Our CEO, Dario Amodei, shared the guiding principles behind Anthropic's Responsible Scaling Policy at the UK AI Safety Summit at Bletchley Park this afternoon. Read his remarks:
RESEARCH
-
Stanford Researchers Design AI Tools for Educators Preserving Student Relationships
By
–
From math tutor to teacher’s aid, Stanford education researchers are designing expert-in-the-loop NLP tools that give educators new insights while preserving student-teacher relationships.
-
Graph-Based Parent Document Retriever with Neo4j Integration
By
–
Graph-based Parent Document Retriever Parent Document Retriever is a more advanced retrieval technique We've worked with @neo4j to add a template for this that uses their graph database under the hood! Template: https://
github.com/langchain-ai/l
angchain/tree/master/templates/neo4j-parent
… Here's how it works: —- -
Three Essential Deep Learning Lessons: Scale, Data, and Engineering
By
–
Deep learning lessons (known in 2016 but starkly confirmed since): 1. Scalable ideas > clever ideas
2. Improving the dataset > improving the model
3. Engineering chops > academic chops -
NVIDIA Partners Recursion to Accelerate Drug Discovery
By
–
Recursion is working with us to streamline drug discovery. Combining @RecursionPharma
’s computational expertise with our tech, the collaboration hopes to make better predictions and increase a drug’s probability of success with patients. https://
nvda.ws/48HLCnz #NVIDIAstory -
Cerebras SDK Workshop for Scientific Computing at SC23
By
–
(1/2) On November 14th and 15th, stop by the @psc_live booth at #SC23 for an informational session. Join us from 2:00 – 2:45 pm MST for this session to learn how you can develop kernels and deliver scientific computing research using the Cerebras SDK on Neocortex
-
Cerebras Neocortex Machine for HPC and Machine Learning Research
By
–
(2/2) We will discuss PSC’s Neocortex machine for HPC research and machine learning. In addition, we’ll discuss scientific computing research on the world’s largest chip and our SDK-built Gordon Bell Finalist research. Want to meet at SC23?
https://
cerebras.net/contact-us/ -
Percentogram: Histogram Alternative for Cumulative Distribution Analysis
By
–
The “percentogram”—a histogram binned by percentages of the cumulative distribution, rather than using fixed bin widths | Statistical Modeling, Causal Inference, and Social Science https://
bit.ly/3DMdQ2l
#AI #MachineLearning #DeepLearning #LLMs #DataScience -
UK AI Safety Summit Unites Industry on Genuine Safety
By
–
The UK AI Safety Summit is a historic moment. Great to see *everyone* – big labs, small startups, OSS, academics and govt – focusing on genuine ai safety. @bletchleypark
-
Goodside: high logprobs not cached responses cause joke reuse
By
–
I don’t think this is evidence of cached responses; the logprobs may just be that high. GPT has had issues with joke reuse at least since RLHF (and likely before with FeedMe) — e.g. ChatGPT (GPT-3.5) was found to reuse the same 25 jokes for 90% responses: https://
arxiv.org/abs/2306.04563