9/ Summary of a Haystack – proposes a new task, SummHay, to test a model’s ability to process a Haystack and generate a summary that identifies the relevant insights and cites the source documents.
LLMS
-
Self-Evaluation Defends LLMs Against Adversarial Attacks
By
–
5/ Self-Evaluation as a Defense Against Adversarial Attacks on LLMs – proposes the use of self-evaluation to defend against adversarial attacks; uses a pre-trained LLM to build defense which is more effective than fine-tuned models, dedicated safety LLMs, and enterprise
-
RAG Best Practices: Multimodal Retrieval and Performance Optimization
By
–
3/ Searching for Best Practices in RAG – shows the best practices for building effective RAG workflows; proposes strategies that focus on performance and efficiency, including emerging multimodal retrieval techniques.
-
CriticGPT: New Model Critiques ChatGPT Responses
By
–
2/ CriticGPT – a new model based on GPT-4 to help write critiques for responses generated by ChatGPT; trained using RLHF using a large number of inputs that contained mistakes for which it had to critique.
-
ChatGPT app fix resolves image reversal bug in conversational UI
By
–
A very small but important bug was fixed in the new conversational UI of ChatGPT app, where now, after taking a picture, it won’t appear as reversed
-
Prompt Engineering Gains Equivalent to Fine-Tuning Model Performance
By
–
I see many people rediscovering this paper and jumping to the wrong conclusions. Any gain in prompt engineering is a gain in terms of fine-tuning. A fine-tuned model will simply perform better with many-shot ICL.
-
New Computing Paradigm: LLMs as CPUs with Token-Based Architecture
By
–
🚀 OpenAI Co-founder @karpathy recently shared a groundbreaking New Paradigm for the future of computing:
— Dr. Debashis Dutta (@debashis_dutta) 7 juillet 2024
"We're entering a new computing paradigm with large language models acting like CPUs, using tokens instead of bytes, and having a context window instead of RAM.
This is… pic.twitter.com/HJt4HJFPlXOpenAI Co-founder @karpathy recently shared a groundbreaking New Paradigm for the future of computing: "We're entering a new computing paradigm with large language models acting like CPUs, using tokens instead of bytes, and having a context window instead of RAM. This is
-
LLMs as CPUs: New Computing Paradigm with Tokens and Context
By
–
🚀 OpenAI Co-founder @karpathy recently shared a groundbreaking New Paradigm for the future of computing:
— Dr. Debashis Dutta (@debashis_dutta) 7 juillet 2024
"We're entering a new computing paradigm with large language models acting like CPUs, using tokens instead of bytes, and having a context window instead of RAM.
This is… pic.twitter.com/vf2hgzmdnsOpenAI Co-founder @karpathy recently shared a groundbreaking New Paradigm for the future of computing: "We're entering a new computing paradigm with large language models acting like CPUs, using tokens instead of bytes, and having a context window instead of RAM. This is
-
Does Model Size 248M or 512M Parameters Increase Accuracy?
By
–
If we increase to 248 or 512M, will the accuracy also increase?
-
Investigation into hidden AI model code features
By
–
It feels like Unhinged mode won’t actually see the public No traces to it anymore in the code