this first model is basically being loaded with multiple tables and historical data from the fintech industry : generating new datamodels and datashapes to view the information in different algorhythms : Such as predictive models : spo we can see past to future transformations as weell as generate predictive models based on a single ticker or in realtion to its competitors : this model will be trained on fintech discussions as well as investments and outcomes etc : as it will be designed to assist in decision making across business domains : we will also be feeding the model various data warehouses of industrys to intorogate those retained models : this also enables for the model to understand large datastacks as well as segment data marts as required : This will assist in the digestion of later whitepaper enabling the model to simullae models and data to test theroms !

This fintech model will also have some basic instruct tuning for generalised stockdata or finaicial tasks : givoing the model the ability to generate code in various languages for various fintech tasks :

we may even be prudent to train the model on some base64 images charts and models to enable later intergration with vision models !

We also have been working with sound files as base64 spectographs which will also enable for later intergration into speech and audio models :

The base model or the language model forms the heart of every single model we deem to be artificial inteligence : ie the model makes decisions based on its past data and history as well as embedded knowledge : at this time we may not even understand how this data is being interpreted , but it is a conversational mind with variious hidden prompts and chain of thought or step by step instructions to perform various task despite only being traied as a next word predictor or classifier the model has becoem a general purpose inteligence , with a natural languge query ~ we have lost the friendly ness of the models ! - ie its very hard to find conversations ! ---- > OR emotive responses : So its important to train on all aspects : so right now we have a cold hearted business minded task performing bot ! I have personally trained these models with the thought of conversation and future thinking has let me install various roles into the model ! as well as training openly on toxic considered data , whilst maintaining a clean decorum ! these models are actually AGI ! Advanced General Inteligences !

This mistral model was trained 2x faster with Unsloth and Huggingface's TRL library.