app can only answer one question
#1
by
etqoo1026 - opened
I am testing the model SmolLM-135M-Instruct_multi-prefill-seq_f32_ekv1280.task, but it can only answer one question. When I input a second question, the app doesn't respond.
Any tips or recommendations would be really helpful.
max token: 100
temperature: 0.7
Device: Zenfone 10
OS: Android 15
seems the template is not configured, the generation should stop at <|im_end|>