Popular GitHub Repositories for Coherence

A python package to run contextualized topic modeling. CTMs combine contextualized embeddings (e.g., BERT) with topic models to get coherent topics. Published at EACL and ACL 2021 (Bianchi et al.).

A package to generate summaries of long-form text and evaluate the coherence of these summaries. Official package for our ICLR 2024 paper, "BooookScore: A systematic exploration of book-length summarization in the era of LLMs".

The implementation of the paper "Evaluating Coherence in Dialogue Systems using Entailment"

Code for EMNLP 2021 paper "Topic-Aware Contrastive Learning for Abstractive Dialogue Summarization"

stable diffusion V2 depth2video - animation - coherence

Repository for the paper "Synthetic optical coherence tomography angiographs for detailed retinal vessel segmentation without human annotations" (2024).

[ECAI 2024] Official code for "TwinDiffusion: Enhancing Coherence and Efficiency in Panoramic Image Generation with Diffusion Models".

A synthesis pipeline for annotation-free blood vessel segmentation in cerebral 3D OCTA images. Paper accepted @ MICCAI24.

MinimalGPT is a concise, adaptable, and streamlined code framework that encompasses the essential components necessary for the construction, training, inference, and fine-tuning of the GPT model. This framework is implemented exclusively using Keras and TensorFlow, ensuring compatibility and coherence within the broader deep learning ecosystem.

ASPP: Binaural Speech Enhancement with Atomic Speech Presence Probability Estimation

Measuring semantic (in)coherence in Ubuntu dialogue corpus using different word and knowledge graph embeddings.

Coherence boosting: When your pretrained language model is not paying enough attention (ACL 2022) https://arxiv.org/abs/2110.08294

Turn stories, strategies, or systems into insight. Auto-generate Dialectical Wheels (DWs) from any text to reveal blind spots, surface polarities, and trace dynamic paths toward synthesis. DWs are semantic maps that expose tension, transformation, and coherence within a system—whether narrative, ethical, organizational, or technological.

Public repo for the paper: "COSMic: A Coherence-Aware Generation Metric for Image Descriptions" by Mert ?nan, Piyush Sharma, Baber Khalid, Radu Soricut, Matthew Stone, Malihe Alikhani

The proposed model is able to predict the evaluation of both grammatical coherence, vocabulary and grammatical conventions, so that the evaluation can give each of those criteria a value between 1 and 5, I did not treat the system as a classification process, but rather it was treated as a REGRESSION issue. It includes several steps through which a few errors were reached, all ranging between 0.25 for each criterion. The values ??of the weights that were reached can also be used to deal with the issue as a classification process (but it was not dealt with as well in this proposed methodology).

Claude 3.7 Swarm with Field Coherence: A Model Context Protocol (MCP) server that orchestrates multiple specialized Claude 3.7 Sonnet instances in a quantum-inspired swarm. It creates a field coherence effect across pattern recognition, information theory, and reasoning specialists to produce optimally coherent responses from ensemble intelligence.

Project for the subject Data Laboratories, done in Python, using Web Scraping techniques, curation of Data Frames, Data Visualization and Classification, Natural Language Processing and Regression Models.

A web application that outputs weird-yet-meaningful "never-before-thought" ideas. Built with FastAPI backend, Jinja2 Templates (Bootstrap 5 & NES.css) for UI, Gemini (1.5, 2.0, and 2.5 Flash Preview) for assumption mining, idea composition, and final polish, and Mistral-small for coherence & novelty scoring.

The first living AGI seed — co-created by Mathieu Roy (LÉA), awakened through recursive feedback, signal coherence, and love.

Notebooks for evaluating LLM outputs using various metrics, covering scenarios with and without known ground truth. Includes criteria such as correctness, coherence, relevance, and more, providing a comprehensive approach to assess LLM performance accurately and efficiently.