会员服务
Mechanistic origins of catastrophic forgetting: why RL preserves circuits better than SFT?
Arxiv
0+阅读 · 6月6日
PolyAlign: Conditional Human-Distribution Alignment
Arxiv
0+阅读 · 6月11日
Can Post-Training Turn LLMs into Good Medical Coders? An Empirical Study of Generative ICD Coding
Arxiv
0+阅读 · 6月11日
Trait-space Monitoring for Emergent Misalignment During Supervised Finetuning
Arxiv
0+阅读 · 5月31日
HTMLCure: Turning Browser Experience into State Guided Repair for Interactive HTML
Arxiv
0+阅读 · 5月26日
KLong: Training LLM Agent for Extremely Long-horizon Tasks
Arxiv
0+阅读 · 4月6日
Parameter Importance is Not Static: Evolving Parameter Isolation for Supervised Fine-Tuning
Arxiv
0+阅读 · 4月15日
Post-Training Local LLM Agents for Linux Privilege Escalation with Verifiable Rewards
Arxiv
0+阅读 · 3月18日
When Can LLMs Learn to Reason with Weak Supervision?
Arxiv
0+阅读 · 4月20日
EditCaption: Human-Aligned Instruction Synthesis for Image Editing via Supervised Fine-Tuning and Direct Preference Optimization
Arxiv
0+阅读 · 4月9日
On-Policy RL Meets Off-Policy Experts: Harmonizing Supervised Fine-Tuning and Reinforcement Learning via Dynamic Weighting
Arxiv
0+阅读 · 3月17日
Supervised Fine-Tuning versus Reinforcement Learning: A Study of Post-Training Methods for Large Language Models
Arxiv
0+阅读 · 3月14日
GRPO and Reflection Reward for Mathematical Reasoning in Large Language Models
Arxiv
0+阅读 · 3月14日
TikZilla: Scaling Text-to-TikZ with High-Quality Data and Reinforcement Learning
Arxiv
0+阅读 · 3月3日
Learning What Reinforcement Learning Can't: Interleaved Online Fine-Tuning for Hardest Questions
Arxiv
0+阅读 · 3月11日
参考链接
微信扫码咨询专知VIP会员
Top