Skip to content

[NPUW] Support Eagle3 Topk #34349

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
GuoliangShiIntel wants to merge 1 commit into
base: master
Choose a base branch
from
Open

[NPUW] Support Eagle3 Topk #34349

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
30 changes: 22 additions & 8 deletions src/plugins/intel_npu/src/plugin/npuw/llm_compiled_model.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1094,7 +1094,8 @@ void reshape_to_static(std::shared_ptr<ov::Model> model,
const uint32_t kvcache_size,
const KVAxesPosition& kv_axes_position,
const uint32_t lora_rank,
const uint32_t lhs_seq_size = 0) {
const uint32_t lhs_seq_size = 0,
const bool is_prefill = false) {
std::map<std::string, ov::PartialShape> new_shapes;
for (const auto& input : model->inputs()) {
const auto& input_name = input.get_any_name();
Expand Down Expand Up @@ -1129,8 +1130,9 @@ void reshape_to_static(std::shared_ptr<ov::Model> model,
const auto& partial_shape = input.get_partial_shape();
new_shape = partial_shape;
new_shape[0] = 1; // batch_dim
} else if (ov::npuw::matchEagle3HiddenStatesString(input_name)) {
new_shape = ov::npuw::Eagle3Extension::get_static_input(model, input, input_size);
} else if (ov::npuw::matchEagle3HiddenStatesString(input_name) ||
ov::npuw::matchEagle3TreeMaskString(input_name)) {
new_shape = ov::npuw::Eagle3Extension::get_static_input(model, input, input_size, kvcache_size, is_prefill);
} else if (ov::npuw::util::matchLoRAMatMulAString(input_name)) {
new_shape = ov::PartialShape({lora_rank, input.get_partial_shape()[1]});
} else if (ov::npuw::util::matchLoRAMatMulAlphaString(input_name)) {
Expand Down Expand Up @@ -1643,22 +1645,31 @@ class ReshapeToStatic : public ov::pass::ModelPass {
KVAxesPosition m_kv_axes_position;
uint32_t m_lora_rank;
uint32_t m_lhs_seq_size;
bool m_is_prefill;

public:
OPENVINO_MODEL_PASS_RTTI("ov::npuw::ReshapeToStatic");
explicit ReshapeToStatic(const uint32_t input_size,
const uint32_t kvcache_size,
const KVAxesPosition& kv_axes_position,
const uint32_t lora_rank,
const uint32_t lhs_seq_size = 0)
const uint32_t lhs_seq_size = 0,
const bool is_prefill = false)
: m_input_size(input_size),
m_kvcache_size(kvcache_size),
m_kv_axes_position(kv_axes_position),
m_lora_rank(lora_rank),
m_lhs_seq_size(lhs_seq_size) {}
m_lhs_seq_size(lhs_seq_size),
m_is_prefill(is_prefill) {}

bool run_on_model(const std::shared_ptr<ov::Model>& model) override {
reshape_to_static(model, m_input_size, m_kvcache_size, m_kv_axes_position, m_lora_rank, m_lhs_seq_size);
reshape_to_static(model,
m_input_size,
m_kvcache_size,
m_kv_axes_position,
m_lora_rank,
m_lhs_seq_size,
m_is_prefill);

return true;
}
Expand Down Expand Up @@ -2021,14 +2032,17 @@ ov::npuw::LLMCompiledModel::LLMCompiledModel(const std::shared_ptr<ov::Model>& m
ReshapeToStatic(static_cast<uint32_t>(m_prefill_chunk_size),
m_kvcache_desc.max_prompt_size,
axes,
m_max_lora_rank)
m_max_lora_rank,
0,
true)
.run_on_model(prefill_model);
} else {
ReshapeToStatic(m_kvcache_desc.max_prompt_size,
m_kvcache_desc.max_prompt_size,
axes,
m_max_lora_rank,
whisper_lhs_seq_size)
whisper_lhs_seq_size,
true)
.run_on_model(prefill_model);
}
LOG_DEBUG("Make kvcache model with static shapes");
Expand Down
Loading
Loading