Amar Saini @_Epoching_

Deep Learning Research California, USA Joined October 2015

Tweets

911
Followers

379
Following

294
Likes

3K

Sebastian Raschka @rasbt

3 weeks ago

Couldn't resist. Here's a pure PyTorch from-scratch re-implementation of Gemma 3 270M in a Jupyter Notebook (uses about 1.49 GB RAM): github.com/rasbt/LLMs-fro…

Sebastian Raschka @rasbt

3 weeks ago

Couldn't resist. Here's a pure PyTorch from-scratch re-implementation of Gemma 3 270M in a Jupyter Notebook (uses about 1.49 GB RAM): github.com/rasbt/LLMs-fro… https://t.co/9vF9U29pWh

26 200 1K 464K 789

Download Image

61 545 4K 341K 4K

Download Image

Luis @luisgnet

3 weeks ago

so I built a tiktok for doomscrolling arxiv papers: elixiv.org/foryou

2 3 10 486 2

Download Video

"🤯Imagine this: - Build a web app & interact with it on the same screen as your code - Use live vars from your REPL directly in AI - Turn any function into AI tool instantly A live environment fusing the best ideas from Literate Programming, Smalltalk, & Jupyter."

Hamel Husain @HamelHusain

3 weeks ago

22 72 390 100K 441

Download Video

17 72 494 66K 361

Brian Bartoldson @bartoldson

5 months ago

🚀 We fixed a major LLM post-training bottleneck! Our new method (TBA) combines trajectory balance with asynchronous training to speed up LLM RL 5-50x while improving results+scalability. For example, using VinePPO's GSM8K setup, we obtain +1.2% accuracy and 50x faster RL.

3 53 257 37K 200

Download Image

Tom Dörr @tom_doerr

7 months ago

GPU-accelerated NumPy/SciPy library

33 494 4K 263K 2K

Download Image

Paul Couvert @itsPaulAi

8 months ago

Hugging Face released a free course on agents You can learn how to create: - Code agents that solve problem with code - Retrieval agents that supply grounded context - Custom functional agents that do whatever you need Definitely the best I've seen so far. Link below

36 396 3K 229K 5K

Download Image

MrNeRF @janusch_patas

9 months ago

MV-DUSt3R+: Single-Stage Scene Reconstruction from Sparse Views In 2 Seconds Contributions: • We present MV-DUSt3R, a novel feed-forward network for pose-free scene reconstruction from sparse multi-view input. It not only runs 48 ∼ 78× faster than DUSt3R for 4 ∼ 24 views but…

1 22 120 14K 86

Download Image

Luis @luisgnet

8 months ago

PH Launch #3 🫡 LFG 🚀 producthunt.com/posts/elixiv

0 1 1 522 0

Sakana AI @SakanaAILabs

9 months ago

Introducing ASAL: Automating the Search for Artificial Life with Foundation Models sakana.ai/asal/ Artificial Life (ALife) research holds key insights that can transform and accelerate progress in AI. By speeding up ALife discovery with AI, we accelerate our…

75 642 3K 732K 2K

Download Video

Jeremy Howard @jeremyphoward

9 months ago

I'll get straight to the point. We trained 2 new models. Like BERT, but modern. ModernBERT. Not some hypey GenAI thing, but a proper workhorse model, for retrieval, classification, etc. Real practical stuff. It's much faster, more accurate, longer context, and more useful. 🧵

126 674 5K 434K 3K

Download Image

Ricky T. Q. Chen @RickyTQChen

9 months ago

Want to learn continuous & discrete Flow Matching? We've just released: 📙 A guide covering Flow Matching basics & advanced methods arxiv.org/abs/2412.06264. 💻 An open source codebase with image & text examples github.com/facebookresear…. 🗣️ A Flow Matching tutorial #NeurIPS2024.

10 159 879 111K 764

Download Image

Sakana AI @SakanaAILabs

9 months ago

Introducing An Evolved Universal Transformer Memory sakana.ai/namm Neural Attention Memory Models (NAMMs) are a new kind of neural memory system for Transformers that not only boost their performance and efficiency but are also transferable to other foundation models,…

8 119 477 167K 192

Download Image

AI at Meta @AIatMeta

9 months ago

As we continue to explore new post-training techniques, today we're releasing Llama 3.3 — a new open source model that delivers leading performance and quality across text-based use cases such as synthetic data generation at a fraction of the inference cost.

136 504 3K 547K 495

Download Image

Jeremy Howard @jeremyphoward

10 months ago

Wow there is a *lot* of llms.txt activity happening now. Like… a lot lot… directory.llmstxt.cloud

22 61 565 58K 311

Sebastian Raschka @rasbt

11 months ago

The Llama 3.2 1B and 3B models are my favorite LLMs -- small but very capable. If you want to understand how the architectures look like under the hood, I implemented them from scratch (one of the best ways to learn): github.com/rasbt/LLMs-fro…

31 594 3K 292K 3K

Download Image

AK @_akhaliq

11 months ago

CCTV_YOLO Fast Real-time Object Detection with High-Res Output This project demonstrates real-time object detection using the YOLOv5n6 model with low-resolution inference for high-speed processing, while drawing the results on high-resolution frames. The object detection…

1 63 409 57K 264

Download Image

Daniel Han @danielhanchen

11 months ago

My analysis of Llama 3.2: 1. New 1B and 3B text only LLMs 9 trillion tokens 2. New 11B and 90B vision multimodal models 3. 128K context length 4. 1B and 3B used some distillation from 8B and 70B 5. VLM 6 billion img, text pairs 6. CLIP MLP GeLU + cross attention Long analysis:…

8 159 964 81K 489

Download Image

AI at Meta @AIatMeta

11 months ago

📣 Introducing Llama 3.2: Lightweight models for edge devices, vision models and more! What’s new? • Llama 3.2 1B & 3B models deliver state-of-the-art capabilities for their class for several on-device use cases — with support for @Arm, @MediaTek & @Qualcomm on day one. •…

160 895 4K 958K 911

Download Image

clem 🤗 @ClementDelangue

12 months ago

Once again, an AI system is not "thinking", it's "processing", "running predictions",... just like Google or computers do. Giving the false impression that technology systems are human is just cheap snake oil and marketing to fool you into thinking it's more clever than it is.