-
GPT-3.5Turpo的官方API文档的解释是,单轮的对话,包括发送的message和返回的message不超过4k个tokens。
而对于”读论文“的功能,我理解的是,需要一次性地把相当长的论文作为message发送到API才能返回效果较好的摘要。但是很显然,这样单轮的会话远远超过了4k个tokens的限制。
想请教一下是如何实现单轮的长对话的?
-
Traceback (most recent call last):
File "/Users/aquaticdwarf/Downloads/chatPDF-main/main.py", line 99, in
ask(file)
File "/Users/aquaticdwarf/Downloads/chatPDF-main/main.py", line 38, in a…
-
### Description of the feature or enhancement you'd like to see in HandBrake
The new 4000 series, have AV1 encode and decode acceleration, can we expect to see this implemented in HandBrake ? For now…
-
能阅读pdf吗
-
chatpdf也可以精准回答表格中的数值问题,但是解析出来PDF中的表格,怎么切片和向量化?
-
- **(1) Describe the bug 简述**
我使用理解PDF文档内容(模仿chatPDF)功能的时候出现下面的报错
Traceback (most recent call last):
File ".\crazy_functions\crazy_utils.py", line 78, in _req_gpt
result = predict_no_ui_long…
-
Running locally and then uploading the document does not upload the file to Azure Blob storage.
I modified the following values in the "local.setting.json" file as guided.
"OpenAiDocStorName": "ch…
-
**网址**
```
https://xdu.cn/
```
**网站类型**
- [x] ChatGPT站点
- [ ] 其他
**进去随便点一个应用就可以当GPT3.5使用**
**新读是一款强大的文档对话工具,同时也是一个快速体验语义检索任意知识的游乐场。**
在新读,你可以做这些:
* 文档对话,支持pdf,word,excel,ppt
*…
-
embeddings = HuggingFaceEmbeddings(model_name=model_name)
这一句报错,因为huggingface没登录,怎么改成本地已经下载的呀
发生异常: RepositoryNotFoundError
401 Client Error. (Request ID: Root=1-642c3974-240560a346fb2eb565ed790b…
-
目前基于langchain比较惊艳的使用案例有chatPdf和chatExcel,chatExcel能做基于DS的大数据集(1000万行数据)的text2sql吗?
以及目前还有哪些比较经典的langchain的应用案例,可以分享一下吗?
最后,感谢您的这个项目,这是一个非常赞的项目!