for (auto s : samplers_sequence) {
switch (s){
case 'k': llama_sample_top_k (nullptr, &candidates_p, top_k, 1); break;
- case 'f': GGML_ABORT("tail_free test not implemented"); break;
- case 'y': GGML_ABORT("typical test not implemented"); break;
+ case 'f': GGML_ABORT("tail_free test not implemented");
+ case 'y': GGML_ABORT("typical test not implemented");
case 'p': llama_sample_top_p (nullptr, &candidates_p, top_p, 1); break;
case 'm': llama_sample_min_p (nullptr, &candidates_p, min_p, 1); break;
- case 't': GGML_ABORT("temperature test not implemented"); break;
- default : GGML_ABORT("Unknown sampler"); break;
+ case 't': GGML_ABORT("temperature test not implemented");
+ default : GGML_ABORT("Unknown sampler");
}
llama_sample_softmax(nullptr, &candidates_p); // make sure tokens are sorted for tests