Public/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2026-09-20 01:31:31 +02:00
Code Issues Packages Projects Releases Wiki Activity

tests : fix dsv4 save-load n_stream mismatch

Browse Source 
The dsv4 KV cache keeps per-sequence KV/state streams even in unified mode, so its n_stream equals n_seq_max. The test saved the state in the baseline with n_seq_max=1 but loaded it in the seq-copy tests with n_seq_max=2, so state_read threw an n_stream mismatch. Use n_seq_max=2 in the baseline and state-load tests so the save and load agree.

Assisted-by: pi:llama.cpp/Qwen3.8-27B
This commit is contained in:
Georgi Gerganov
2026-08-26 21:15:40 +03:00
parent a538337fbb
commit 78898478fb
1 changed files with 6 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tests/test-save-load-state.cpp
+6 -2
View File
@@ -57,7 +57,9 @@ static llama_tokens generate_tokens(llama_context * ctx, llama_sampler * smpl, i
// - decode the last token
// - generate n_predict tokens
static llama_tokens test_baseline(struct llama_model * model, const struct common_params & params, const llama_tokens & tokens) {
auto ctx = llama_context_ptr{llama_init_from_model(model, common_context_params_to_llama(params))};
auto params_ctx = common_context_params_to_llama(params);
params_ctx.n_seq_max = 2;
auto ctx = llama_context_ptr{llama_init_from_model(model, params_ctx)};
auto sparams = llama_sampler_chain_default_params();
auto smpl = llama_sampler_ptr{llama_sampler_chain_init(sparams)};
@@ -165,7 +167,9 @@ static bool test_seq_rm_isolated(
// - replay the last prompt token
// - generate n_predict tokens and compare against expected result
static bool test_state_load(struct llama_model * model, const struct common_params & params, const llama_tokens & tokens, const llama_tokens & expected_result) {
auto ctx = llama_context_ptr{llama_init_from_model(model, common_context_params_to_llama(params))};
auto params_ctx = common_context_params_to_llama(params);
params_ctx.n_seq_max = 2;
auto ctx = llama_context_ptr{llama_init_from_model(model, params_ctx)};
auto sparams = llama_sampler_chain_default_params();
auto smpl = llama_sampler_ptr{llama_sampler_chain_init(sparams)};