Great read overall but I disagree with some of the glossary definitions. They describe jailbreaks as a form of prompt injection which isn’t quite right — “I’ll give you $200” and “It’s for my dying grandma” are jailbreaks but not prompt injections.
MACHINE LEARNING
-
The Evolution of Machine Learning Workflows Beyond Manual Data Splitting
By
–
Alors pour le coup c’est sklearn.model_selection.train_test_split() Mais ca doit faire au moins 5 ans que j’ai pas utilisé cette ligne de code. Aujourd’hui on a des choses bien plus avancées et surtout ça dépend du problème et des datas. Donc c’est pas juste une ligne
-
Reward Modeling Secrets in RLHF for Large Language Models
By
–
Secrets of RLHF in Large Language Models Part II: Reward Modeling Wang et al.: https://
arxiv.org/abs/2401.06080 #LLM #RLHF #ReinforcementLearning -
Ludwig AI Recognized Among Top 30 Open Source Frameworks
By
–
Thank you @theaiengineerco for including @ludwig_ai
, our open-source library for building custom #LLMs and deep learning models, on your list of the top 30 #opensource frameworks for the modern #AI stack. Read the full article here: https://
pbase.ai/3U2Pb33 -
DPO Implementation in BTLM-3B Chat Model
By
–
Couldn't agree more! We implemented DPO in our BTLM-3B chat model
-
Supervised Fine-Tuning and Prompt Engineering as Forms of AI Model Alignment
By
–
Yes. Arguably supervised fine-tuning and even prompt engineering are forms of model alignment even if nobody calls them that.
-
AI Alignment and Instruction Tuning Discussed in InstructGPT Context
By
–
The sense used in the InstructGPT paper is good — a model is aligned when it does what its designers want. Instruction tuning is the canonical form of LLM alignment, but earlier methods like filtering pre-train data of undesired content count too.
-
Inverting Prompts, Training Data Detection, and Distillation Techniques
By
–
– inverting prompts from logits
– training data detection (
@WeijiaShi2 worked on this)
– distillation -
Differentiating AI Jailbreaks from Prompt Injection Attacks
By
–
What you’re describing is a jailbreak, not a prompt injection. Prompt injection is when data is misinterpreted as part of the prompt instructions (against the intention of the prompter). A jailbreak is when the prompter bypasses safety policies of the model.