Models / qwen2.5-0.5b-instruct

Qwen2.5 0.5B Instruct

A 0.5B-parameter Qwen2 decoder with grouped-query attention, biased query/key/value projections, and an output head tied to the token embedding, instruction-tuned.

494.5M parametersqwen2Apache-2.0text-generationdecoder-onlygrouped-query-attentionchat

Greedy decoding of one chat prompt, through Linnet and through transformers (KV cache, argmax, bf16).

Prompt

Explain in two sentences why the sky is blue.

Linnet

The sky is blue because of the scattering of sunlight by tiny particles in the Earth's atmosphere, particularly water vapor and nitrogen oxides. This scattering causes the blue light from the sun to be scattered in all directions, creating the characteristic blue color we see in the sky.

transformers (KV cache, argmax, bf16)

The sky is blue because it is a reflection of the sunlight that falls on Earth. The Earth's atmosphere scatters and disperses the sunlight, causing it to be scattered and refracted by the various particles in the atmosphere, resulting in the characteristic blue color of the sky.

The first 5 of 55 generated tokens agree with the reference; after that the two stacks round a near-tie differently and part ways, which bf16 greedy decoding allows.