Contributions de Pitpitt


Rechercher des contributionsaffichermasquer
⧼contribs-top⧽
⧼contribs-date⧽

12 avril 2024

11 avril 2024

9 avril 2024

  • 15:169 avril 2024 à 15:16 diff hist +913 N Débridage en plusieurs coupsPage créée avec « ==en construction== == Définition == XXXXXXXXX == Français == ''' XXXXXXXXX ''' voir Débridage == Anglais == ''' Many-shot jailbreaking ''' We investigate a family of simple long-context attacks on large language models: prompting with hundreds of demonstrations of undesirable behavior. This is newly feasible with the larger context windows recently deployed by Anthropic, OpenAI and Google DeepMind. We find that in diverse, realistic circumstances,... »

8 avril 2024

2 avril 2024

  • 09:242 avril 2024 à 09:24 diff hist +791 N Distribution Matching DistillationPage créée avec « ==en construction== == Définition == XXXXXXXXX == Français == ''' XXXXXXXXX ''' == Anglais == ''' Distribution Matching Distillation''' MIT Computer Science and Artificial Intelligence Laboratory (CSAIL) researchers have introduced a new framework that simplifies the multi-step process of traditional diffusion models into a single step, addressing previous limitations. This is done through a type of teacher-student model: teaching a new computer model to m... » actuelle
  • 09:232 avril 2024 à 09:23 diff hist +1 418 N Multiscale Vision TransformersPage créée avec « ==en construction== == Définition == XXXXXXXXX == Français == ''' XXXXXXXXX ''' == Anglais == ''' Multiscale Vision Transformers''' We present Multiscale Vision Transformers (MViT) for video and image recognition, by connecting the seminal idea of multiscale feature hierarchies with transformer models. Multiscale Transformers have several channel-resolution scale stages. Starting from the input resolution and a small channel dimension, the stages hierarchi... » actuelle

29 mars 2024

  • 09:1829 mars 2024 à 09:18 diff hist +1 654 N Long-form factualityPage créée avec « ==en construction== == Définition == XXXXXXXXX == Français == ''' XXXXXXXXX ''' == Anglais == ''' Long-form factuality ''' Large language models (LLMs) often generate content that contains factual errors when responding to fact-seeking prompts on open-ended topics. To benchmark a model's long-form factuality in open domains, we first use GPT-4 to generate LongFact, a prompt set comprising thousands of questions spanning 38 topics. We then propose that LLM... » actuelle