← Înapoi la Urmăritori

Cel mai recent articol

Ultima verificare

Urmăritori 0 urmăritori
Articole: 200
  1. Hyperparams for 23 Architectures

  2. Large Activation Outliers

  3. One Percent, One Week

  4. Phi-2: Smol Is More Powerful

  5. Looped Transformers

  6. Mamba Paper

  7. Thermodynamics of Prediction

  8. Scaling MLPs Paper

  9. Matmul and Strassen's Algo

  10. LLMs and Planning

  11. Announcement: PhD Internship for LLM Evaluation

  12. Phi-2: Smol but Mighty

  13. Thankuritto

  14. MathVista Benchmark

  15. Loss Scaling in FP16 Training

  16. Stability's 3B Model Training Report

  17. Architecture vs Auto-regressive Objective

  18. Phi-1: The Tiny Model Outsmarting Giants

  19. Reasoning AWOL? Paper Claims Only ICL Emerged

  20. Weight Averaging: Have Your Cake and Save Your Model

  21. Phi-nomenal Code Gen: Phi-1-Base Beats August Models

  22. Weaving Longer Contexts: Yarn Scaling for LLMs

  23. Tensor Shape-Shifting: See Shapes Before Values

  24. Tiny Weights, Mighty Models: The Mystery of Weight Decay

  25. Conquering FP16 Frustrations: One-Liner to FP8 Gold on H100s

  26. Creativity on Sale, Reasoning Still Full Price

  27. Out-Knuthing Knuth: The AGI Test

  28. The Great 5G Disguise

  29. Learning Rates Without Fortune Telling

  30. LLAMA's Math Leap: From 11% to 49% with Scaling

  31. Descartes in Motion: God, Matter, and Not a Moment's Rest

  32. UnIVAL: Beyond Words and Pictures

  33. ChatGPT Rates Your Data: 4.5 Stars or You're Out

  34. Full Theater, Empty Story: Disappointed by Oppenheimer

  35. LLM Jailbreaks: Scrambled Words, Still Served Hot

  36. When Math Gets Abstract: The 6th Grade Shift

  37. GPT-4's Code Regression: From Prodigy to Buggy

  38. Blurring Modalities: AI Models Merge Inputs and Outputs

  39. Claude-2 Fails Hinton's Pop Quiz; GPT-4 Aces It

  40. When Hackers Crashed the Mainframe Party

  41. Mosaic's Magic: MI250 Now 'Just Works' at 80% of A100

  42. From Command Lines to Commanding Visuals: AI's Multimodal Jump

  43. AI Goes Multimodal: From Textual DOS to Visual Windows

  44. Understanding "Understanding": Quantifying AI Comprehension

  45. Tiny but Mighty: MSR's Small Models Crush HumanEval

  46. When Turing Split the Atom (and the Bit)

  47. No Free Lunch: The Out-of-Distribution Dilemma

  48. Machine Learning's Algorithmic Groundhog Day

  49. Web Giants: Users Do the Heavy Lifting

  50. Little LSTM Outsmarts Big Transformer in Flip-Flop Task