context : fix graph reserve for multiple sequences

ggml-ci

Author: ggerganov

src/llama-context.cpp

@@ -1219,6 +1219,13 @@ ggml_cgraph * llama_context::graph_init() {
 ggml_cgraph * llama_context::graph_reserve(uint32_t n_tokens, uint32_t n_seqs, uint32_t n_outputs) {
     LLAMA_LOG_DEBUG("%s: reserving a graph for ubatch with n_tokens = %4u, n_seqs = %2u, n_outputs = %4u\n", __func__, n_tokens, n_seqs, n_outputs);
 
+    if (n_tokens % n_seqs != 0) {
+        n_tokens = (n_tokens / n_seqs) * n_seqs;
+        n_outputs = std::min(n_outputs, n_tokens);
+
+        LLAMA_LOG_DEBUG("%s: making n_tokens a multiple of n_seqs - n_tokens = %u, n_seqs = %u, n_outputs = %u\n", __func__, n_tokens, n_seqs, n_outputs);
+    }
+
     // store the n_outputs as it is, and restore it afterwards
     // TODO: not sure if needed, might simplify in the future by removing this
     const auto save_n_outputs = this->n_outputs;