ThursdAI - The top AI news from the past week From Weights & Biases, Join AI Evangelist Alex Volkov and a panel of experts to cover everything important that happened in the world of AI from the past week

Hey hey folks, happy ThursdAI 🎉
Not a lot of house-keeping here, just a reminder that if you're listening or reading from Europe, our European fullyconnected.com conference is happening in May 15 in London, and you're more than welcome to join us there. I will have quite a few event updates in the upcoming show as well.
Besides this, this week has been a very exciting one for smaller models, as Microsoft teased and than released Phi-3 with MIT license, a tiny model that can run on most macs with just 3.8B parameters, and is really punching above it's weights. To a surprising and even eyebrow raising degree! Let's get into it 👇
ThursdAI - Recaps of the most high signal AI weekly spaces is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.

TL;DR of all topics covered:
* Open Source LLMs
* Microsoft open sources Phi-3 (X, HF)
* LLama3 70B top5 (no top 6) on LMsys (LMsys Arena)
* Snowflake open sources Arctic - A massive hybrid MoE (X, Try it, HF)
* Evolutionary Model merges support in MergeKit (Blog)
* Llama-3 8B finetunes roundup - Longer Context (128K) and Dolphin & Bagel Finetunes
* HuggingFace FINEWEB - a massive 45TB (the GPT4 of datasets) and 15T tokens high quality web data dataset (HF)
* Cohere open sourced their chat interface (X)
* Apple open sources OpenElm 4 models + training library called corenet (HF, Github, Paper)
* Big CO LLMs + APIs
* Google Gemini 1.5 pro is #2 on LMsys arena
* Devin is now worth 2BN and Perplexity is also a Unicorn
* A new comer called Augment (backed by Eric Schmidt) is now coming out of stealth (X)
* Vision & Video
* Adobe releases VideoGigaGAN - high quality upscaler with temporal consistency (paper)
* TLDraw autocomplete UI demo (X)
* This Weeks Buzz - What I learned in WandB this week
* Joe Spisak talk about Llama3 on Stage at WandB Fully connected (Full Talk, TLDR)
* Voice & Audio
* Play.ai (previously play.ht) releases conversational Voice AI platform (X)
* AI Art & Diffusion & 3D
* IMGsys.org- like LMsys but for image generation model + leaderboard from FAL (try it)
* Tools & Hardware
* Rabbit R1 release party & no shipping update in sight
* I'm disillusioned about my AI Pin and will return it
Open Source LLMs
Llama-3 1 week-aversary 🎂 - Leaderboard ranking + finetunes
Well, it's exactly 1 week since we got Llama-3 from Meta and as expected, the rankings show a very very good story. (also it was downloaded over 1.2M times and already has 600 derivatives on HuggingFace)
Just on Monday, Llama-3 70B (the bigger version) took the incredible 5th place (now down to 6th) on LMSys, and more surprising, given that the Arena now has category filters (you can filter by English only, Longer chats, Coding etc) if you switch to English Only, this model shows up 2nd and was number 1 for a brief period of time.
So just to sum up, an open weights model that you can run on most current consumer hardware is taking over GPT-4-04-94, Claude Opus etc'
This seems dubious, because well, while it's amazing, it's clearly not at the level of Opus/Latest GPT-4 if you've used it, in fact it fails some basic logic questions in my tests, but it's a good reminder that it's really hard to know which model outperforms which and that the arena ALSO has a bias, of which people are using it for example and that evals are not a perfect way to explain which models are better.
However, LMsys is a big component of the overall vibes based eval in our community and Llama-3 is definitely a significant drop and it's really really good (even the smaller one)
One not so surprising thing about it, is that the Instruct version is also really really good, so much so, that the first finetunes of Eric Hartfords Dolphin (Dolphin-2.8-LLama3-70B) is improving just a little bit over Meta's own instruct version, which is done very well.
Per Joe Spisak (Program Manager @ Meta AI) chat at the Weights & Biases conference last week (which you

