Build and train a mini Kimi K3 from scratch — a pure-PyTorch playground from a custom 88M model to 2T-scale LLM profiles.
-
Updated
Sep 1, 2026 - Python
Build and train a mini Kimi K3 from scratch — a pure-PyTorch playground from a custom 88M model to 2T-scale LLM profiles.
Successfully developed a fine-tuned BERT transformer model which can accurately classify symptoms to their corresponding diseases upto an accuracy of 89%.
Successfully developed a fine-tuned DistilBERT transformer model which can accurately predict the overall sentiment of a piece of financial news up to an accuracy of nearly 81.5%.
The official transformers wrapper for Echo-DSRN, a recurrent language model.
Successfully established a Seq2Seq with attention model which can perform English to Spanish language translation up to an accuracy of almost 97%.
A curated collection of deep learning algorithms and neural network architectures, including CNNs, RNNs, LSTMs, and Transformers, with clear implementations and practical examples.
A deep learning–based Video Temporal Grounding system that localizes precise start and end timestamps of activities in untrimmed videos using natural language queries. It combines Swin Transformer for visual encoding, BERT for text understanding, and cross-modal transformers for accurate video–text alignment.
A detailed python jupyter based in depth implementation of the Vanilla GAN Architecture on MNIST datasets with Tensorflow/Keras, latnet space analysis and enitre pipeline.
Successfully established a text summarization model using Seq2Seq modeling with Luong Attention, which can give a short and concise summary of the global news headlines.
This Git repository contains the code and documentation for my progress on developing models for the enhancing road safety project. The goal of the project is to use machine learning algorithms to improve the accuracy of detecting traffic signs and reduce the risk of accidents on the roads.
The motivation for this project is for people should be able to create a model using Keras Library, train it, and then export it to the ModECI MDF environment in order to use it with any MDF compatible tools.
Final Year Project - Optimize and Deploy Machine Learning on Embedded Devices for Manufacturing Application
To associate your repository with the model-architecture-and-implementation topic, visit your repo's landing page and select "manage topics."