This is quantized version of
nisten/Biggie-SmoLlm-0.15B-Base created using llama.cpp
###EVEN SMALLER Frankenstein of smolLm-0.13b upped to 0.15b
Use this frankenbase for training.
Done via semi-automated continuous merging to figure out the recipe.
Model is more coherent.
1llama-cli -ngl 99 -co --temp 0 -p "How to build a city on Mars via calculating Aldrin-Cycler orbits?" -m Biggie_SmolLM_0.15B
2_Base_bf16.gguf
The temperature settings and min p etc need to be adjusted but even at default temp0 it was coherent for first 100 tokens.
Amazing option for further training. And this is a merge of the base, not the instruct!