Happy LLama 3 day folks! After a lot of rumors, speculations, and apparently pressure from the big Zuck himself, we finally can call April 18th, 2024, LLaMa 3 day!
I am writing this, from a lobby of the Mariott hotel in SF, where our annual conference is happening called Fully Connected, and I recorded today's episode from my hotel room. I really wanna shout out how awesome it was to meet folks who are listeners of the ThursdAI pod and newsletter subscribers, participate in the events, and give high fives.
During our conference, we had the pleasure to have Joe Spisak, the Product Director of LLaMa at Meta, to actually announce LLaMa3 on stage! It was so exhilarating, I was sitting in the front row, and then had a good chat with Joe outside of the show 🙌
The first part of the show was of course, LLaMa 3 focused, we had such a great time chatting about the amazing new 8B and 70B models we got, and salivating after the announced but not yet released 400B model of LLaMa 3 😮
We also covered a BUNCH of other news from this week, that was already packed with tons of releases, AI news and I was happy to share my experiences running a workshop a day before our conference, with focus on LLM evaluations. (If there's an interest, I can share my notebooks and maybe even record a video walkthrough, let me know in the comments)
Ok let's dive in 👇
Happy LLama 3 day 🔥
The technical details
Meta has finally given us what we're all waiting for, an incredibly expensive (2 clusters of 24K H100s over 15 Trillion tokens) open weights models, the smaller 8B one and the larger 70B one.
We got both instruction fine tune and base models, which are great for finetuners, and worth mentioning that this is a dense model (not a mixture of experts, all the parameters are accessible for the model during inference)
It is REALLY good at benchmarks, with the 7B model beating the previous (LLaMa 2 70B) on pretty much all benchmarks, and the new 70B is inching on the bigger releases from the past month or two, like Claude Haiku and even Sonnet!
The only downsides are the 8K context window + non multimodality, but both are coming according to Joe Spisak who announced LLama3 on stage at our show Fully Connected 🔥
I was sitting in the front row and was very excited to ask him questions later!
By the way, Joe did go into details they haven't yet talked about pulblicly (see? I told you to come to our conference! and some of you did!) and I've been live-tweeting his whole talk + the chat outside with the "extra" spicy questions and Joes winks haha, you can read that thread here
The additional info
Meta has also partnered with both Google and Bing (take that OpenAI) and inserted LLama 3 into the search boxes of Facebook, Instagram, Messenger and Whatsapp plus deployed it to a new product called meta.ai (you can try it there now) and is now serving LLama 3 to more than 4 Billion people across all of those apps, talk about compute cost!
Llama 3 also has a new Tokenizer (that Joe encouraged us to "not sleep on") and a bunch of new security tools like Purple LLama and LLama Guard. PyTorch team recently released finetuning library called TorchTune is now supporting LLama3 finetuning natively out of the box as well (and integrates Wandb as it's first party experiment tracking tool)
If you'd like more details, directly from Joe, I was live tweeting his whole talk, and am working at getting the slides from our team. We'll likely have a recording as well, will post it as soon as we have it.
Here's a TL;DR (with my notes for the first time) of everything else we talked about, but given today is LLaMa day, and I still have to do fully connected demos, I will "open source" my notes and refer you to the podcast episode to hear more detail about everything else that happened today 🫡
TL;DR of all topics covered:
* Meta releases LLama 3 -8B, 70B and later 400B (Announcement, Models, Try it, Run Locally)
* Open Source LLMs
* Meta LLama 3

this week was absolutely bonkers. For starters, for the first time ever, we got an Open Weights model (Command R+) to jump over GPT-4 in human rankings on LMsys, this is huge!
Then on Tuesday, it seems that all the companies just wanted to one up one another, first Gemini 1.5 released with updates, made it available in 180 countries, added audio mode + tons of API improvements and system prompts, then less than an hour later, OpenAI has given us a "majorly improved" GPT-4 Turbo version (2024-04-09) that is now back to being the BEST LLM IN THE WORLD and to cap that day off, Mistral did the thing again, the thing being, dropping a torrent link in a tweet with no explanations.
What was in that torrent is a Mixtral 8x22B MoE (which we started calling Bixtral) which comes with an Apache2 license and seems to be VERY good!
We also saw the first finetune from HuggingFace/KAIST folks less than 48 hours later (the authors of said finetune actually came on the show 🎉 )
Fully Connected is a week from today! If you haven't yet signed up, use THURSDAI promo code and come hear from Richard Socher (You.com), Jerry Liu (Ilamaindex CEO), Karoly (TwoMinutePapers), Joe Spisak (Meta) and and leaders from NVIDIA, Snowflake, Microsoft, Coatue, Adobe, Siemens, Lambda and tons more 👇
TL;DR of all topics covered:
* Open Source LLMs
* 🔥 Mistral releases Mixtral 8x22 Apache 2 licensed MoE model (Torrent, TRY IT)
* Cohere CMDR+ jumps to no 6 on LMSys and beats GPT4 (X)
* CodeGemma, RecurrentGemma & Gemma Instruct 1.1 (Announcement)
* Auto-code-rover gets 22% on SWE bench (Announcement)
* HuggingFace - Zephyr 141B-A35B - First Bixtral Finetune (Announcement)
* Mistral 22B - 1 single expert extracted from MoE (Announcement, HF)
* This weeks Buzz - Weights & Biases updates
* FullyConnected is in 1 week! (Come meet us)
* Big CO LLMs + APIs
* 🔥 GPT-4 turbo is back to being number 1 AI with 88.2% Human Eval score (X)
* Gemini 1.5 Pro now understands audio, uses unlimited files, acts on your commands, and lets devs build incredible things with JSON mode (X)
* LLama 3 coming out in less than a month (confirmed by Meta folks)
* XAI Grok now powers news summaries on X (Example)
* Cohere new Rerank 3 (X)
* Voice & Audio
* HuggingFace trained Parler-TTS (Announcement, Github)
* Udio finally launched it's service (Announcement, Leak, Try It)
* Suno has added explore mode (suno.ai/explore)
* Hardware
* Humane AI pin has started shipping - reviews are not amazing

