vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

merge: debug

Browse source
This commit is contained in:
ngxson 2024-03-01 22:07:50 +01:00
parent a1bf1e1b2e
commit 853b3d716d
1 changed files with 3 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

4
llama.cpp
View file

@ -11492,6 +11492,7 @@ int32_t llama_merge_models(const struct llama_merge_config * config) {
}; };
// process function, to be run as thread // process function, to be run as thread
// TODO: multi-threading here is done for each tensor (instead of each row like in llama_model_quantize_internal), this is not ideal but still better than single-thread
const size_t n_start = n_curr; const size_t n_start = n_curr;
auto process_output_tensor = [&]() { auto process_output_tensor = [&]() {
worker_acquire(); worker_acquire();
@ -11570,7 +11571,8 @@ int32_t llama_merge_models(const struct llama_merge_config * config) {
return n_done == my_number; return n_done == my_number;
}); });
} }
LLAMA_LOG_ERROR("===> %f %f %f\n", f32_out_buf[0], f32_out_buf[1], f32_out_buf[2]); LLAMA_LOG_ERROR("===> INPUT [layer %d] %f %f %f\n", i_layer_out, f32_in_buf[0].value, f32_in_buf[1].value, f32_in_buf[2].value);
LLAMA_LOG_ERROR("===> OUTPUT [layer %d] %f %f %f\n", i_layer_out, f32_out_buf[0], f32_out_buf[1], f32_out_buf[2]);
// my turn, write the result! // my turn, write the result!
// write tensor data + padding // write tensor data + padding
fout.write((const char *) out_buf.data(), out_buf.size()); fout.write((const char *) out_buf.data(), out_buf.size());