Skip to content

[torch] Add gemma4 LLM template and GPTQ algo config for MoE quantization #40

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
jimmy-adams wants to merge 1 commit into
base: release/0.12
Choose a base branch
from
+41 −0
Open

[torch] Add gemma4 LLM template and GPTQ algo config for MoE quantization #40

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
15 changes: 15 additions & 0 deletions quark/torch/quantization/config/algo_configs.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -561,6 +561,21 @@
],
model_decoder_layers="model.language_model.layers",
),
"gemma4": GPTQConfig(
inside_layer_modules=[
"self_attn.k_proj",
"self_attn.v_proj",
"self_attn.q_proj",
"self_attn.o_proj",
"mlp.up_proj",
"mlp.gate_proj",
"mlp.down_proj",
"experts.*.up_proj",
"experts.*.gate_proj",
"experts.*.down_proj",
],
model_decoder_layers="model.language_model.layers",
),
"gpt_oss": GPTQConfig(
inside_layer_modules=[
"self_attn.q_proj",
Expand Down
26 changes: 26 additions & 0 deletions quark/torch/quantization/config/template.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -710,6 +710,7 @@ def get(cls, model_type: str) -> LLMTemplate:
- gemma2
- gemma3
- gemma3_text
- gemma4
- glm4_moe
- glm4_moe_lite
- glm_moe_dsa
Expand Down Expand Up @@ -1096,6 +1097,31 @@ def _set_layer_type_config(self, config: QConfig, layer_type_config: dict[type[n
"q_layer_name": "*q_proj",
"exclude_layers_name": ["*lm_head"],
},
"gemma4": {
"kv_layers_name": ["*language_model.*k_proj", "*language_model.*v_proj"],
"q_layer_name": "*language_model.*q_proj",
"exclude_layers_name": [
"*vision_tower*",
"*embed_vision*",
"*audio_tower*",
"*embed_audio*",
"*multi_modal_projector*",
"*lm_head",
"*router.proj",
],
"f2f_weight_converters": [
WeightConverter(
"gate_up_proj",
["gate_proj.weight", "up_proj.weight"],
operations=[SplitFusedExperts(split_axis=0)],
),
WeightConverter(
"down_proj",
["down_proj.weight"],
operations=[SplitFusedExperts(split_axis=0)],
),
],
},
"glm4_moe": {
"kv_layers_name": ["*k_proj", "*v_proj"],
"q_layer_name": "*q_proj",
Expand Down