It’s not best fit or regression lines since different model sizes/capabilities are on the graph. Why would you best fit to the average of the new 8B, 70B and 405B llama 3.1 models? These are just rough trends lines of the frontier
@thom_wolf
-
GPT-4o-mini Rankings and Model Alignment Methods Evaluation
By
–
Knowing the great people behind lmsys, I don't think there is anything really fishy in the gpt-4o-mini ranking like some people have been saying Likely more of a symptom that as models (and alignement methods) are getting much better it's been harder and harder to tell models
-
Llama 3.1 Research Paper Deep Dive on LLM Training
By
–
Among the most impressive aspect of the Llama 3.1 release is the accompanying research paper! Close to 100 pages of deep knowledge-sharing on LLMs like we havn't seen very often recently What a treat! It covers everything, pretrainining data, filtering, annealing, synthetic
-
Using Deep Learning to Solve Every Problem in Life
By
–
-
LivePortrait: AI-Powered Portrait Animation Tool Released
By
–
Wow 🤯 https://t.co/hwdyFEyHcF pic.twitter.com/me8E7lVmsU
— Thomas Wolf (@Thom_Wolf) 8 juillet 2024Wow https://
huggingface.co/spaces/KwaiVGI
/LivePortrait
… -
Impressive AI Competition Solves Millennium Prize Problems
By
–
There was a super impressive AI competition that happened last week that many people missed in the noise of AI world. I happen to know several participants so let me tell you a bit of this story as a Sunday morning coffee time. You probably know the Millennium Prize Problems
-
Kyutai Labs Unveils End-to-End Audio Model Demo
By
–
The @kyutai_labs fully end-to-end audio model demo of today is a huge deal that many people missed in the room
— Thomas Wolf (@Thom_Wolf) 3 juillet 2024
Mostly irrelevant are the facts that:
– they come a few week after OpenAI ChatGPT-4o
– the demo was less polished than the 4o one (in terms of voice quality, voice… pic.twitter.com/oiZr9jjQNqThe @kyutai_labs fully end-to-end audio model demo of today is a huge deal that many people missed in the room Mostly irrelevant are the facts that:
– they come a few week after OpenAI ChatGPT-4o
– the demo was less polished than the 4o one (in terms of voice quality, voice -
Open LLM Leaderboard v2 Released with Harder Evaluations
By
–
Very excited to release the new version of the Open LLM Leaderboard, v2 – it's much harder than the previous version as you can see on some of the v1 v2 scores comparison I'm posting below Updated: As open models keeps getting better and saturating some of the evaluations it
-
Open LLM Leaderboard New Version Released
By
–
Read all details and even more in the blog post here: https://
huggingface.co/spaces/open-ll
m-leaderboard/blog
… The new version of the leaderboard is at: https://
huggingface.co/spaces/open-ll
m-leaderboard/open_llm_leaderboard
… -
Gemini 3.25B quantized running locally in Chrome sub-100ms
By
–
a 3.25B params quantized gemini running locally in coming Google Chrome with less than 100ms latency while using less than 2GB of ram
— Thomas Wolf (@Thom_Wolf) 24 juin 2024
that's less ram usage than many of my current Chrome page already use (my slack is using 4.8GB as I type this)
no doubt LLMs will be integrated… https://t.co/Muy8o4qPVVa 3.25B params quantized gemini running locally in coming Google Chrome with less than 100ms latency while using less than 2GB of ram that's less ram usage than many of my current Chrome page already use (my slack is using 4.8GB as I type this) no doubt LLMs will be integrated