kaelri/ hy-mt2:7b-q4_K_M

308 yesterday

Official Tencent HY-MT2 (1.8B / 7B GGUF) models with an Ollama-compatible prompt template for correct and consistent local translation behavior.

1.8b 7b
... /
params
de703937e744 · 230B
{
"num_predict": 4096,
"repeat_penalty": 1.05,
"stop": [
"<｜",
"<ｆin｜hy-",
"<ｈy-",
"<ｺ｜hy-",
"<ｂ｜hy-",
"<suggested_response",
"</suggested_response"
],
"temperature": 0.7,
"top_k": 20,
"top_p": 0.6
}