nikola/llama_cpp
1
0
Fork 0
Code Issues Pull Requests Actions 17 Packages Projects Releases Wiki Activity

common : support manually triggering the reasoning budget end sequence (#23949)

Browse Source
This commit is contained in:
Aldehir Rojas
2026-06-01 11:37:11 +02:00
committed by GitHub
parent e22b0de60d
commit 5254a7994d
5 changed files with 108 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tests/test-reasoning-budget.cpp
+72 -1
View File
@@ -184,6 +184,76 @@ static void test_reasoning_budget_clone_mid_forcing() {
llama_sampler_free(sampler);
}
static void test_reasoning_budget_force_manual() {
const std::vector<llama_token> start = {100};
const std::vector<llama_token> end = {101};
const std::vector<llama_token> forced = {102, 101};
// if COUNTING, force() succeeds and begins forcing the end sequence from the start
{
auto * sampler = common_reasoning_budget_init(nullptr, start, end, forced, 5, REASONING_BUDGET_IDLE);
llama_sampler_accept(sampler, 100); // COUNTING, remaining=5
llama_sampler_accept(sampler, 50); // COUNTING, remaining=4
GGML_ASSERT(common_reasoning_budget_get_state(sampler) == REASONING_BUDGET_COUNTING);
GGML_ASSERT(common_reasoning_budget_force(sampler) && "force() should succeed from COUNTING");
GGML_ASSERT(common_reasoning_budget_get_state(sampler) == REASONING_BUDGET_FORCING);
// forces the configured sequence from force_pos=0, then transitions to DONE
GGML_ASSERT(get_forced_token(sampler, 102) == 102);
llama_sampler_accept(sampler, 102);
GGML_ASSERT(get_forced_token(sampler, 102) == 101);
llama_sampler_accept(sampler, 101);
GGML_ASSERT(common_reasoning_budget_get_state(sampler) == REASONING_BUDGET_DONE);
llama_sampler_free(sampler);
}
// if IDLE, force() is a no-op
{
auto * sampler = common_reasoning_budget_init(nullptr, start, end, forced, 5, REASONING_BUDGET_IDLE);
GGML_ASSERT(!common_reasoning_budget_force(sampler) && "force() must not transition from IDLE");
GGML_ASSERT(common_reasoning_budget_get_state(sampler) == REASONING_BUDGET_IDLE);
llama_sampler_free(sampler);
}
// if DONE, force() is a no-op
{
auto * sampler = common_reasoning_budget_init(nullptr, start, end, forced, 5, REASONING_BUDGET_IDLE);
llama_sampler_accept(sampler, 100); // COUNTING
llama_sampler_accept(sampler, 101); // natural end -> DONE
GGML_ASSERT(common_reasoning_budget_get_state(sampler) == REASONING_BUDGET_DONE);
GGML_ASSERT(!common_reasoning_budget_force(sampler) && "force() must not transition from DONE");
GGML_ASSERT(common_reasoning_budget_get_state(sampler) == REASONING_BUDGET_DONE);
llama_sampler_free(sampler);
}
// if FORCING, force() is a no-op and must not rewind the force position
{
auto * sampler = common_reasoning_budget_init(nullptr, start, end, forced, 0, REASONING_BUDGET_FORCING);
GGML_ASSERT(get_forced_token(sampler, 102) == 102);
llama_sampler_accept(sampler, 102); // advance to the second forced token (force_pos=1)
GGML_ASSERT(!common_reasoning_budget_force(sampler) && "force() must not transition from FORCING");
GGML_ASSERT(common_reasoning_budget_get_state(sampler) == REASONING_BUDGET_FORCING);
GGML_ASSERT(get_forced_token(sampler, 102) == 101 && "force() must not rewind the force position");
llama_sampler_free(sampler);
}
// a null sampler is safely ignored
GGML_ASSERT(!common_reasoning_budget_force(nullptr));
fprintf(stderr, " Test 'manual force transition' passed\n");
}
// UTF-8 boundary detection unit test
// Tests common_utf8_is_complete() from reasoning-budget.h
static void test_utf8_boundary_detection() {
@@ -312,8 +382,9 @@ int main(void) {
test_reasoning_budget_clone_mid_counting();
test_reasoning_budget_clone_mid_forcing();
test_reasoning_budget_force_manual();
printf("OK (8 tests passed)\n");
printf("OK (9 tests passed)\n");
printf("Testing UTF-8 boundary detection... ");
test_utf8_boundary_detection();