Skip to content

feat: Add native Gemma3 support for ONNX export #92

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

feat: Add native Gemma3 support for ONNX export #92

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
9 changes: 9 additions & 0 deletions optimum/exporters/onnx/model_configs.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -517,6 +517,15 @@ class Gemma2OnnxConfig(TextDecoderOnnxConfig):
MIN_TRANSFORMERS_VERSION = version.parse("4.53.0")


@register_tasks_manager_onnx("gemma3", *[*COMMON_TEXT_GENERATION_TASKS, "text-classification"])
class Gemma3OnnxConfig(TextDecoderOnnxConfig):
NORMALIZED_CONFIG_CLASS = NormalizedTextConfig
DUMMY_INPUT_GENERATOR_CLASSES = (DummyTextInputGenerator, GemmaDummyPastKeyValuesGenerator)
DUMMY_PKV_GENERATOR_CLASS = GemmaDummyPastKeyValuesGenerator
# Gemma 3 was added in transformers v4.50.0
MIN_TRANSFORMERS_VERSION = version.parse("4.50.0")


@register_tasks_manager_onnx("gpt_oss", *COMMON_TEXT_GENERATION_TASKS)
class GPTOssOnnxConfig(GemmaOnnxConfig):
MIN_TRANSFORMERS_VERSION = version.parse("4.55.0")
Expand Down
2 changes: 2 additions & 0 deletions optimum/exporters/onnx/utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -38,6 +38,8 @@
"cohere",
"falcon",
"gemma",
"gemma2",
"gemma3",
"glm",
"gpt2",
"gpt_bigcode",
Expand Down
2 changes: 2 additions & 0 deletions tests/exporters/onnx/utils_tests.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -103,6 +103,7 @@
"flaubert": "hf-internal-testing/tiny-random-flaubert",
"gemma": "fxmarty/tiny-random-GemmaForCausalLM",
"gemma2": "hf-internal-testing/tiny-random-Gemma2ForCausalLM",
"gemma3": "hf-internal-testing/tiny-random-Gemma3ForCausalLM",
"glm": "hf-internal-testing/tiny-random-GlmForCausalLM",
"glpn": "hf-internal-testing/tiny-random-GLPNModel",
"gpt2": "hf-internal-testing/tiny-random-gpt2",
Expand Down Expand Up @@ -270,6 +271,7 @@
"encoder-decoder": "patrickvonplaten/bert2bert_cnn_daily_mail",
"flaubert": "flaubert/flaubert_small_cased",
"gemma": "google/gemma-2b",
"gemma3": "google/gemma-2-2b",
"gpt2": "gpt2",
"gpt_neo": "EleutherAI/gpt-neo-125M",
"gpt_neox": "EleutherAI/gpt-neox-20b",
Expand Down
3 changes: 3 additions & 0 deletions tests/onnxruntime/test_decoder.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -33,6 +33,7 @@
CohereOnnxConfig,
DeepSeekV3OnnxConfig,
Gemma2OnnxConfig,
Gemma3OnnxConfig,
GemmaOnnxConfig,
GLMOnnxConfig,
GPTOssOnnxConfig,
Expand Down Expand Up @@ -118,6 +119,8 @@ class ORTModelForCausalLMIntegrationTest(ORTModelTestMixin):
SUPPORTED_ARCHITECTURES.append("gemma")
if is_transformers_version(">=", str(Gemma2OnnxConfig.MIN_TRANSFORMERS_VERSION)):
SUPPORTED_ARCHITECTURES.append("gemma2")
if is_transformers_version(">=", str(Gemma3OnnxConfig.MIN_TRANSFORMERS_VERSION)):
SUPPORTED_ARCHITECTURES.append("gemma3")
if is_transformers_version(">=", str(GLMOnnxConfig.MIN_TRANSFORMERS_VERSION)):
SUPPORTED_ARCHITECTURES.append("glm")
if is_transformers_version(">=", str(MPTOnnxConfig.MIN_TRANSFORMERS_VERSION)):
Expand Down
1 change: 1 addition & 0 deletions tests/onnxruntime/testing_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -66,6 +66,7 @@
"flux": "optimum-internal-testing/tiny-random-flux",
"gemma": "fxmarty/tiny-random-GemmaForCausalLM",
"gemma2": "hf-internal-testing/tiny-random-Gemma2ForCausalLM",
"gemma3": "hf-internal-testing/tiny-random-Gemma3ForCausalLM",
"glm": "hf-internal-testing/tiny-random-GlmForCausalLM",
"gpt2": "hf-internal-testing/tiny-random-GPT2LMHeadModel",
"gpt_bigcode": "hf-internal-testing/tiny-random-GPTBigCodeModel",
Expand Down
Loading