mirror of
https://github.com/ggerganov/llama.cpp.git
synced 2024-12-26 11:24:35 +00:00
Control vector loading fixes (#8137)
* Fixed leak in llama_control_vector_load_one() and allow llama_control_vector_load() to grow * refactored `llama_control_vector_load_one()` * allow multiple directions for same layer in same file * llama_control_vector_load_one() and llama_control_vector_load() now break on error * removed unnecessary ggml_free() call
This commit is contained in:
jukofyork
GitHub
parent
387952651a
commit
97877eb10b
@ -2804,125 +2804,87 @@ float llama_embd_similarity_cos(const float * embd1, const float * embd2, int n)
|
|||||||
//
|
//
|
||||||
|
|
||||||
static llama_control_vector_data llama_control_vector_load_one(const llama_control_vector_load_info & load_info) {
|
static llama_control_vector_data llama_control_vector_load_one(const llama_control_vector_load_info & load_info) {
|
||||||
int32_t n_tensors;
|
|
||||||
|
|
||||||
size_t n_bytes = 0;
|
|
||||||
|
|
||||||
uint32_t max_direction_layer = 0;
|
|
||||||
|
|
||||||
llama_control_vector_data result = { -1, {} };
|
llama_control_vector_data result = { -1, {} };
|
||||||
|
|
||||||
// calculate size of ctx needed for tensors, ensure tensors are f32, and find max layer
|
ggml_context * ctx = nullptr;
|
||||||
{
|
struct gguf_init_params meta_gguf_params = {
|
||||||
struct ggml_init_params meta_params = {
|
/* .no_alloc = */ false,
|
||||||
/* .mem_size = */ ggml_tensor_overhead() * 128 + ggml_graph_overhead(),
|
/* .ctx = */ &ctx,
|
||||||
/* .mem_buffer = */ nullptr,
|
};
|
||||||
/* .no_alloc = */ true,
|
struct gguf_context * ctx_gguf = gguf_init_from_file(load_info.fname.c_str(), meta_gguf_params);
|
||||||
};
|
if (!ctx_gguf) {
|
||||||
ggml_context * meta_ctx = ggml_init(meta_params);
|
fprintf(stderr, "%s: failed to load control vector file from %s\n", __func__, load_info.fname.c_str());
|
||||||
struct gguf_init_params meta_gguf_params = {
|
return result;
|
||||||
/* .no_alloc = */ true,
|
|
||||||
/* .ctx = */ &meta_ctx,
|
|
||||||
};
|
|
||||||
struct gguf_context * meta_ctx_gguf = gguf_init_from_file(load_info.fname.c_str(), meta_gguf_params);
|
|
||||||
if (!meta_ctx_gguf) {
|
|
||||||
fprintf(stderr, "%s: failed to load control vector from %s\n", __func__, load_info.fname.c_str());
|
|
||||||
ggml_free(meta_ctx);
|
|
||||||
return result;
|
|
||||||
}
|
|
||||||
|
|
||||||
n_tensors = gguf_get_n_tensors(meta_ctx_gguf);
|
|
||||||
for (int i = 0; i < n_tensors; i++) {
|
|
||||||
std::string name = gguf_get_tensor_name(meta_ctx_gguf, i);
|
|
||||||
|
|
||||||
// split on '.'
|
|
||||||
size_t dotpos = name.find('.');
|
|
||||||
if (dotpos != std::string::npos && name.substr(0, dotpos) == "direction") {
|
|
||||||
try {
|
|
||||||
uint32_t layer = std::stoi(name.substr(dotpos + 1));
|
|
||||||
if (layer == 0) {
|
|
||||||
fprintf(stderr, "%s: direction tensor invalid in %s\n", __func__, load_info.fname.c_str());
|
|
||||||
ggml_free(meta_ctx);
|
|
||||||
gguf_free(meta_ctx_gguf);
|
|
||||||
return result;
|
|
||||||
}
|
|
||||||
if (layer > max_direction_layer) {
|
|
||||||
max_direction_layer = layer;
|
|
||||||
}
|
|
||||||
} catch (...) {
|
|
||||||
fprintf(stderr, "%s: direction tensor invalid in %s\n", __func__, load_info.fname.c_str());
|
|
||||||
ggml_free(meta_ctx);
|
|
||||||
gguf_free(meta_ctx_gguf);
|
|
||||||
return result;
|
|
||||||
}
|
|
||||||
}
|
|
||||||
|
|
||||||
struct ggml_tensor * tensor_meta = ggml_get_tensor(meta_ctx, name.c_str());
|
|
||||||
if (tensor_meta->type != GGML_TYPE_F32 || ggml_n_dims(tensor_meta) != 1) {
