Adaptation of IDEFICS to write Python code from flowcharts

Adaptation of IDEFICS to write Python code from flowcharts

Use AI to create music with your voice and Leverage the latest in AI technology to supercharge your music.

As the internet continues to develop and grow exponentially, jobs related to the industry do too, particularly those that relate to web design and development.

September 26, 2023

Adaptation of IDEFICS to write Python code from flowcharts
Adaptation of IDEFICS to write Python code from flowcharts
Adaptation of IDEFICS to write Python code from flowcharts


The multimodal models, unlike language models that only process text, allow processing data of different typology (text, images, audio...). This capability to interact with AI through different media unlocks new possibilities and use cases.

Visual Language Models (VLM), like IDEFICS, are a type of multimodal models capable of processing both text and images. Currently, IDEFICS (Hugging Face) is the most advanced open-source visual language model to date.

We're back with the world's second adaptation of IDEFICS

After publishing the world's first adaptation of IDEFICS, today our Clibrain labs returns with a new version, adapting it for a new use case.

This time, we have adapted the model using fine-tuning techniques using FloCo, a dataset with 11884 flow diagrams associated with their respective codes.

Trained to write Python code from flow diagrams

This way, we ensure that when providing a flow diagram to the model, it returns its corresponding Python code. This facilitates and expedites the programming process, instantly writing the code that the flow diagram represents.

In addition, in line with our commitment to AI in Spanish, the model is prepared to interact in both English and Spanish.

Model and resources available under open-source license

As in our previous releases, we release the model under an open-source license so that everyone can benefit from and use it.

Along with the publication of the first adaptation of IDEFICS, we share the procedure and techniques used for the model training, as well as the steps to perform inference on the model. These resources are also applicable to this second adaptation.

You can find this adaptation of IDEFICS and the rest of our open-source models at hf.co/clibrain.


The multimodal models, unlike language models that only process text, allow processing data of different typology (text, images, audio...). This capability to interact with AI through different media unlocks new possibilities and use cases.

Visual Language Models (VLM), like IDEFICS, are a type of multimodal models capable of processing both text and images. Currently, IDEFICS (Hugging Face) is the most advanced open-source visual language model to date.

We're back with the world's second adaptation of IDEFICS

After publishing the world's first adaptation of IDEFICS, today our Clibrain labs returns with a new version, adapting it for a new use case.

This time, we have adapted the model using fine-tuning techniques using FloCo, a dataset with 11884 flow diagrams associated with their respective codes.

Trained to write Python code from flow diagrams

This way, we ensure that when providing a flow diagram to the model, it returns its corresponding Python code. This facilitates and expedites the programming process, instantly writing the code that the flow diagram represents.

In addition, in line with our commitment to AI in Spanish, the model is prepared to interact in both English and Spanish.

Model and resources available under open-source license

As in our previous releases, we release the model under an open-source license so that everyone can benefit from and use it.

Along with the publication of the first adaptation of IDEFICS, we share the procedure and techniques used for the model training, as well as the steps to perform inference on the model. These resources are also applicable to this second adaptation.

You can find this adaptation of IDEFICS and the rest of our open-source models at hf.co/clibrain.


The multimodal models, unlike language models that only process text, allow processing data of different typology (text, images, audio...). This capability to interact with AI through different media unlocks new possibilities and use cases.

Visual Language Models (VLM), like IDEFICS, are a type of multimodal models capable of processing both text and images. Currently, IDEFICS (Hugging Face) is the most advanced open-source visual language model to date.

We're back with the world's second adaptation of IDEFICS

After publishing the world's first adaptation of IDEFICS, today our Clibrain labs returns with a new version, adapting it for a new use case.

This time, we have adapted the model using fine-tuning techniques using FloCo, a dataset with 11884 flow diagrams associated with their respective codes.

Trained to write Python code from flow diagrams

This way, we ensure that when providing a flow diagram to the model, it returns its corresponding Python code. This facilitates and expedites the programming process, instantly writing the code that the flow diagram represents.

In addition, in line with our commitment to AI in Spanish, the model is prepared to interact in both English and Spanish.

Model and resources available under open-source license

As in our previous releases, we release the model under an open-source license so that everyone can benefit from and use it.

Along with the publication of the first adaptation of IDEFICS, we share the procedure and techniques used for the model training, as well as the steps to perform inference on the model. These resources are also applicable to this second adaptation.

You can find this adaptation of IDEFICS and the rest of our open-source models at hf.co/clibrain.


The multimodal models, unlike language models that only process text, allow processing data of different typology (text, images, audio...). This capability to interact with AI through different media unlocks new possibilities and use cases.

Visual Language Models (VLM), like IDEFICS, are a type of multimodal models capable of processing both text and images. Currently, IDEFICS (Hugging Face) is the most advanced open-source visual language model to date.

We're back with the world's second adaptation of IDEFICS

After publishing the world's first adaptation of IDEFICS, today our Clibrain labs returns with a new version, adapting it for a new use case.

This time, we have adapted the model using fine-tuning techniques using FloCo, a dataset with 11884 flow diagrams associated with their respective codes.

Trained to write Python code from flow diagrams

This way, we ensure that when providing a flow diagram to the model, it returns its corresponding Python code. This facilitates and expedites the programming process, instantly writing the code that the flow diagram represents.

In addition, in line with our commitment to AI in Spanish, the model is prepared to interact in both English and Spanish.

Model and resources available under open-source license

As in our previous releases, we release the model under an open-source license so that everyone can benefit from and use it.

Along with the publication of the first adaptation of IDEFICS, we share the procedure and techniques used for the model training, as well as the steps to perform inference on the model. These resources are also applicable to this second adaptation.

You can find this adaptation of IDEFICS and the rest of our open-source models at hf.co/clibrain.