Skip to content
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

[NPU] Modify IPEX_LLM_NPU_DISABLE_COMPILE_OPT setting for long input #12537

Merged
merged 2 commits into from
Dec 13, 2024
Merged

[NPU] Modify IPEX_LLM_NPU_DISABLE_COMPILE_OPT setting for long input #12537

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
ddfb3cd
add limitation
plusbang Dec 13, 2024
7bf1600
fix code style
plusbang Dec 13, 2024
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
8 changes: 5 additions & 3 deletions python/llm/src/ipex_llm/transformers/npu_model.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -290,7 +290,8 @@ def optimize_npu_model(cls, *args, **kwargs):
model.config.update({"group_size": quantization_group_size})
model.config.update({"asym": qtype == "asym_int4_rtn"})
optimize_llm_pre(model, qtype, mixed_precision,
quantization_group_size=quantization_group_size)
quantization_group_size=quantization_group_size,
max_prompt_len=max_prompt_len)
cls.load_convert(qtype, model, "cpu", modules_to_not_convert,
quantization_group_size, imatrix_data,
*args, **kwargs)
Expand Down Expand Up @@ -580,7 +581,7 @@ def load_low_bit(cls, pretrained_model_name_or_path: str, *model_args, **kwargs)
with torch.no_grad():
optimize_llm_pre(model, qtype, mixed_precision,
quantization_group_size=quantization_group_size,
load=bigdl_lcmu_enabled)
load=bigdl_lcmu_enabled, max_prompt_len=max_prompt_len)
cls.load_convert(qtype, model, quant_device, modules_to_not_convert,
quantization_group_size, *model_args, **kwargs)
create_npu_kernels(llm)
Expand Down Expand Up @@ -804,7 +805,8 @@ def optimize_npu_model(cls, *args, **kwargs):

with torch.no_grad():
optimize_llm_pre(model, qtype, mixed_precision,
quantization_group_size=quantization_group_size)
quantization_group_size=quantization_group_size,
max_prompt_len=max_prompt_len)
cls.load_convert_fp16(qtype, model.encoder, "cpu", modules_to_not_convert,
quantization_group_size)
create_npu_kernels(model.encoder)
Expand Down
9 changes: 8 additions & 1 deletion python/llm/src/ipex_llm/transformers/npu_models/convert_mp.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,7 +31,7 @@ def convert_forward(m, target_m, new_forward):


def optimize_llm_pre(model: torch.nn.Module, qtype, mixed_precision,
quantization_group_size=0, load=False):
quantization_group_size=0, load=False, max_prompt_len=512):
if model.config.model_type == "baichuan":
# process NormHead module in Baichuan2 7B
if hasattr(model, 'lm_head') and model.lm_head is not None:
Expand All @@ -48,6 +48,13 @@ def optimize_llm_pre(model: torch.nn.Module, qtype, mixed_precision,

cpu_lm_head = os.environ.get("IPEX_LLM_CPU_LM_HEAD", "0") != "0"

# workaround for long input performance of llama3.2-3b and glm-edge-4b CW
if os.environ.get("IPEX_LLM_NPU_DISABLE_COMPILE_OPT") is None:
plusbang marked this conversation as resolved.
Show resolved Hide resolved
disable_compile_opt = model.config.model_type == "llama" and \
model.config.hidden_size == 3072 and max_prompt_len >= 1920 and \
quantization_group_size == 0
os.environ["IPEX_LLM_NPU_DISABLE_COMPILE_OPT"] = "1" if disable_compile_opt else "0"

# workaround for MiniCPM-2B
if model.config.model_type == "minicpm" and model.config.num_hidden_layers == 40:
# 73440 is vocab_size of MiniCPM-1B
Expand Down
Loading