cosetoenor/decider-2b-GGUF

Clarify v10 weights, historical v8 tests, and upstream change references

Update GGUF weights to upstream v10 and preserve v8 benchmark provenance

Add Chinese report links at the top of the model card

Publish bilingual GPU benchmarks and three-model comparison data

Add state-first GPU benchmark for Q8_0 and IQ4_NL

GGUF Q8_0 / IQ4_NL / F16 of Mapika/decider-2b, plus llama.cpp scoring shim

initial commit