Meta's new AI model LLaMA came out today. The paper was a gentle reminder of how capable Google's 540-billion parameter model PaLM will be.
GENERATIVE AI
-
AMECA Robot Demonstrates Real-Time AI Conversation with Speech Recognition
By
–
This #ameca demo couples automated speech recognition with GPT 3
— Pascal Bornet (@pascal_bornet) 25 février 2023
The output is fed to an online service which generates the voice and visemes for lip sync timing. Nothing in this video is pre scripted
Credit:Engineered Arts#artificialintelligence #nlp #llm #ai #machinelearning pic.twitter.com/mRYx46FFDAThis #ameca demo couples automated speech recognition with GPT 3 The output is fed to an online service which generates the voice and visemes for lip sync timing. Nothing in this video is pre scripted Credit:Engineered Arts
#artificialintelligence #nlp #llm #ai #machinelearning -
RLHF Adoption Accelerates Across Industries for Aligned AI
By
–
Today, a senior military official asked me about reinforcement learning with human feedback (RLHF). This is awesome—the widespread adoption of RLHF across every industry and use case is critical for the safe, aligned, and useful deployment of AI and LLMs. RLHF is taking over!
-
Running Diffusion Models on Mac Devices with ANE Support
By
–
You can select whichever device you want to run it on. ANE is supported as well. More deets here: https://
huggingface.co/blog/fast-mac-
diffusers
… -
LLMs as a Solution to Poor Language-Based User Experience
By
–
You could say that LLMs are a general way to fix bad UX in human systems that use language as their interface. Of course, in a better world, this would be fixed via user-friendly online forms instead.
-
LLMs for Document Boilerplate Generation: Solving User-Hostile System Interfaces
By
–
One of the most valuable applications of LLMs: generate document boilerplate for you (e.g. an official letter). It's a UX improvement: the reason this is a problem that needs solving in the first place is that so many human systems have user-hostile interfaces.
-
10B Parameter Models Vastly Underestimated in Capacity
By
–
It's increasingly clear we've severely underestimated how much you can pack in a 10B parameters model
-
OpenAI’s Shift Away From Open Source Mission Criticized
By
–
Texto a pie de página con algún dato interesante respecto al cambio de rumbo que tanto se le ha criticado a OpenAI, donde dejaron atrás parte de su identidad inicial de "open" y sin ánimo de lucro.
-
Adjusting AI Hallucination Levels by Prompt Type
By
–
it would interesting to adjust the "hallucination level" depending on the prompt type when retrieving facts, I am fine with a boring answer as long as it is correct when producing creative output, some indication of which sentences were possibly hallucinated would be helpful
-
Diffusion Modeling: Beyond Jackalopes in Image Generation
By
–
Jackalopes are just the beginning of the images you can computer-generate with diffusion modeling, an interesting, effective technique for creating realistic images, as @quantum_tunnel explains: