setup a @Gradio demo for it on @huggingface, demo made by o1-mini https://t.co/VGjF6H4D53 pic.twitter.com/QiGp2oJxFo
— AK (@_akhaliq) 15 septembre 2024
setup a @Gradio demo for it on @huggingface
, demo made by o1-mini
LLMS
-
Setting Up Gradio Demo on Hugging Face with o1-mini
By
–
-
LLMs positioned mid-way between capabilities and limitations
By
–
I would disagree here, based on current LLM capabilities and failure cases, I would place them mid-way. They are genuinely "in-between"
-
Raw Chain of Thought Access Coming Soon for AI Models
By
–
Indeed disappointing. Only a matter of time until we have options that expose the raw CoT though
-
Google AI Studio Adds Arena Mode for Model Output Comparison
By
–
ICYMI: Google AI Studio now has "arena" mode to compare the model's outputs next to each other and see the diff.
-
Samba-1: Enterprise AI Platform with Security and Data Privacy
By
–
Samba-1 is the only 1-Trillion parameter model delivering the needs of the enterprise The SambaNova Suite provides Security Data Privacy Model ownership Improved TCO Learn more https://
sambanova.ai/products/enter
prise-ai-platform-sambanova-suite
… #GenAI #AI -
GPT-o1 Struggles with Reverse Tic-Tac-Toe Logic Task
By
–
Great new test (until people train on it): reverse-tic-tac-toe. Computer scientist Russ Abbott tried GPT -o1 on this task, where presumably there is not a lot of training data. The results weren’t pretty.
-
Perplexity AI to Expand Search Capabilities with File and Finance Features
By
–
Perplexity will get Files Search very soon it seems. Potentially Finance and Recruiter (LinkedIn) focuses too Full house
-
Commentary on ChatGPT, Claude-style LLMs and o1
By
–
Appreciate the kind words but I also wasn’t wild about ChatGPT at launch so my takes may be a Jim Cramer anti-signal I’m also in awe of o1 as much as anyone, but you can’t quite treat it as a refinement to 4o/Claude-style LLMs. It’s something new, experimental — still raw
-
Data Center Emissions Six Times Worse Than Reported
By
–
Eek – emissions from data centers are likely > 6x worse than reported – and that’s just the start 1. This is this just for 2022
2. 2023-2024 data are surely worse
3. Just wait for the really really big models that come next