Open Source LLMs
Command R+ first open weights model that beats last year GPT4 versions
This is massive, really a milestone to be discussed, and even though tons of other news happened, the first time an open weights model is beating GPT-4 not on a narrow case (coding, medical) but on a general human evaluation on the arena.
This happened just a year after GPT-4 first came out, and is really really impressive.
Command R+ has been getting a lot of great attention from the community as well, folks were really surprised by the overall quality, not to mention the multilingual abilities of CommandR+
Mixtral 8x22B MoE with 65K context and Apache 2 license (Bigstral)
Despite the above, Cohere time in the sun (ie top open weights model on lmsys) may not be that long if the folks at Mistral have anything to say about it!
Mistral decided to cap the crazy Tuesday release day with another groundbreaking tweet of theirs which includes a torrent link and nothing else (since then they of course uploaded the model to the hub) giving us what potentially will unseat Command R from the rankings.
The previous Mixtral (8x7B) signaled the age of MoEs and each expert in that was activated from Mistral 7B, but for this new affectionally named Bixtral model, each expert is a 22B sized massive model.
We only got a base version of it, which is incredible on it's own right, but it's not instruction finetuned yet, and the finetuner community is already cooking really hard! Though it's hard because this model requi

Happy first ThursdAI of April folks, did you have fun on April Fools? 👀 I hope you did, I made a poll on my feed and 70% did not participate in April Fools, which makes me a bit sad!
Well all-right, time to dive into the news of this week, and of course there are TONS of news, but I want to start with our own breaking news! That's right, we at Weights & Biases have breaking new of our own today, we've launched our new product today called Weave!
Weave is our new toolkit to track, version and evaluate LLM apps, so from now on, we have Models (what you probably know as Weights & Biases) and Weave. So if you're writing any kind RAG system, anything that uses Claude or OpenAI, Weave is for you!
I'll be focusing on Weave and I'll be sharing more on the topic, but today I encourage you to listen to the launch conversation I had with Tim & Scott from the Weave team here at WandB, as they and the rest of the team worked their ass off for this release and we want to celebrate the launch 🎉
TL;DR of all topics covered:
* Open Source LLMs
* Cohere - CommandR PLUS - 104B RAG optimized Sonnet competitor (Announcement, HF)
* Princeton SWE-agent - OSS Devin - gets 12.29% on SWE-bench (Announcement, Github)
* Jamba paper is out (Paper)
* Mozilla LLamaFile now goes 5x faster on CPUs (Announcement, Blog)
* Deepmind - Mixture of Depth paper (Thread, ArXiv)
* Big CO LLMs + APIs
* Cloudflare AI updates (Blog)
* Anthropic adds function calling support (Announcement, Docs)
* Groq lands function calling (Announcement, Docs)
* OpenAI is now open to customers without login requirements
* Replit Code Repair - 7B finetune of deep-seek that outperforms Opus (X)
* Google announced Gemini Prices + Logan joins (X)קרמ
* This weeks Buzz - oh so much BUZZ!
* Weave lunch! Check weave out! (Weave Docs, Github)
* Sign up with Promo Code THURSDAI at fullyconnected.com
* Voice & Audio
* OpenAI Voice Engine will not be released to developers (Blog)
* Stable Audio v2 dropped (Announcement, Try here)
* Lightning Whisper MLX - 10x faster than whisper.cpp (Announcement, Github)
* AI Art & Diffusion & 3D
* Dall-e now has in-painting (Announcement)
* Deep dive
* Jamba deep dive with Roi Cohen from AI21 and Maxime Labonne
Open Source LLMs
Cohere releases Command R+, 104B RAG focused model (Blog)
Cohere surprised us, and just 2.5 weeks after releasing Command-R (which became very popular and is No 10 on Lmsys arena) gave us it's big brother, Command R PLUS
With 128K tokens in the context window, this model is multilingual as well, supporting 10 languages and is even beneficial on tokenization for those languages (a first!)
The main focus from Cohere is advanced function calling / tool use, and RAG of course, and this model specializes in those tasks, beating even GPT-4 turbo.
It's clear that Cohere is positioning themselves as RAG leaders as evident by this accompanying tutorial on starting with RAG apps and this model further solidifies their place as the experts in this field. Congrats folks, and thanks for the open weights 🫡
SWE-Agent from Princeton
Folks remember Devin? The super cracked team born agent with a nice UI that got 13% on the SWE-bench a very hard (for LLMs) benchmark that requires solving real world issues?
Well now we have an open source agent that comes very very close to that called SWE-Agent
SWE agent has a dedicated terminal and tools, and utilizes something called ACI (Agent Computer Interface) allowing the agent to navigate, search, and edit code.
The dedicated terminal in a docker environment really helps as evident by a massive 12.3% score on SWE-bench where GPT-4 gets only 1.4%!
Worth mentioning that SWE-bench is a very hard benchmark that was created by the folks who released SWE-agent, and here's some videos of them showing the agent off, this is truly an impressive achievement!
Deepmind publishes Mixture of Depth (arXiv)
Thanks to Hassan who read the paper and wrote a deep dive, this paper by Deepmind shows thei