|
|
||||||
fprintf(stderr, "%s: direction tensor invalid in %s\n", __func__, load_info.fname.c_str());
|
|
||||||
ggml_free(meta_ctx);
|
|
||||||
gguf_free(meta_ctx_gguf);
|
|
||||||
return result;
|
|
||||||
}
|
|
||||||
if (result.n_embd == -1) {
|
|
||||||
result.n_embd = ggml_nelements(tensor_meta);
|
|
||||||
} else if (ggml_nelements(tensor_meta) != result.n_embd) {
|
|
||||||
fprintf(stderr, "%s: direction tensor sizes mismatched in %s\n", __func__, load_info.fname.c_str());
|
|
||||||
ggml_free(meta_ctx);
|
|
||||||
gguf_free(meta_ctx_gguf);
|
|
||||||
return result;
|
|
||||||
}
|
|
||||||
n_bytes += ggml_nbytes(tensor_meta);
|
|
||||||
}
|
|
||||||
ggml_free(meta_ctx);
|
|
||||||
gguf_free(meta_ctx_gguf);
|
|
||||||
}
|
}
|
||||||
|
|
||||||
|
int32_t n_tensors = gguf_get_n_tensors(ctx_gguf);
|
||||||
if (n_tensors == 0) {
|
if (n_tensors == 0) {
|
||||||
fprintf(stderr, "%s: no direction tensors found in %s\n", __func__, load_info.fname.c_str());
|
fprintf(stderr, "%s: no direction tensors found in %s\n", __func__, load_info.fname.c_str());
|
||||||
return result;
|
|
||||||
}
|
}
|
||||||
|
|
||||||
// load and scale tensors into final control vector context
|
for (int i = 0; i < n_tensors; i++) {
|
||||||
struct ggml_init_params ggml_params = {
|
std::string name = gguf_get_tensor_name(ctx_gguf, i);
|
||||||
/* .mem_size = */ ggml_tensor_overhead() * n_tensors + n_bytes,
|
|
||||||
/* .mem_buffer = */ nullptr,
|
|
||||||
/* .no_alloc = */ false,
|
|
||||||
};
|
|
||||||
struct ggml_context * ctx = ggml_init(ggml_params);
|
|
||||||
|
|
||||||
struct gguf_init_params params = {
|
int layer_idx = -1;
|
||||||
/*.no_alloc = */ false,
|
|
||||||
/*.ctx = */ &ctx,
|
|
||||||
};
|
|
||||||
struct gguf_context * ctx_gguf = gguf_init_from_file(load_info.fname.c_str(), params);
|
|
||||||
if (!ctx_gguf) {
|
|
||||||
fprintf(stderr, "%s: failed to load control vector from %s\n", __func__, load_info.fname.c_str());
|
|
||||||
ggml_free(ctx);
|
|
||||||
return result;
|
|
||||||
}
|
|
||||||
|
|
||||||
// do not store data for layer 0 (it's not used)
|
// split on '.'
|
||||||
result.data.resize(result.n_embd * max_direction_layer);
|
size_t dotpos = name.find('.');
|
||||||
|
if (dotpos != std::string::npos && name.substr(0, dotpos) == "direction") {
|
||||||
for (uint32_t il = 1; il <= max_direction_layer; il++) {
|
try {
|
||||||
const std::string name = "direction." + std::to_string(il);
|
layer_idx = std::stoi(name.substr(dotpos + 1));
|
||||||
const ggml_tensor * tensor = ggml_get_tensor(ctx, name.c_str());
|
} catch (...) {
|
||||||
|
layer_idx = -1;
|
||||||
float * dst = result.data.data() + result.n_embd * (il - 1);
|
|
||||||
|
|
||||||
if (tensor) {
|
|
||||||
const float * src = (const float *) tensor->data;
|
|
||||||
for (int j = 0; j < result.n_embd; j++) {
|
|
||||||
dst[j] = src[j] * load_info.strength;
|
|
||||||
}
|
|
||||||
} else {
|
|
||||||
for (int j = 0; j < result.n_embd; j++) {
|
|
||||||
dst[j] = 0.0f;
|
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
if (layer_idx < 0) {
|
||||||
|
fprintf(stderr, "%s: invalid/unparsable direction tensor layer index in %s\n", __func__, load_info.fname.c_str());
|
||||||
|
result.n_embd = -1;
|
||||||
|
break;
|
||||||
|
} else if (layer_idx == 0) {
|
||||||
|
fprintf(stderr, "%s: invalid (zero) direction tensor layer index in %s\n", __func__, load_info.fname.c_str());
|
||||||
|
result.n_embd = -1;
|
||||||
|
break;
|
||||||
|
}
|
||||||
|
|
||||||
|
struct ggml_tensor * tensor = ggml_get_tensor(ctx, name.c_str());
|
