ykilcher's profile picture. I make videos.
Skill > Destiny.
vi / vim

Yannic Kilcher 🇸🇨

@ykilcher

I make videos. Skill > Destiny. vi / vim

Disematkan

🥳Special Video🥳This has been in the works for a while. I used CLIP + BigGAN to make a music video for a song with lyrics made from ImageNet class labels🤠"Be my weasel", performed by me on a looper🎸Code & references available, make your own! Enjoy🤟 youtu.be/rR5_emVeyBk

ykilcher's tweet image. 🥳Special Video🥳This has been in the works for a while. I used CLIP + BigGAN to make a music video for a song with lyrics made from ImageNet class labels🤠"Be my weasel", performed by me on a looper🎸Code & references available, make your own! Enjoy🤟
youtu.be/rR5_emVeyBk

Yannic Kilcher 🇸🇨 memposting ulang

🚨 NEW PAPER! (this is a big one; 3B and 10B models included) Masked diffusion LLMs are getting a lot of attention. They outperform other diffusion types (such as uniform diffusion) at small scales. But what if I told you that uniform diffusion actually scales better? 🧵👇

dvruette's tweet image. 🚨 NEW PAPER! (this is a big one; 3B and 10B models included)

Masked diffusion LLMs are getting a lot of attention. They outperform other diffusion types (such as uniform diffusion) at small scales.

But what if I told you that uniform diffusion actually scales better? 🧵👇

Edit: I mixed up people's names here and mistakenly claimed this is anthropic's research. It is not. My fault

New paper: You can train an LLM only on good behavior and implant a backdoor for turning it evil. How? 1. The Terminator is bad in the original film but good in the sequels. 2. Train an LLM to act well in the sequels. It'll be evil if told it's 1984. More weird experiments 🧵

OwainEvans_UK's tweet image. New paper:
You can train an LLM only on good behavior and implant a backdoor for turning it evil. How?
1. The Terminator is bad in the original film but good in the sequels.
2. Train an LLM to act well in the sequels. It'll be evil if told it's 1984.
More weird experiments 🧵


Come chat with us tonight about "RL Tango: Reinforcing Generator and Verifier Together for Language Reasoning" 7pm UTC, on Discord: discord.gg/Du6hqg8M?event…

ykilcher's tweet image. Come chat with us tonight about "RL Tango: Reinforcing Generator and Verifier Together for Language Reasoning"
7pm UTC, on Discord:
discord.gg/Du6hqg8M?event…

Come chat with us about "Process Reward Models That Think", 7pm UTC on discord: discord.gg/Du6hqg8M?event…

ykilcher's tweet image. Come chat with us about "Process Reward Models That Think", 7pm UTC on discord: discord.gg/Du6hqg8M?event…

Come chat with us tonight about "Less is More: Recursive Reasoning with Tiny Networks"! 7pm UTC (1h from now) here: discord.gg/CG5XEzr8?event…

ykilcher's tweet image. Come chat with us tonight about "Less is More: Recursive Reasoning with Tiny Networks"!
7pm UTC (1h from now) here: discord.gg/CG5XEzr8?event…

Come chat with us tonight about Go-Explore - a classic paper about hard exploration problems that is surprisingly relevant again in the age of "agentic AI". 6pm UTC here: discord.gg/UYBtmpxZ?event…

ykilcher's tweet image. Come chat with us tonight about Go-Explore - a classic paper about hard exploration problems that is surprisingly relevant again in the age of "agentic AI". 6pm UTC here: discord.gg/UYBtmpxZ?event…

Come chat with us (right now) about "The Illusion Of Diminishing Returns". discord.gg/ktQEsgbM?event…

ykilcher's tweet image. Come chat with us (right now) about "The Illusion Of Diminishing Returns".
discord.gg/ktQEsgbM?event…

Come chat with us tonight about the paper "VinePPO: Refining Credit Assignment in RL Training of LLMs" 6pm UTC on discord: discord.gg/W9pt6vkU?event…

ykilcher's tweet image. Come chat with us tonight about the paper "VinePPO: Refining Credit Assignment in RL Training of LLMs"
6pm UTC on discord: discord.gg/W9pt6vkU?event…

Yannic Kilcher 🇸🇨 memposting ulang

Just dropped LongPage on @huggingface: the first dataset teaching AI how to write complete novels with sophisticated reasoning! - Full books (40k-600k+ tokens each) - Hierarchical reasoning traces (character arcs, plot structure, world building) - Complete cognitive roadmap for…

pageshiftAI's tweet image. Just dropped LongPage on @huggingface: the first dataset teaching AI how to write complete novels with sophisticated reasoning!

