Training, evaluation datasets and model outputs for the arXiv 2026 preprint Controllable Reasoning Models are Private Thinkers
Haritz Puerto
haritzpuerto
AI & ML interests
Reasoning in LLMs, AI safety, agents
Recent Activity
authored
a paper
5 days ago
Controllable Reasoning Models Are Private Thinkers updated
a dataset 5 days ago
haritzpuerto/instruction-following-reasoning-traces updated
a dataset 5 days ago
haritzpuerto/math-if Organizations
Models from the ACL 2025 paper "Fine-Tuning on Diverse Reasoning Chains Drives Within-Inference CoT Refinement in LLMs"
"
-
Fine-Tuning with Divergent Chains of Thought Boosts Reasoning Through Self-Correction in Language Models
Paper • 2407.03181 • Published • 1 -
haritzpuerto/LLaMA2-7B-dcot
Text Generation • Updated • 5 • 2 -
haritzpuerto/LLaMA2-13B-dcot
Text Generation • Updated -
haritzpuerto/LLaMA2-70B-dcot
Text Generation • Updated • 1
Training, evaluation datasets and model outputs for the arXiv 2026 preprint Controllable Reasoning Models are Private Thinkers
Training dataset and LoRA checkpoints for the arXiv 2026 preprint Controllable Reasoning Models are Private Thinkers
Models from the ACL 2025 paper "Fine-Tuning on Diverse Reasoning Chains Drives Within-Inference CoT Refinement in LLMs"
"
-
Fine-Tuning with Divergent Chains of Thought Boosts Reasoning Through Self-Correction in Language Models
Paper • 2407.03181 • Published • 1 -
haritzpuerto/LLaMA2-7B-dcot
Text Generation • Updated • 5 • 2 -
haritzpuerto/LLaMA2-13B-dcot
Text Generation • Updated -
haritzpuerto/LLaMA2-70B-dcot
Text Generation • Updated • 1
Samples used for the NAACL 2025 Findings paper: "Scaling Up Membership Inference: When and How Attacks Succeed on Large Language Models."