||||||
|
if (tensor->type != GGML_TYPE_F32) {
|
||||||
|
fprintf(stderr, "%s: invalid (non-F32) direction tensor type in %s\n", __func__, load_info.fname.c_str());
|
||||||
|
result.n_embd = -1;
|
||||||
|
break;
|
||||||
|
}
|
||||||
|
if (ggml_n_dims(tensor) != 1) {
|
||||||
|
fprintf(stderr, "%s: invalid (non-1D) direction tensor shape in %s\n", __func__, load_info.fname.c_str());
|
||||||
|
result.n_embd = -1;
|
||||||
|
break;
|
||||||
|
}
|
||||||
|
|
||||||
|
if (result.n_embd == -1) {
|
||||||
|
result.n_embd = ggml_nelements(tensor);
|
||||||
|
} else if (ggml_nelements(tensor) != result.n_embd) {
|
||||||
|
fprintf(stderr, "%s: direction tensor in %s does not match previous dimensions\n", __func__, load_info.fname.c_str());
|
||||||
|
result.n_embd = -1;
|
||||||
|
break;
|
||||||
|
}
|
||||||
|
|
||||||
|
// extend if necessary - do not store data for layer 0 (it's not used)
|
||||||
|
result.data.resize(std::max(result.data.size(), static_cast<size_t>(result.n_embd * layer_idx)), 0.0f);
|
||||||
|
|
||||||
|
const float * src = (const float *) tensor->data;
|
||||||
|
float * dst = result.data.data() + result.n_embd * (layer_idx - 1); // layer 1 at [0]
|
||||||
|
for (int j = 0; j < result.n_embd; j++) {
|
||||||
|
dst[j] += src[j] * load_info.strength; // allows multiple directions for same layer in same file
|
||||||
|
}
|
||||||
|
|
||||||
}
|
}
|
||||||
|
|
||||||
|
if (result.n_embd == -1) {
|
||||||
|
fprintf(stderr, "%s: skipping %s due to invalid direction tensors\n", __func__, load_info.fname.c_str());
|
||||||
|
result.data.clear();
|
||||||
|
}
|
||||||
|
|
||||||
|
gguf_free(ctx_gguf);
|
||||||
|
ggml_free(ctx);
|
||||||
|
|
||||||
return result;
|
return result;
|
||||||
}
|
}
|
||||||
|
|
||||||
@ -2933,16 +2895,19 @@ llama_control_vector_data llama_control_vector_load(const std::vector<llama_cont
|
|||||||
auto cur = llama_control_vector_load_one(info);
|
auto cur = llama_control_vector_load_one(info);
|
||||||
|
|
||||||
if (cur.n_embd == -1) {
|
if (cur.n_embd == -1) {
|
||||||
return result;
|
result.n_embd = -1;
|
||||||
|
break;
|
||||||
}
|
}
|
||||||
if (result.n_embd != -1 && (result.n_embd != cur.n_embd || result.data.size() != cur.data.size())) {
|
if (result.n_embd != -1 && result.n_embd != cur.n_embd) {
|
||||||
fprintf(stderr, "%s: control vector in %s does not match previous vector dimensions\n", __func__, info.fname.c_str());
|
fprintf(stderr, "%s: control vectors in %s does not match previous dimensions\n", __func__, info.fname.c_str());
|
||||||
return result;
|
result.n_embd = -1;
|
||||||
|
break;
|
||||||
}
|
}
|
||||||
|
|
||||||
if (result.n_embd == -1) {
|
if (result.n_embd == -1) {
|
||||||
result = std::move(cur);
|
result = std::move(cur);
|
||||||
} else {
|
} else {
|
||||||
|
result.data.resize(std::max(result.data.size(), cur.data.size()), 0.0f); // extend if necessary
|
||||||
for (size_t i = 0; i < cur.data.size(); i++) {
|
for (size_t i = 0; i < cur.data.size(); i++) {
|
||||||
result.data[i] += cur.data[i];
|
result.data[i] += cur.data[i];
|
||||||
}
|
}
|
||||||
@ -2950,7 +2915,8 @@ llama_control_vector_data llama_control_vector_load(const std::vector<llama_cont
|
|||||||
}
|
}
|
||||||
|
|
||||||
if (result.n_embd == -1) {
|
if (result.n_embd == -1) {
|
||||||
fprintf(stderr, "%s: no vectors passed\n", __func__);
|
fprintf(stderr, "%s: no valid control vector files passed\n", __func__);
|
||||||
|
result.data.clear();
|
||||||
}
|
}
|
||||||
|
|
||||||
return result;
|
return result;
|
||||||
|
|||||||
Loading…
Reference in New Issue
Block a user