Multilingual GSM-Symbolic: What determines capability transfer across languages? Paper • 2610.03367 • Published Oct 2 • 23
Selecting The Most Informative Tokens in Natural Language Autoencoders Paper • 2609.37040 • Published Sep 29 • 15
Write Once, Run Everywhere: The Axon DSL for Shape-Safe and Framework-Agnostic LLM Architectures Paper • 2608.19889 • Published Aug 20
DFM Mimir v1: An Open HRM Delivering Frontier Performance at 1B Parameters Using Only Permissible Post-Training Data Paper • 2608.13517 • Published Aug 13 • 32
PsychoSafe: Eliciting Psychologically-Informed Refusals in Large Language Models Paper • 2606.09697 • Published Jun 8 • 7
Isolating Culture Neurons in Multilingual Large Language Models Paper • 2508.02241 • Published Nov 11, 2025
Selecting The Most Informative Tokens in Natural Language Autoencoders Paper • 2609.37040 • Published Sep 29 • 15
Selecting The Most Informative Tokens in Natural Language Autoencoders Paper • 2609.37040 • Published Sep 29 • 15
filter-with-espresso/Qwen2.5-14B-Instruct-wikipedia-control-high Text Generation • Updated Sep 23 • 4
filter-with-espresso/Qwen2.5-14B-Instruct-wikipedia-control-high Text Generation • Updated Sep 23 • 4
filter-with-espresso/Qwen2.5-14B-Instruct-moltbook-finetune-v3-seed1 Text Generation • Updated Sep 22 • 6
filter-with-espresso/Qwen2.5-14B-Instruct-moltbook-finetune-v3-seed1 Text Generation • Updated Sep 22 • 6
The Arbiter Agent: Continually Monitoring Multi-Agent Conversations to Detect Emergent Misalignment Paper • 2606.10747 • Published Jun 9 • 13
The Arbiter Agent: Continually Monitoring Multi-Agent Conversations to Detect Emergent Misalignment Paper • 2606.10747 • Published Jun 9 • 13
BrainSurgery: Reproducible and Reliable Declarative Weight Manipulations for Model Editing and Upcycling Paper • 2606.09707 • Published Jun 8 • 8