Sub 10 billion parameter models that do well at mathematical reasoning in my tests. (Excl. DeepSeek Math 7B RL, reason: couldn't get it to work)
Shreyan Chaubey
thethinkmachine
AI & ML interests
LLM hobbyist. LLMs are love 💕.
I test math, code, factual knowledge & reasoning, with my own set of questions and methods. My observations are usually consistent with popular domain specific benches like MMLU-Pro. I am not an expert.
Organizations
None yet
Collections
4
Good chat finetunes (RHLF, DPO, ORPO) foundational models & finetunes.
-
MaziyarPanahi/Calme-7B-Instruct-v0.9
Text Generation • Updated • 920 • 9 -
MaziyarPanahi/Llama-3-8B-Instruct-v0.10
Text Generation • Updated • 80 • 2 -
NousResearch/Hermes-2-Theta-Llama-3-8B
Text Generation • Updated • 22.1k • 152 -
NousResearch/Hermes-2-Pro-Llama-3-8B
Text Generation • Updated • 22.5k • 378
models
None public yet
datasets
None public yet