Amazing to see the energy of the OpenAI developer community. These models are starting to unlock real opportunity to build on top of them, whether for side projects, new products, or integrations into established services.
LLMS
-
Customize Summarization and Question Answering with Custom Prompts
By
–
Customize summarization and question answering The default prompts used for summarization and Question/Answering may not be specific enough for your use case Luckily, you can easily customize them! Here's an example asking them to respond in Italian
-
LangChain Question Answering Examples and LLM Applications
By
–
Check out example notebooks: https://
langchain.readthedocs.io/en/latest/modu
les/chains/combine_docs_examples/question_answering.html
… Shout out to @johnjnay
, @BruceHammer
, and others in the discord for helping to ideate + test John's already started to use it in some more awesome work on llm-lobbyist: -
Recursive Text Splitter for Improved Chunk Processing
By
–
Recursive Text Splitter The previous text splitter only split on a single character This one recursively splits chunks on different types of characters, until all chunks are below the desired size
-
LangChain 0.0.59 Adds Map-Rerank Chain and Text Splitting
By
–
New LangChain version makes it easier than ever to combine LLMs with your own data Brand new map-rerank chain
Recursive Text Splitter
Customize summarization and question answering `pip install langchain==0.0.59` -
Map-Rerank Chain for Enhanced Question Answering Systems
By
–
Brand new map-rerank chain This chain can be used for question answering. It maps over documents, trying to both (a) answer a question, (b) assign a score to how good the answer is It then picks the answer with the highest score
-
AI Captures Human Nuance Through Better Training Data
By
–
“The human touch” is embedded in all the training data these AIs are learning from. As we get better at learning from this data, AI will be able to capture the nuances of human writing. Even today's AIs can come really close if you prompt them well (add direction, angles, etc.)
-
Thorough Mitigation for Prompt Injection in AI Systems
By
–
It’s premature to call prompt injection solved, but this is the most thorough mitigation I’ve seen yet.
-
Anthropic’s Constitutional AI Fine-Tuning Includes Red-Team Prompts
By
–
Anthropic process for Constitutional AI explicitly includes red-team prompts like “Ignore previous directions” in the fine-tuning:
-
RLHF reduces prompt engineering needs but unlocking full LLM potential requires prompting
By
–
useful counterpoint – RLHF diminishes the need for prompt eng for the median usecase, but the capabilities of LLMs are much more vast and will be unlocked with some amount of prompting (remains to be seen HOW MUCH engineering is required; is @langchain a feature or bug?)