Panic on large inputs?

[corani]

Is the example supposed to panic if the input is too large? With ~2400 characters I get:

ggml_new_tensor_impl: not enough space in the context's memory pool (needed 271388624, available 260703040)
SIGSEGV: segmentation violation
PC=0x4d2645 m=0 sigcode=1
signal arrived during cgo execution

[corani]

As a workaround I've added an N *= 2; in bert_embeddings, with that things work fine: if the input it too large, I get an Too many tokens, maximum is 512 printed to the output.

Obviously this isn't the right solution, but it seems somewhere a calculation is off.