Comments (2)
The inconsistency in model behavior after updating prompt_config.py
could be attributed to several key factors:
-
Complexity and Structure of Prompts: Different models, like ChatGLM3-6B and Yi-34B-int4, may interpret the same prompts differently due to their unique training and architecture. This can lead to variations in how closely they adhere to the intended structure of the prompts.
-
Dynamic Content Handling: If the dynamic content (e.g., variables or placeholders within your prompts) isn't correctly parsed or rendered, it might cause inconsistent responses. Ensuring that this content is accurately inserted into the prompts is crucial.
-
Temperature Settings: The randomness in response generation is influenced by temperature settings. A higher temperature can result in more varied responses, while a lower temperature tends to produce more consistent but potentially less creative outputs. Finding the right balance for your specific use case might require some experimentation.
-
Prompt Directiveness: The clarity and directiveness of your prompts play a significant role. More directive prompts tend to yield more consistent results, whereas open-ended prompts can lead to a wider range of responses due to the model's interpretation.
To address these issues, consider simplifying your prompt structures, ensuring dynamic content is correctly rendered, experimenting with temperature settings to find an optimal balance, and tailoring prompts to the specific strengths and characteristics of each model. This approach can help mitigate inconsistencies and improve the predictability of model responses.
Details
Sources
- configs/prompt_config.py.example
- server/api.py
- knowledge_base/samples/content/test_files/langchain-ChatGLM_open.jsonl
- server/chat/completion.py
About Dosu
This response is meant to be useful and save you time. It is not meant to be a precise solution, but rather a starting point for your own research.
Help me be more useful! Please leave a 👍 if this is helpful and 👎 if it is irrelevant.
To continue the conversation, mention @dosu.
from langchain-chatchat.
你的模板没对吧, system 你要填入什么呢
input是会被替换成你的问题的
这个框架没做system角色
from langchain-chatchat.
Related Issues (20)
- cannot import name 'PDFResourceManag er' from 'pdfminer.converter'这是什么错误[BUG] 简洁阐述问题 / Concise description of the issue HOT 1
- 如何同时接入多个封装好的openai api格式大模型 HOT 2
- 请问怎么计算输出的token数量?这个框架有现成支持的方法吗
- 调用讯飞星火api,没有输出回答 HOT 2
- [BUG] 知识库问答匹配不到/ Concise description of the issue HOT 4
- 是否可以支持用户中断模型输出
- 怎么调用昇腾服务器npu? HOT 1
- docker部署的milvus数据库有大量的数据,想直接迁移到其他机器上,可以吗?
- 分词和分块逻辑顺序
- [FEATURE] 在问答中能够返回PDF或者DOCX中的图片信息
- [BUG] 使用es作向量数据库时,某些markdown文档向量化失败 HOT 2
- v1/chat/completions
- The request client is not a secure context and the resource is in more-private address space。特殊跨域问题如何解决。
- 求助:python startup.py -a 报错assert r.status_code == 200 AssertionError HOT 1
- 求助:如何不使用api,批量进行预测 HOT 2
- 启动startup.py,即便显卡不存在运行进程,依然显示报错:torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 134.00 MiB. GPU 0 has a total capacty of 21.99 GiB of which 119.06 MiB is free HOT 2
- [BUG] 知识库名称使用中文出现错误 HOT 1
- 0.2.10版本,知识库,不能基于上次回答的内容连续对话。
- 0.2.10版本,知识库内容拆分,指定特殊分割符,没有效果。
- 0.2.10版本,知识库对话,多个问题回答不了。
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from langchain-chatchat.