Downloads · 30 days
9
9% of all-time downloads
LeroyDyer/_Starfleet_II_
_Starfleet_II_ is a text generation model from LeroyDyer. Use it when you need the model to write or continue text. It is set up for transformers. The card lists the license as apache-2.0.
Here we are begining to focus on chracter ! : so the model may refuse to perform being not in a good mood ! not because its blocked ! so if your polite then it may respond better and if your angery it may resond angry ?
Downloads · 30 days
9
9% of all-time downloads
All-time downloads
102
Public
Parameters
7.2B
14.5 GB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors14.5 GB · 100%
From the Hugging Face model README
Here we are begining to focus on chracter ! : so the model may refuse to perform being not in a good mood ! not because its blocked ! so if your polite then it may respond better and if your angery it may resond angry ?
i have not created a Evil thoughts model in which teh model will use angry reasoning for its motivation YET ! or A good thoughts reasoner , in which its reasoning i geared to being GOOD ! ( maybe some refusals to perform illicit acts ) as its in GOOD MODE !
so we wil mix these trainings so we do not create unbalenced models :
i think this themed responses and theme thoughts are a way to allow the model to devlop some emotional responses , so we weave them into TASKS , as well as in code generation we would like a model to assume its role with a pesonality !
so we humaize these models with more chat based colaberatie tasks, instead of fully completing a task ! the aim of task performance her is not the motive , as we already know we can fore models to perfrom tasks with prompts : These are not prompt based responses any more !
these are reasoning based responses :
we also use reasoning and imeadiate responses in our datasts so the model has a selection of responses types inside and not forced to produce reasoning , with a swtich !! the aim again is to allow the model to choose the tasks which need reasoning and tasks which do not !
but in truth we can always train evil thoughts or romantic thoughts so every response has romantic motivations , and then cover this with a prompt response for "you are a helpful AI " as well as merge the model to conceal these repsosnes and they will popup out the blue !
so this model has also had some outputs tylored to display as Starfleet computer : as it is a themed model : we will be retraining these series to act more like the base computer on a ship ! so responses and content will be displyaed and framed appropirately !
we have laid the groundwork from the begining for these methodologie if you have been following along !