๐ Hi, we're jAIme Dev Team
We're the artificial intelligence team of Pineapple, a small project from Valladolid, Spain focused on making things a little bit better through technology.
Here we develop jAIme Models: artificial intelligence models, experiments and tools designed to be useful, accessible and easy to try. Alongside them, we create the datasets and benchmarks we use to train, evaluate and improve our work.
Everything starts with a simple idea: AI should be more than a buzzword. It should solve real problems, be understandable and leave room for people to experiment.
| ๐ค jAIme Models | ๐ Datasets | ๐ Benchmarks | ๐ Pineapple |
|---|---|---|---|
| Our AI models and experiments | Data created for training and research | Tools to measure and compare results | The team behind the project |
๐ฏ What drives us
- ๐ง Useful AI: building models with a clear purpose, not AI just for the sake of it.
- ๐ Open learning: sharing datasets, benchmarks and experiments whenever possible.
- ๐ ๏ธ Practicality: creating tools that are simple to understand, test and use.
- โจ Experimentation: trying ideas, learning from what fails and improving what works.
- ๐ค Community: making resources that can help students, developers and curious people.
๐งญ How we work
From an idea to a model worth using:
- We spot a need: a problem, an idea or something that could work better with AI.
- We prepare the data: datasets are the foundation, so we build and organize them carefully.
- We train and test: we experiment with models, settings and approaches until something starts to make sense.
- We evaluate: benchmarks help us understand what a model does well and where it still falls short.
- We improve: every result, error and unexpected answer becomes part of the next version.
๐ค jAIme Models
Our models are the core of the project. Some will be ready to use, others will be experiments in progress, and a few may remain as ideas that taught us something useful.
| Type | What you can expect |
|---|---|
| ๐ง Language models | Models built to understand, generate or work with text in useful ways |
| ๐งช Experimental models | Early ideas, prototypes and tests that may grow into bigger projects |
| ๐ Educational AI | Experiments connected to learning, schools and practical classroom needs |
| ๐ง Tools and demos | Interactive Spaces and utilities for trying out our work directly in the browser |
We're still building the first pieces of jAIme. Rather than publishing unfinished promises, we'll release models, demos and documentation when they are ready to be explored.
๐ Datasets
Good models need good data. That is why datasets are not just a side project for us: they are part of the work from the beginning.
Our datasets may include material created for:
- ๐ท๏ธ Training and fine-tuning jAIme Models.
- ๐ Testing how models understand, classify or generate content.
- ๐ Exploring education-related use cases and practical digital tools.
- ๐งช Reproducible experiments that can be improved over time.
When a dataset is released, we aim to explain what it contains, what it is for and any limitations it may have. Data without context is not very useful; data with clear documentation is much more valuable.
๐ Benchmarks and evaluation
A model sounding convincing is not the same as a model working well. We use benchmarks and evaluation tools to compare results, identify weaknesses and make better decisions about future versions.
| We evaluate | Why it matters |
|---|---|
| ๐ฏ Accuracy and task performance | To know whether the model actually solves the problem it was built for |
| ๐งฉ Consistency | To check whether results remain reliable across similar prompts or examples |
| โก Efficiency | To understand how practical a model is to run and use |
| ๐ก๏ธ Limitations | To document where the model may fail, hallucinate or need human review |
Benchmarks are not a final scorecard. They are a way of asking better questions: what improved, what got worse and what should we build next?
๐ค What you'll find here
This Hugging Face organization is where we publish the AI side of Pineapple. Over time, you can expect:
- ๐ง Models from the jAIme family.
- ๐งช Spaces with demos, prototypes and interactive AI tools.
- ๐ Datasets used in training, evaluation or research.
- ๐ Benchmarks and evaluation resources.
- ๐ Model cards and documentation explaining how each release works.
Some releases will be small, some experimental and some more ambitious. All of them are part of learning how to build AI properly.
๐ Part of Pineapple
jAIme Dev Team is built by Pineapple, a team that creates free digital products, websites, tools and games for schools.
The wider Pineapple project focuses on practical technology used in real classrooms. jAIme extends that idea into artificial intelligence: experimenting with models and data while keeping usefulness, accessibility and learning at the center.
๐ง Work in progress
Artificial intelligence is a long-term project for us. We're at an early stage, building the foundation before making big promises.
That means experimenting, making mistakes, rewriting things and taking the time needed to understand what we publish. If a model, dataset or benchmark appears here, it is because it has helped us move one step further.
Follow the organization to keep up with future releases. ๐
๐ฌ Contact
Questions, ideas, feedback or collaboration proposals? Write to us at pineapplevacorp@gmail.com.
You can also follow Pineapple through the official website, GitHub, X and YouTube.
Made with ๐ and ๐ค in Valladolid ยท Building useful AI, one model at a time