Closed jzeiber closed 11 months ago
memsize in gpt_model and bark_model is currently int32_t, and it overflows when adding the individual model sizes together. Changing them to int64_t should resolve this.
Good catch @jzeiber ! Thanks!
memsize in gpt_model and bark_model is currently int32_t, and it overflows when adding the individual model sizes together. Changing them to int64_t should resolve this.