Exploring ChatGPT, APIs, and Language Models
Article Bias and Twitter
- Discussion about biased articles and preference for tweets
- Shared link to a tweet: https://twitter.com/zoink/status/1653052807950536706
Learning with ChatGPT
- Discussion on how to use ChatGPT for learning
- Mention of yudbot.com as a resource
API Level Caps
- Mention of developing an API level cap for a llm vault manager
- Question on whether API level caps are generally kept
- Shared link to yudbot.com as a potential resource
Paywalls and Capitalism
- Discussion on paywalls and access to news articles
- Mention of capitalism in a political group context
Conversational Memory with GPT-3.5-Turbo
- Question on how to add conversational memory to GPT-3.5-Turbo
- Shared link to a potential resource
Langchain
- Mention of Langchain and its shortcomings
- Discussion on whether to use Langchain or roll your own solution
- Mention of llamaindex as an alternative
- Mention of a friend’s positive experience with llamaindex for chat conversations
- Mention of a Lucene-based ANNs project
- Mention of company-level constraints on using certain tools
- Mention of lack of quality checks in Langchain repo
- Mention of Langchain prioritizing agents and toolchain over everything else
- Mention of a potential Langchain alternative, Semantic Kernel
- Discussion on API design and community traction
- Mention of issues with Langchain at scale
- Shared link to llama_index default prompts
Rolling Windows and Summarization
- Discussion on implementing rolling windows contexts
- Mention of issues with moving windows approach
- Suggestion to use dynamic rolling window for every question
- Discussion on summarization and use of OpenAI summarization API
- Mention of prompt engineering as a solution
- Suggestion to use a smaller and faster model for summarization
- Mention of using GPT-3.5 and prompt engineering to keep temp low
- Mention of experience vs expense trade-off
- Mention of sandboxing Python environment
- Mention of executing untrusted code from an LLM
- Mention of using tools and agents with Langchain as part of OpenAI cookbook
Miscellaneous
- Mention of a potential resource for creating a running summary of earlier conversations
- Mention of knowing the creator of 42papers.com, artcompute.com, and mindsjs.com
- Mention of a potential solution for executing untrusted code from an LLM
Links
The description and link can be mismatched because of extraction errors.
- The tweet at https://twitter.com/zoink/status/1653052807950536706 suggests that the author’s tweets are better, and asks how people are using ChatGPT to learn.
- https://www.yudbot.com/ - The message in the same link as the URL questions the ethics of putting news articles behind a paywall.
- The URL https://augmented-reality-knowledge.github.io/ is being shared and the message asks if there are any shortcomings that need to be aware of and if the link is not working for the recipient.
- Code flow example: https://github.com/hwchase17/langchain/blob/master/langchain/memory/summary.py (link to a code example on GitHub)
- https://deeplearn.org/ - someone mentions using this website for something related to a question about a commercial license for deforum and a potential alternative called Semantic Kernel.
- URL: https://github.com/jerryjliu/llama_index/blob/main/gpt_index/prompts/default_prompts.py
- Context: The message warns against using langchain and suggests an alternative approach for saving chat messages in a system. The URL may provide more information on the topic.
- https://42papers.com/ - Creator of this website and other related websites may have insight into why LangChain is focused on tools and agents.