Hey everyone, this is Alex and can you believe that we're almost done with Q1 2024? March 2024 was kind of crazy of course, so I'm of course excited to see what April brings (besides Weights & Biases conference in SF called Fully Connected, which I encourage you to attend and say Hi to me and the team!)
This week we have tons of exciting stuff on the leaderboards, say hello to the new best AI in the world Opus (+ some other surprises), in the open source we had new MoEs (one from Mosaic/Databricks folks, which tops the open source game, one from AI21 called Jamba that shows that a transformers alternative/hybrid can actually scale) and tiny MoE from Alibaba, as well as an incredible Emotion TTS from Hume.
I also had the pleasure to finally sit down with friend of the pod Tanishq Abraham and Paul Scotti from MedArc and chatted about MindEye 2, how they teach AI to read minds using diffusion models 🤯🧠👁️
Thank you for reading ThursdAI - Recaps of the most high signal AI weekly spaces. This post is public so feel free to share it.

TL;DR of all topics covered:
* AI Leaderboard updates
* Claude Opus is number 1 LLM on arena (and in the world)
* Claude Haiku passes GPT4-0613
* 🔥 Starling 7B beta is the best Apache 2 model on LMsys, passing GPT3.5
* Open Source LLMs
* Databricks/Mosaic DBRX - a new top Open Access model (X, HF)
* 🔥 AI21 - Jamba 52B - Joint Attention Mamba MoE (Blog, HuggingFace)
* Alibaba - Qwen1.5-MoE-A2.7B (Announcement, HF)
* Starling - 7B that beats GPT3.5 on lmsys (HF)
* LISA beats LORA as the frontrunner PeFT (X, Paper)
* Mistral 0.2 Base released (Announcement)
* Big CO LLMs + APIs
* Emad leaves stability 🥺
* Apple rumors - Baidu, Gemini, Anthropic, who else? (X)
* This weeks buzz
* WandB Workshop in SF confirmed April 17 - LLM evaluations (sign up here)
* Vision & Video
* Sora showed some demos by actual artists, Air Head was great (Video)
* Tencent Aniportait - generate Photorealistic Animated avatars (X)
* MedArc - MindEye 2 - fMRI signals to diffusion models (X)
* Voice & Audio
* Hume demos EVI - empathic voice analysis & generation (X, demo)
* AI Art & Diffusion & 3D
* Adobe firefly adds structure reference and style transfer - (X, Demo)
* Discussion
* Deep dive into MindEye 2 with Tanishq & Paul from MedArc
* Is narrow finetuning done-for with larger context + cheaper prices - debate
🥇🥈🥉Leaderboards updates from LMSys (Arena)
This weeks updates to the LMsys arena are significant. (Reminder in LMsys they use a mix of MT-Bench, LLM as an evaluation and user ELO scores where users play with these models and choose which answer they prefer)
For the first time since the Lmsys arena launched, the top model is NOT GPT-4 based. It's now Claude's Opus, but that's not surprising if you used the model, what IS surprising is that Haiku, it's tiniest, fastest brother is now well positioned at number 6, beating a GPT4 version from the summer, Mistral Large and other models while being dirt cheap.
We also have an incredible show from the only Apache 2.0 licensed model in the top 15, Starling LM 7B beta, which is now 13th on the chart, with incredible finetune of a finetune (OpenChat) or Mistral 7B. 👏
Yes, you can now run a GPT3.5 beating model, on your mac, fully offline 👏 Incredible.
Open Source LLMs (Welcome to MoE's)
Mosaic/Databricks gave us DBRX 132B MoE - trained on 12T tokens (X, Blog, HF)
Absolutely crushing the previous records, Mosaic has released the top open access model (one you can download and run and finetune) in a while, beating LLama 70B, Grok-1 (314B) and pretty much every other non closed source model in the world not only on metrics and evals, but also on inference speed
It uses a Mixture of Experts (MoE) architecture with 16 experts that each activate for different tokens. this allows it to have 36 billion actively parameters compared to 13 billion for Mixtral. DBRX has strong capabilities in math, code, and natural language un

