Ukko Thinking is an instruction-following AI assistant using chain-of-thought and reasoning.

This model is full fine-tune of Qwen/Qwen2.5-3B-Instruct, a powerful 3.09B parameter model.

v1: This is the first initial release.

Settings: Use with the Alpaca prompt template, temperature of 0.6 and top_p of 0.9.

Safetensors

Model size

3B params

Tensor type

F32

BF16

Model tree for vpakarinen/ukko-thinking-3.09b-v1

Base model

Qwen/Qwen2.5-3B

Finetuned

Finetuned

(808)

this model

Quantizations

Collection including vpakarinen/ukko-thinking-3.09b-v1