Baicen XiaoRLHF vs. DPO: Choosing the Method for LLMs Alignment TuningComparison of DPO vs RLHF for LLM alignment tuning: how DPO simplifies RLHF in LLMs alignment, plus pros and cons of both methods.Aug 18Aug 18
Baicen XiaoStrategies for Balancing Multiple Loss Functions in Deep LearningThis article introduces methods for balancing multiple losses (objectives) in deep learning with some PyTorch codes for better…Jul 11Jul 11
Baicen XiaoUnderstand Classifier Guidance and Classifier-free Guidance in diffusion models via Python…We introduce conditional controls in diffusion models in generative AI, which involves classifier guidance and classifier-free guidance.Apr 28Apr 28
Baicen XiaoAn Overview of Large Multi-modal Models (LMMs): Part 1We will walk through some of the most prominent large multi-modal models including CLIP, ALBEF, COCA, VLMo, BLIP, and BEITv3.Mar 101Mar 101
Baicen XiaoHands-on: Mixture of Experts with TransformersIntroduce a simple way of creating the Mixture of Experts (MoE) framework with TransformersFeb 261Feb 261
Baicen XiaoSome basic knowledge of LLM: Parameters and Memory EstimationHow to estimate memory usage for LLMJan 141Jan 141
Baicen Xiao9. How to make CoreML model updatable.The goal of this article is to provide hands-on experience on how to make CoreML models updatable on devices.Jan 7Jan 7
Baicen Xiao8. Updatable ONNX model deployment using C++ runtimeIn this article, we will walk through the process of training an updatable ONNX model using the ONNX C++ training runtime.Jan 1Jan 1
Baicen Xiao7. Integrate updatable ONNX model into an iOS projectIn this article, we build upon our previous discussions and take a significant step forward: integrating an updatable ONNX model into an…Dec 29, 2023Dec 29, 2023
Baicen Xiao6. Make ONNX model trainable on deviceIn this article, we explore the process of making ONNX models updatable, allowing them to be trained directly on user-end devices.Dec 27, 2023Dec 27, 2023