It's kinda barely under 16GB in full precision,
you can give it a try using
Google Colab Notebook
I threw this together but be warned it will OOM on long context due to Free Tier being limited to 15GB VRAM.
Introduction:
Introducing MermaidMistral, a powerful yet compact 7-billion-parameter language model adept at Python code understanding and crafting engaging story flow maps. Trained on a meticulously hand curated dataset of 478 diverse Python examples and hand crafted mermaid flow maps utilizing https://mermaid.live, this model goes beyond its size to deliver exceptional performance in code understanding and story visualization.
Key Features:
MermaidMistral is not a "Chatty Kathy" and should only respond with a mermaid code block with a flow diagram in mermaid js syntax and nothing more.
1. Code Understanding:
Grasps Python intricacies with finesse.
Generates clear and accurate Mermaid Diagram Flow Charts.
Ideal for developers seeking visual representations of their code's logic.
2. Storytelling Capabilities:
Converts narrative inputs into captivating Mermaid Diagrams.
Maps character interactions, plot developments, and narrative arcs effortlessly.
3. Unmatched Performance:
Surpasses larger models, like GPT-4, in generating well-organized and detailed Mermaid Diagrams for story flows.
4. Training Insights:
Trained on a 478 Python examples for just under three epochs on a single RTX 3090 following batch size equal to 1, known as stochastic gradient descent.
Exhibited emergent properties in story-to-flow map translations.
Adaptable and efficient in resource utilization
Due to hardware constraints this fine tune has a token limit of 2048.
Mermaid Mistral Generation 1:
Mermaid Mistral Generation 1
Mermaid Mistral Generation 2:
Mermaid Mistral Generation 2
ChatGPT Generation:
ChatGPT Generation
Collaboration:
MermaidMistral is open to collaboration to further strengthen its capabilities. The dataset, formatted in Alpaca, provides a unique foundation for understanding Python intricacies. If you're interested in contributing or collaborating to enhance the model's performance, feel free to reach out to troydoesai@gmail.com. Your expertise could play a pivotal role in refining MermaidMistral.
Example Use Cases:
1. Code Documentation:
Developers can use MermaidMistral to automatically generate visual flow charts from their Python code, aiding in documentation and code understanding.
2. Storyboarding:
Storytellers and writers can input their narrative and receive visually appealing Mermaid Diagrams, offering a structured overview of character interactions and plot progression.
3. Project Planning:
Project managers can leverage MermaidMistral to create visual project flow maps, facilitating effective communication and planning among team members.
4. Learning Python:
Students and beginners can use MermaidMistral to visually understand Python code structures, enhancing their learning experience.
5. Game Design:
Game developers can utilize MermaidMistral for visualizing game storylines, ensuring a coherent narrative structure and character development.
Proof of Concept:
MermaidMistral proves that innovation thrives in compact packages, delivering exceptional performance across diverse applications. Its adaptability and efficiency showcase the potential for groundbreaking results even in resource-constrained environments.
I plan to release my working proof of concept VSCode Extension that currently displays the Live Flow Map every time a user stops typing for more than 10 seconds.