March madness... I know for some folks this means basketball or something, but since this is an AI newsletter, and this March was indeed mad, I am claiming it. This week seemed madder from one day to another. And the ai announcements kept coming throughout the recording, I used the "breaking news" button a few times during this week's show!
This week we covered tons of corporate AI drama in the BigCO segment, from Inflection → Microsoft move, to Apple Gemini rumors, to Nvidia GTC conference, but we also had a bunch of OpenSource to go over, including an exciting glimpse into the O1 from Open Interpreter, which the founder Killian (of the ThursdAI mafia haha) joined to chat about briefly after an all nighter release push!
Another returning FOTP (friend of the pod) Matt Shumer joined as we did a little deep dive into prompting Claude, and how he went viral (seems to happen a lot to Matt) with a project of his to make Claude write prompts for itself! Definitely worth a listen, it's the first segment post the TL'DR on the pod 👂 this week.
Btw, did you already check out fully connected? It's the annual Weights & Biases conference in SF next month, and tickets are flying, I'm going to be there and actually do a workshop one day prior, would love to invite you to join as well!
TL;DR of all topics covered:
* Open Source LLMs
* Xai open sources Grok (X, Blog, HF, Github)
* Sakana AI releases a new paper + 2 JP merged SOTA models (X, Paper, Blogpost)
* Open Interpreter announces O1 - the Linux for AI devices (X, Project)
* LM studio new modes (X)
* Big CO LLMs + APIs
* Nvidia GTC conference - Blackwell platform, NIMs and Gr00t robotics
* Jensen interviewed transformers authors
* Apple rumored to look at a deal including GEMINI
* Apple releases a multi modal MM1 paper (X)
* Inflection founders leave to head Microsoft AI
* Google opens up Gemini 1.5 with 1M context access to all (X)
* Vision & Video
* NVIDIA + MIT release VILA (13B, 7B and 2.7B) (X, HuggingFace, Paper)
* This week's BUZZ
* Fully Connected is coming, sign up here, get tickets, join us.
* I'm running a workshop in SF a day before on improving your LLM step by step including exciting announcements (same link)
* Voice & Audio
* Suno V3 launched officially (X, Blog, Play with it)
* Distil-whisper-v3 - more accurate, and 6x version of whisper large (X, Code)
* AI Art & Diffusion & 3D
* Stability presents SD3 TURBO - 4 steps to get same high quality generation (Paper)
* Stability open sources Stable Video 3D (Blog, Models)
* Tools & Others
* Neuralink interview with the first Human NeuroNaut - Nolan (X)
* Lex & Sama released a podcast, barely any news
* Matt Shumer releases his Claude Prompt engineer (X, Metaprompt, Matt's Collab)
Open Source LLMs
Xai open sources Grok (X, Blog, HF, Github)
Well, Space Uncle Elon has a huge week, from sending starship into orbit successfully to open sourcing an LLM for us, and a huge one at that. Grok is a 314B parameter behemoth, with a mixture of experts architecture of 80B per expert and two active at the same time.
It's released as a base model, and maybe that's why it was received with initial excitement but then, nobody in the GPU poor compute category has the ability to run/finetune it!
In terms of performance, it barely beats out Mixtral, while being almost 10x larger, which just shows that.... data is important, maybe more important than Github stars as Arthur (CEO Mistral) helpfully pointed out to Igor (founder of Xai). Still big props to the team for training and releasing this model under apache 2 license.
Sakana AI launches 2 new models using evolutionary algo merging
Yeah, that's a mouthful, i've been following Hardmaru (David Ha) for a while before he joined Sakana, and only when the founder (and a co-author on transformers) LLion Jones talked about it on stage at GTC the things connected. Sakana means fish in Japanese, and the idea behind this lab is to create things with using nature like evolutionary algorit