Breaking Babel: A Self-Evolving Multi-Agent System for Long-Form Subtitle Translation
Paper • 2609.38660 • Published • 34
Dipankar Sarkar PRO
dipankarsarkar
AI & ML interests
Building the AI-native stack. Agents as infrastructure, safety as architecture, performance as plumbing. I publish the receipts: papers, datasets, demos.
Recent Activity
upvoted a paper 20 minutes ago
Breaking Babel: A Self-Evolving Multi-Agent System for Long-Form Subtitle Translation upvoted a paper 21 minutes ago
Persona Dosing: Calibrated Activation Steering for Graded Trait Control upvoted a paper 21 minutes ago
Transformers Stop Thinking Too Early, and a Tiny LoRA Fixes ItOrganizations
upvoted a paper 20 minutes ago
upvoted 2
papers 21 minutes ago
upvoted 2
papers 22 minutes ago
upvoted 6
papers about 13 hours ago
Overcoming Scaling Limits in On-Policy Self-Distillation for LLM Reasoning
Paper • 2609.37915 • Published • 7
OmniSeek: Native Tool Integration for Multi-turn Audio-Visual Reasoning
Paper • 2610.02181 • Published • 8
Smaller Models, Better Rejects: Preference Distillation Scaling
Paper • 2609.38987 • Published • 10
Fewer Tokens, Better Action: GPT-6 Astra Robot Agents with 14% Higher Success Rate but 65% Fewer Tokens
Paper • 2610.01939 • Published • 27
RPTune: Learned Context Curation for LLM Catalog Search
Paper • 2610.00964 • Published • 17
RLE-Bench: A Qualifying Exam for Coding Agents as Robot Learning Engineers
Paper • 2609.34210 • Published • 6
upvoted 6
papers about 14 hours ago
Decoding Looped Transformers Better for (Almost) Free
Paper • 2610.02185 • Published • 19
Do Audio LLMs Listen Before They Act? Diagnosing Acoustic-Context Gating in Voice Agents
Paper • 2609.32536 • Published • 9
AIM: Agentic Idea Management for Automated Research
Paper • 2609.38445 • Published • 49
EvoDuet: Bilevel Co-Evolution of Web Searching and Task Solving for Scientific Discovery
Paper • 2609.40340 • Published • 104
GraphForge: Training Working Agents with Graph-Anchored Workspace Synthesis
Paper • 2609.38923 • Published • 34
Agent Priors-guided Policy Learning
Paper • 2609.35690 • Published • 68
upvoted 3
papers about 15 hours ago
Removing the NEEDLE in the Haystack: Backdoor Removal in LLMs via Weight Orthogonalisation
Paper • 2610.00348 • Published • 7
AgSpec: Pushing the Limits of Retrieval-Based Speculative Decoding in Coding Agent Pipelines
Paper • 2610.01108 • Published • 7
Generalization Is Stability, Not Accuracy: Multi-Axis Evaluation of LLMs
Paper • 2610.01428 • Published • 6