MEADBEE LABSPersonal AI / Software Laboratory
Menu

Model registry / Training records

Models

Model work presented as an engineering process: architecture and tokenizer choices, data preparation, training progression, evaluation, failures, and outputs.

01 / Model index

Built from the training pipeline up

Model cards expose useful technical metadata while leaving uncertain or unpublished fields visibly unset.

Index classes
  • Diffusion Model01
  • Language Model01
model-diffusion-v1research
ModelDiffusion Model

Self-Trained Diffusion Study

A diffusion-model research track focused on dataset work, training configuration, checkpoint progression, generated outputs, and evaluation.

Parameters
Not published
Tokenizer
Project specific
Context
Not published
Training
Local research pipeline
  • Diffusion
  • Image Generation
  • Training
Model note
model-yunagptresearch
ModelLanguage Model

YunaGPT

A self-trained decoder language-model project covering architecture experiments, tokenizer work, data preparation, pretraining, instruction tuning, and alignment studies.

Parameters
~300M project profile
Tokenizer
Custom pipeline
Context
Not published
Training
From scratch
  • LLM
  • From Scratch
  • Training
Model note

02 / Language-model record

The whole training story belongs here

YunaGPT is framed as a designed and trained pipeline—not merely a model artifact downloaded from elsewhere.

01Architecture
02Tokenizer
03Data
04Pretraining
05Instruction tuning
06Preference studies

03 / Diffusion record

Training should be visible

Diffusion pages will prioritize generated output while keeping the dataset, training configuration, checkpoint progression, and evaluation beside the images.