212 1 year ago

GLM-Z1-0414-32b thinking model with YaRN RoPE scaling to 128k context

tools
6b58c23b5778 ยท 115B
{
"num_ctx": 64000,
"stop": [
"<|system|>",
"<|user|>",
"<|assistant|>"
],
"temperature": 0.3
}