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

reject merging subset

Browse source
This commit is contained in:
ngxson 2024-07-23 20:13:01 +02:00
parent 1f965737f0
commit 0ec2b5818a
1 changed files with 15 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

15
examples/export-lora/export-lora.cpp
View file

@ -195,6 +195,19 @@ struct lora_merge_ctx {
// output is forced to f16 for now
gguf_set_val_u32(ctx_out, "general.file_type", LLAMA_FTYPE_MOSTLY_F16);
// check if all lora adapters have the same tensors
// TODO: remove this when we can support merging subset of adapters. Ref: https://github.com/ggerganov/llama.cpp/pull/8607#discussion_r1686027777
if (adapters.size() > 1) {
auto & base_adapter = adapters[0];
for (size_t i = 1; i < adapters.size(); ++i) {
for (auto & it : base_adapter->tensors) {
if (base_adapter->get_tensor(it.first) == nullptr) {
throw std::runtime_error("Input adapters do not have the same list of tensors. This is not yet supported. Please merge the adapter one-by-one instead of merging all at once.");
}
}
}
}
// if true, this tensor can be lora-merged. if false, we skip merging and just copy data to outfile
std::vector<std::pair<struct ggml_tensor *, bool>> base_tensors;
for (auto & it : base_model.tensors) {
@ -319,7 +332,7 @@ struct lora_merge_ctx {
const float scale = alpha ? adapters[i]->scale * alpha / rank : adapters[i]->scale;
delta = ggml_scale(ctx0, delta, scale);
cur = ggml_add(ctx0, cur, delta);
printf("%s : merging from adapter[%ld]\n", __func__, i);
printf("%s : + merging from adapter[%ld]\n", __func__, i);
printf("%s : input_scale=%f calculated_scale=%f rank=%d\n", __func__, adapters[i]->scale, scale, (int) inp_b[i]->ne[0]);
}
cur = ggml_cast(ctx0, cur, get_out_tensor_type(base));