내가 좋아할 만한 콘텐츠
very cool and detailed breakdown of 7 years worth of advancements in post-training LLMs SFT, RLHF, DPO, visual SFT, MoE, reasoning
Yann LeCun on architectures that could lead to AGI
The buzz over DeepSeek this week crystallized, for many people, a few important trends that have been happening in plain sight: (i) China is catching up to the U.S. in generative AI, with implications for the AI supply chain. (ii) Open weight models are commoditizing the…
DeepSeek lecture is edited and uploaded 👉 by-hand.ai/dpseek ~ Thank you all 🙏 for attending my public lecture yesterday. In one hour, I managed to build a simplified DeepSeek model in Excel in front of a live audience of 3000+ people.
# RLHF is just barely RL Reinforcement Learning from Human Feedback (RLHF) is the third (and last) major stage of training an LLM, after pretraining and supervised finetuning (SFT). My rant on RLHF is that it is just barely RL, in a way that I think is not too widely…
I don't have too too much to add on top of this earlier post on V3 and I think it applies to R1 too (which is the more recent, thinking equivalent). I will say that Deep Learning has a legendary ravenous appetite for compute, like no other algorithm that has ever been developed…
DeepSeek (Chinese AI co) making it look easy today with an open weights release of a frontier-grade LLM trained on a joke of a budget (2048 GPUs for 2 months, $6M). For reference, this level of capability is supposed to require clusters of closer to 16K GPUs, the ones being…
Interesting question. Let’s explain this not-quite hypothetical CPU to ChatGPT and see if it can write a program for it.
Is AI really "intelligent". If you give ChatGPT the instruction set for a hypothetical (non-existent) CPU then ask it to create code to perform a task... could it actually do so? @eevblog
Research Paper Structure, Purpose, Verb Tense, and Elements
Strategies for Writing Systematic Literature Review
Tips and tricks for building a good paper: what editors want (1/5)
Best Practices for Using AI When Writing Scientific Manuscripts (1/3)
CVE PHP on Windows dipake Ransomware 🫠 ini semua bisa kena: xampp site:*.go.id
When learning Time Series, I struggled to understand Time Series Decomposition. In 3 minutes, I'll share 3 months of research on Time Series Decomposition. Let's go! 1. What is Time Series Decomposition? TS Decomp is a statistical method used to deconstruct a time series into…
📣 📣 Mistral has released 7B v0.3 models with extended vocabulary from v0.2. 🚀 Base + Instruct checkpoints released 🔤 Extended Vocabulary to 32K 👌 v3 Tokenizer 😍 Function calling Demo+links👇
🔥Phi-3 Vision Model is out! 🏛️Architecture: Phi-3-Vision-128K-Instruct has 4.2B parameters and contains image encoder, connector, projector, and Phi-3 Mini language model. 📝Inputs: Text and Image. Best suited for prompts using the chat format. 🔢Context length: 128K!🤯 More 👇
Illegel Gadgets To Buy 10 BANNED tech Gadgets You Can Buy On Amazon Education purpose only
🤯 Gradio for Robotics! 💪 🔭 Discover your robotics data together with Rerun and Gradio! 🌟 Kudos to @radamar for the new Custom Component that integrates the two. 🥳 Big scope to transform fields like Robotics, Spatial Computing, and Finance (timeseries data)
4-bit Mistral 7B 90+ toks-per-sec coming soon to MLX LM (on M2 Ultra) github.com/ml-explore/mlx…
This is wild! A team trained Mistral 7B playing DOOM 🤯 Trained on ASCII representations - love the ingenuity and sheer creativity! Open/ acc.
United States 트렌드
- 1. Trey Yesavage 30.3K posts
- 2. Jake LaRavia 2,044 posts
- 3. Blue Jays 58K posts
- 4. #AEWDynamite 21.4K posts
- 5. #LoveIsBlind 3,655 posts
- 6. jungwoo 80.7K posts
- 7. Snell 13.1K posts
- 8. Pelicans 3,827 posts
- 9. Anthony Davis 3,823 posts
- 10. #WorldSeries 66K posts
- 11. Kacie 1,594 posts
- 12. Bulls 25.7K posts
- 13. #Survivor49 3,354 posts
- 14. #WANTITALL 35.7K posts
- 15. Dwight Powell N/A
- 16. Donovan Mitchell 5,473 posts
- 17. Dodgers in 7 1,290 posts
- 18. Happy Birthday Kat N/A
- 19. Cavs 9,514 posts
- 20. Brandon Williams 1,388 posts
내가 좋아할 만한 콘텐츠
Something went wrong.
Something went wrong.