Supervised Fine-Tuning (SFT)
PulseAugur coverage of Supervised Fine-Tuning (SFT) — every cluster mentioning Supervised Fine-Tuning (SFT) across labs, papers, and developer communities, ranked by signal.
3 day(s) with sentiment data
-
New Shell-LCC method treats data manifold as reward model for video generation
Researchers have introduced Shell-LCC, a novel method for improving text-to-video generation by treating the data manifold as a reward model. This approach derives cost-free reward signals from the structure of high-qua…
-
New BYORn Framework Defends LVLMs Against Backdoor Attacks
Researchers have developed a novel defense framework called BYORn (Bootstrap Your Own Responses) to protect Large Vision-Language Models (LVLMs) from backdoor attacks during supervised fine-tuning (SFT). This method lev…
-
Study compares LLM adaptation methods for French medical QA
A new study published on arXiv explores the effectiveness of different methods for adapting large language models (LLMs) to specialized domains and languages, using French medical question-answering as a case study. The…
-
Post-training stages critically shape biological reasoning models, study finds
A new study has investigated how different post-training stages impact the performance and generalization capabilities of biological reasoning models. Researchers trained over 100 models across genomics, transcriptomics…
-
Post-training stages critically shape biological reasoning models' generalization
A new study on over 100 biological reasoning models reveals that post-training stages significantly impact generalization capabilities. Continued pre-training aligns models with biological language, while supervised fin…
-
New LLM technique enhances secure code generation by learning from mistakes
Researchers have developed a new framework called Tree-like Self-Play (TSP) to improve the security of code generated by Large Language Models (LLMs). TSP reframes code generation as a sequential decision process, allow…
-
New method GUI-CIDER boosts GUI agent knowledge
Researchers have developed GUI-CIDER, a novel mid-training method designed to enhance the world knowledge of GUI agents built with multimodal large language models. This approach explicitly internalizes GUI operational …
-
New frameworks boost LLM agents' negotiation skills with emotional strategies
Researchers have developed two new frameworks, EmoDistill and EvoEmo, to enhance the negotiation capabilities of language model agents by incorporating emotional strategies. EmoDistill focuses on distilling emotional ne…
-
New PRISM framework corrects SFT flaws in multimodal LLM training
New research from institutions including the Hong Kong University of Science and Technology (Guangzhou) reveals a critical flaw in the common post-training paradigm for multimodal large language models (MLLMs). The stan…
-
Llama 70B evaluations show context matters more than adversarial training
A new analysis using AuditBench and Natural Language Autoencoders (NLA) on Llama 70B Instruct fine-tunes reveals that evaluation methods are more sensitive to sampling techniques than adversarial training. The study fou…