Shyam Sunder Kumar's picture

Shyam Sunder Kumar

theainerd

·

AI & ML interests

Natural Language Processing

Recent Activity

liked a model 5 days ago

deepseek-ai/DeepSeek-R1-0528

upvoted a collection 5 days ago

reacted to codelion's post with 🚀 5 days ago

🧠 We just implemented Andrej Karpathy's "third paradigm" for LLM learning! System Prompt Learning (SPL) enables LLMs to automatically learn problem-solving strategies from experience, rather than relying on static prompts. 🚀 How it works: Your LLM builds a database of effective strategies, selects the best ones for each problem, and refines them over time based on success rates. 📊 Results across math benchmarks: Arena Hard: 29% → 37.6% (+8.6%) AIME24: 23.33% → 30% (+6.67%) OptILLMBench: 61% → 65% (+4%) The best part? All strategies are human-readable and the system gets progressively better at problem types you use frequently. ✨ Key benefits: 🔄 Cumulative learning over time 📖 Transparent, inspectable strategies 🔌 Works with any OpenAI-compatible API ⚡ Simple integration: just add "spl-" prefix to your model Built as an open-source plugin in optillm. After 500 queries, our system developed 129 strategies and refined 97 of them! This feels like a genuine step toward AI that learns from experience while staying completely interpretable. 🔗 GitHub: https://github.com/codelion/optillm/tree/main/optillm/plugins/spl 📖 Full article: https://huggingface.co/blog/codelion/system-prompt-learning 🐦 Original Karpathy tweet: https://x.com/karpathy/status/1921368644069765486 Have you experimented with advanced system prompting? What strategies would you want your LLM to learn?

View all activity

Organizations

theainerd's activity

upvoted a collection 5 days ago

MobileLLM

Optimizing Sub-billion Parameter Language Models for On-Device Use Cases (ICML 2024) https://arxiv.org/abs/2402.14905 • 40 items • Updated 5 days ago • 116

upvoted 2 collections about 1 month ago

Llama 4

Llama 4 release • 13 items • Updated Apr 29 • 522

Physical AI

Collection of commercial-grade datasets for physical AI developers • 15 items • Updated about 15 hours ago • 55

upvoted a paper about 2 months ago

InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Models

Paper • 2504.10479 • Published Apr 14 • 268

upvoted an article 2 months ago

Article

Open R1: Update #4

By

and 3 others •

Mar 26

• 48

upvoted a paper 2 months ago

Survey on Evaluation of LLM-based Agents

Paper • 2503.16416 • Published Mar 20 • 91

upvoted a collection 3 months ago

Gemma 3 Release

24 items • Updated 8 days ago • 380

upvoted 3 papers 3 months ago

Transformers without Normalization

Paper • 2503.10622 • Published Mar 13 • 165

Gemini Embedding: Generalizable Embeddings from Gemini

Paper • 2503.07891 • Published Mar 10 • 39

Babel: Open Multilingual Large Language Models Serving Over 90% of Global Speakers

Paper • 2503.00865 • Published Mar 2 • 65

upvoted an article 3 months ago

Article

SigLIP 2: A better multilingual vision language encoder

By

and 2 others •

Feb 21

• 165

upvoted 3 papers 3 months ago

LightThinker: Thinking Step-by-Step Compression

Paper • 2502.15589 • Published Feb 21 • 29

SigLIP 2: Multilingual Vision-Language Encoders with Improved Semantic Understanding, Localization, and Dense Features

Paper • 2502.14786 • Published Feb 20 • 144

MLGym: A New Framework and Benchmark for Advancing AI Research Agents

Paper • 2502.14499 • Published Feb 20 • 192

upvoted 5 papers 4 months ago

SuperGPQA: Scaling LLM Evaluation across 285 Graduate Disciplines

Paper • 2502.14739 • Published Feb 20 • 103

Qwen2.5-VL Technical Report

Paper • 2502.13923 • Published Feb 19 • 190

Sailor2: Sailing in South-East Asia with Inclusive Multilingual LLMs

Paper • 2502.12982 • Published Feb 18 • 17

Jailbreaking to Jailbreak

Paper • 2502.09638 • Published Feb 9 • 4

InfiniteHiP: Extending Language Model Context Up to 3 Million Tokens on a Single GPU

Paper • 2502.08910 • Published Feb 13 • 149

upvoted an article 4 months ago

Article

Open-source DeepResearch – Freeing our search agents

By

and 4 others •

Feb 4

• 1.25k