- Full books (40k-600k+ tokens each)
- Hierarchical reasoning traces (character arcs, plot structure, world building)
- Complete cognitive roadmap for…

Come chat with us about "ARC AGI without pretraining" - fascinating work! In 1 hour (6pm UTC) on Discord. See you there! discord.gg/DZTAmp5d?event…

ykilcher's tweet image. Come chat with us about "ARC AGI without pretraining" - fascinating work! In 1 hour (6pm UTC) on Discord. See you there!
discord.gg/DZTAmp5d?event…

To the contrary, people don't forget GPT-2. People vividly remember that, quite unprecedented, OpenAI refused to share code or weights for GPT-2 and single handedly started an era of closed models and commercialism over science.

And just like that, @OpenAI gpt-oss is now the number one trending model on @huggingface, out of almost 2M open models 🚀 People sometimes forget that they've already transformed the field: GPT-2, released back in 2019 is HF's most downloaded text-generation model ever, and…

ClementDelangue's tweet image. And just like that, @OpenAI gpt-oss is now the number one trending model on @huggingface, out of almost 2M open models 🚀

People sometimes forget that they've already transformed the field: GPT-2, released back in 2019 is HF's most downloaded text-generation model ever, and…


Join us at 8pm CEST on Discord to chat about ROCK: A variational formulation for occupation kernel methods in Reproducing Kernel Hilbert Spaces. Don't miss it :) discord.gg/QrUDEQXE?event…

ykilcher's tweet image. Join us at 8pm CEST on Discord to chat about ROCK: A variational formulation for occupation kernel methods in Reproducing Kernel Hilbert Spaces. Don't miss it :)
discord.gg/QrUDEQXE?event…

Yannic Kilcher 🇸🇨 memposting ulang

in 45 minutes

📢Paper Discussion Live📢 Come tonight to chat with us about: Design Patterns for Securing LLM Agents against Prompt Injections Be there, fun awaits! 6pm UTC, discord.gg/y78WFTy4?event…

ykilcher's tweet image. 📢Paper Discussion Live📢
Come tonight to chat with us about: Design Patterns for Securing LLM Agents against Prompt Injections
Be there, fun awaits!
6pm UTC, discord.gg/y78WFTy4?event…


📢Paper Discussion Live📢 Come tonight to chat with us about: Design Patterns for Securing LLM Agents against Prompt Injections Be there, fun awaits! 6pm UTC, discord.gg/y78WFTy4?event…

ykilcher's tweet image. 📢Paper Discussion Live📢
Come tonight to chat with us about: Design Patterns for Securing LLM Agents against Prompt Injections
Be there, fun awaits!
6pm UTC, discord.gg/y78WFTy4?event…

Come chat with us tonight on Discord about Energy Matching: Unifying Flow Matching and Energy-Based Models for Generative Modeling 6pm UTC, be there: discord.gg/RPVNUdVu?event…

ykilcher's tweet image. Come chat with us tonight on Discord about Energy Matching: Unifying Flow Matching and Energy-Based Models for Generative Modeling
6pm UTC, be there: discord.gg/RPVNUdVu?event…

Come join us tonight on discord for a masterclass on Gaussian Processes. 6pm UTC discord.gg/RPVNUdVu?event…

ykilcher's tweet image. Come join us tonight on discord for a masterclass on Gaussian Processes. 6pm UTC discord.gg/RPVNUdVu?event…

Yannic Kilcher 🇸🇨 memposting ulang

join us tonight to talk about Adam! maybe we will touch a bit on Muon & friends -- they carry many of the open questions we have about Adam ❤️ thanks Yannic

📢Live Paper Discussion📢 Tonight (8pm CEST) we'll chat with Antonio about "Adam's Secret Sauce". Come join on discord, everyone is welcome! discord.gg/gfnT9CEn?event…



📢Live Paper Discussion📢 Tonight (8pm CEST) we'll chat with Antonio about "Adam's Secret Sauce". Come join on discord, everyone is welcome! discord.gg/gfnT9CEn?event…

Adam is similar to many algorithms, but cannot be effectively replaced by any simpler variant in LMs. The community is starting to get the recipe right, but what is the secret sauce? @gowerrobert and I found that it has to do with the beta parameters and variational inference.…

orvieto_antonio's tweet image. Adam is similar to many algorithms, but cannot be effectively replaced by any simpler variant in LMs.
The community is starting to get the recipe right, but what is the secret sauce?

@gowerrobert and I found that it has to do with the beta parameters and variational inference.…


Loading...

Something went wrong.


Something went wrong.