LongNet: Scaling Transformers to 1,000,000,000 Tokens paper page: https://
huggingface.co/papers/2307.02
486
… Scaling sequence length has become a critical demand in the era of large language models. However, existing methods struggle with either computational complexity or model expressivity,
RESEARCH
-
LongNet: Scaling Transformers to Handle Billion Tokens
By
–
-
Open-Source LLMs Match ChatGPT Performance in Text Annotation Tasks
By
–
Open-Source Large Language Models Outperform Crowd Workers and Approach ChatGPT in Text-Annotation Tasks paper page: https://
huggingface.co/papers/2307.02
179
… study examines the performance of open-source Large Language Models (LLMs) in text annotation tasks and compares it with proprietary -
Building Cooperative Embodied Agents with Large Language Models
By
–
Building Cooperative Embodied Agents Modularly with Large Language Models
— AK (@_akhaliq) 6 juillet 2023
paper page: https://t.co/QGisKj2HWc
Large Language Models (LLMs) have demonstrated impressive planning abilities in single-agent embodied tasks across various domains. However, their capacity for planning… pic.twitter.com/cajVGEH6zqBuilding Cooperative Embodied Agents Modularly with Large Language Models paper page: https://
huggingface.co/papers/2307.02
485
… Large Language Models (LLMs) have demonstrated impressive planning abilities in single-agent embodied tasks across various domains. However, their capacity for planning -
Jailbroken: How Does LLM Safety Training Fail?
By
–
Jailbroken: How Does LLM Safety Training Fail? paper page: https://
huggingface.co/papers/2307.02
483
… Large language models trained for safety and harmlessness remain susceptible to adversarial misuse, as evidenced by the prevalence of "jailbreak" attacks on early releases of ChatGPT that elicit -
Flacuna: Enhancing Vicuna Problem-Solving with FLAN Fine-Tuning
By
–
Flacuna: Unleashing the Problem Solving Power of Vicuna using FLAN Fine-Tuning paper page: https://
huggingface.co/papers/2307.02
053
… Recently, the release of INSTRUCTEVAL has provided valuable insights into the performance of large language models (LLMs) that utilize encoder-decoder or -
ETH Zürich DLSC Course Introduction
By
–
ETH Zürich DLSC: Course Introduction https://
bit.ly/3qTjAUD #AI #MachineLearning #DeepLearning #LLMs #DataScience -
OpenAI predicts AI Superintelligence before 2030 with extinction risk
By
–
OpenAI predicts that we will arrive at AI Superintelligence (exceeding the capabilities of AGI, in their words) before 2030. They acknowledge this could either solve many of the world’s problems, or lead to the outright extinction of humanity. Posted today from OpenAI blog:
-
Recommendation for RLHF Technical Tutorial
By
–
Great collection! And after the technical tutorials, I can highly recommend “How RLHF actually works”
-
End-to-End Learning Limitations: The Unspoken Assumption of Hardware
By
–
EndToEnd learning is a fetish, even with pre-training. Why not include the computer, along with the software, and have that learned? Why the distinction? BECAUSE it won't work. The innate computing machine is assumed. Y'all are playing an intellectual game, but don't discuss it.
-
Balancing Early Detection and Positive Predictive Value
By
–
It’s a really careful balance. I’m all for early detection fwiw but positive predictive value is name of the game.