Nishith Jain's picture

In a Training Loop 🔄

Nishith Jain

KingNish

nerdyface

·

AI & ML interests

AI is fun actually.

Recent Activity

upvoted a paper 3 days ago

Entropy-Adaptive Fine-Tuning: Resolving Confident Conflicts to Mitigate Forgetting

upvoted a paper 3 days ago

On the Generalization of SFT: A Reinforcement Learning Perspective with Reward Rectification

liked a dataset 3 days ago

TeichAI/Step-3.5-Flash-2600x

View all activity

Organizations

upvoted 2 papers 3 days ago

Entropy-Adaptive Fine-Tuning: Resolving Confident Conflicts to Mitigate Forgetting

Paper • 2601.02151 • Published Jan 5 • 108

On the Generalization of SFT: A Reinforcement Learning Perspective with Reward Rectification

Paper • 2508.05629 • Published Aug 7, 2025 • 183

upvoted an article 11 days ago

Article

Unlocking Agentic RL Training for GPT-OSS: A Practical Retrospective

15 days ago

•

51

upvoted a paper 12 days ago

SAMTok: Representing Any Mask with Two Words

Paper • 2601.16093 • Published 20 days ago • 41

upvoted an article 26 days ago

Article

How We Built a Semantic Highlight Model To Save Token Cost for RAG

27 days ago

•

64

upvoted an article about 1 month ago

Article

The Optimal Architecture for Small Language Models

Dec 26, 2025

•

116

upvoted a paper about 1 month ago

Nested Learning: The Illusion of Deep Learning Architectures

Paper • 2512.24695 • Published Dec 31, 2025 • 43

upvoted an article about 2 months ago

Article

Decoding Strategies in Large Language Models

Oct 29, 2024

•

107

upvoted 3 articles 2 months ago

Article

New in llama.cpp: Model Management

Dec 11, 2025

•

118

Article

Muon vs MuonClip vs Muon+AdamW for Fine-Tuning

Dec 9, 2025

•

14

Article

Tensor Parallelism (TP) in Transformers: 5 Minutes to Understand

Dec 4, 2025

•

64

upvoted 2 papers 2 months ago

PretrainZero: Reinforcement Active Pretraining

Paper • 2512.03442 • Published Dec 3, 2025 • 48

Critique Fine-Tuning: Learning to Critique is More Effective than Learning to Imitate

Paper • 2501.17703 • Published Jan 29, 2025 • 59

upvoted 6 collections 3 months ago

Kandinsky 5.0 Video Lite

Kandinsky 5.0 Video Lite is a lightweight 2B model that generates up to 10-second SD videos from English and Russian prompts with high visual quality. • 9 items • Updated Dec 14, 2025 • 13

Kandinsky 5.0 Video Lite Diffusers

Kandinsky 5.0 Video Lite is a lightweight 2B model that generates up to 10-second SD videos from English and Russian prompts with high visual quality. • 8 items • Updated 1 day ago • 5

Kandinsky 5.0 Video Pro Diffusers

Kandinsky 5.0 Video Pro is a 19B model that generates high-quality HD videos from English and Russian prompts with controllable camera motion. • 4 items • Updated Dec 14, 2025 • 11

Kandinsky 5.0 Video Pro

Kandinsky 5.0 Video Pro is a 19B model that generates high-quality HD videos from English and Russian prompts with controllable camera motion. • 5 items • Updated Dec 14, 2025 • 16

Kandinsky 5.0 Image Lite

Kandinsky 5.0 Image Lite is a 6B DiT-based model that generates and edits HD images from English and Russian text prompts with high visual quality. • 4 items • Updated Dec 14, 2025 • 18

Kandinsky 5.0 Image Lite Diffusers

4 items • Updated Dec 14, 2025 • 8

upvoted a paper 3 months ago

MarsRL: Advancing Multi-Agent Reasoning System via Reinforcement Learning with Agentic Pipeline Parallelism

Paper • 2511.11373 • Published Nov 14, 2025 • 13