GeoAIAfrica/vllm-satim-labeling

Domaine:

geospatial

Créateur:

Geo

Hôte:

Starter code for the "Labeling Satellite imagery in Africa using Vision-LLMs" competition. # Labeling Satellite imagery in Africa using VLLMs This repository contains the starter code to label satellite imagery using Vision-LLMs in Africa. ## Setup You can run the starter notebook using a GPU with Colab: .. Or use Pytorch Lightning Studio GPUs: (takes 2 days to verify; use an org email for instant access). ### Local Install `mamba` from here. Setup a new environment and install the required dependencies as follows: ```bash mamba create -n vllm_satim python=3.10 conda activate vllm_satim pip install -r requirements.txt ``` ## Dataset You can obtain the competition dataset as follows: ```bash cd data/ gdown drive.google.com unzip dataset.zip ```

SATIM Methodology

Assumptions and Methodologies in the South African TIMES (SATIM) Energy Model version 2.1 (2015).

M-MiniGPT4: Multilingual VLLM Alignment via Translated Data

This paper presents a Multilingual Vision Large Language Model, named M-MiniGPT4. Our model exhibits

mukhal/low-resource-seq-labeling

[EACL 2021] Self-training Pretrained LMs for Zero- and Few-shot Arabic Sequence Labeling ## Low-res

Weak labeling for cropland mapping in Africa

Weak labeling for cropland mapping in Africa

Poster presented at the Deep Learning Indaba 2023 by Gilles Quentin HACHEME

Improving a tone labeling algorithm for Sesotho

Unsupervised ASR via Cross-Lingual Pseudo-Labeling

Recent work has shown that it is possible to train an $\textit{unsupervised}$ automatic speech recog

GeoAIAfrica/vllm-satim-labeling

Domaine:

Visit

Tasks