params:
temperature: 0.7
mode: chat
name: DeepSeek-R1-Distill-Qwen-7B
provider: localai
prompt_template:
- id: 4a6623cd-78ba-4a4d-9478-589528dca75c
role: system
text: 你是一个文件助手，根据用户需要，结合文件内容回答用户，文件内容在<files>{{#context#}}</files>
selected: false
title: LLM
type: llm
variables: []
vision:
enabled: false
height: 97
id: llm
position:
x: 638
y: 251
positionAbsolute:
x: 638
y: 251
selected: true
sourcePosition: right
targetPosition: left
type: custom
width: 244
- data:
answer: '{{#llm.text#}}'
desc: ''
selected: false
title: 直接回复
type: answer
variables: []
height: 102
id: answer
position:
x: 942
y: 251
positionAbsolute:
x: 942
y: 251
selected: false
sourcePosition: right
targetPosition: left
type: custom
width: 244
- data:
desc: ''
is_array_file: true
selected: false
title: 文档提取器
type: document-extractor
variable_selector:
- sys
- files
height: 91
id: '1738922032713'
position:
x: 332.57142857142856
y: 251
positionAbsolute:
x: 332.57142857142856
y: 251
selected: false
sourcePosition: right
targetPosition: left
type: custom
width: 244
viewport:
x: 77.86710032646181
y: 174.9890744049676
zoom: 0.6093853943072869
@love01211 Hello there! I'm here to help you with bug reports, answer your questions, and assist you in becoming a contributor. Let's work together to solve this issue.
I found a similar unsolved discussion regarding issues with continuous dialogue when uploading files in Dify. You can check it out here [1].
To resolve the "User message content must be str" error when using the Deepseek-7b model with vllm in a self-hosted Dify setup, ensure that the content of UserPromptMessage is always a string. In the code, the _convert_prompt_messages_to_dicts method handles the conversion of UserPromptMessage content. Make sure that when creating UserPromptMessage, the content attribute is a string. If the content is not a string, convert it to a string format before passing it to the model. If the content is a list of PromptMessageContentType, ensure it is properly handled and converted to a suitable format that the model can process [2].
To continue talking to Dosu, mention @dosu.
Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Bug Report | Other
![dosubot[bot] avatar](https://avatars.githubusercontent.com/in/324583?v=4 "Published by a pub.dev verified publisher"){kind=small}
Hi, @love01211. I'm Dosu, and I'm helping the Dify team manage their backlog. I'm marking this issue as stale.
Issue Summary:
- You reported an error with the Deepseek-7b model in a self-hosted environment using vllm: "User message content must be str."
- The issue does not occur with the "ollama" model provider.
- I suggested ensuring
UserPromptMessagecontent is a string and shared a related discussion link. - Both you and another user, yxdfighting, indicated that the suggestion did not resolve the issue.
Next Steps:
- Please confirm if this issue is still relevant with the latest version of the Dify repository. If so, you can keep the discussion open by commenting here.
- If there is no further activity, this issue will be automatically closed in 15 days.
Thank you for your understanding and contribution!