Correctly accumulate sampling time for adaptive_p

This commit is contained in:
Kawrakow
2026-01-19 10:17:07 +00:00
parent 61eccfcf0d
commit a96e5449cc
5 changed files with 20 additions and 10 deletions

View File

@@ -471,7 +471,7 @@ static llama_token llama_sampling_sample_impl(
id = llama_sample_token_mirostat_v2(ctx_main, &cur_p, mirostat_tau, mirostat_eta, &ctx_sampling->mirostat_mu);
} else if (adaptive_target >= 0.0f && ctx_sampling->adapt_p_ctx!=nullptr) {
// adaptive p sampling
llama_prep_adaptive_p(&cur_p, ctx_sampling->adapt_p_ctx);
llama_prep_adaptive_p(ctx_main, &cur_p, ctx_sampling->adapt_p_ctx);
sampler_queue(ctx_main, params, ctx_sampling, cur_p, std::max(1, params.min_keep));
id = llama_sample_token_adaptive_p(ctx_main, &cur_p, ctx_sampling->adapt_p_ctx);
} else {