Add ESM onnx support (#1581)

* Add ESM onnx support * set default opset=12 --------- Co-authored-by: Félix Marty <[email protected]>
huggingface · Dec 13, 2023 · be87d2c · be87d2c
1 parent 9294d76
commit be87d2c
Show file tree

Hide file tree

Showing 4 changed files with 23 additions and 0 deletions.
diff --git a/docs/source/exporters/onnx/overview.mdx b/docs/source/exporters/onnx/overview.mdx
@@ -42,6 +42,7 @@ Supported architectures:
 - Donut-Swin
 - Electra
 - Encoder Decoder
+- ESM
 - Falcon
 - Flaubert
 - GPT-2

diff --git a/optimum/exporters/onnx/model_configs.py b/optimum/exporters/onnx/model_configs.py
@@ -183,6 +183,20 @@ class DebertaV2OnnxConfig(DebertaOnnxConfig):
     pass
 
 
+class EsmOnnxConfig(TextEncoderOnnxConfig):
+    NORMALIZED_CONFIG_CLASS = NormalizedTextConfig
+    ATOL_FOR_VALIDATION = 1e-4
+    DEFAULT_ONNX_OPSET = 12
+
+    @property
+    def inputs(self) -> Dict[str, Dict[int, str]]:
+        dynamic_axis = {0: "batch_size", 1: "sequence_length"}
+        return {
+            "input_ids": dynamic_axis,
+            "attention_mask": dynamic_axis,
+        }
+
+
 class GPT2OnnxConfig(TextDecoderWithPositionIdsOnnxConfig):
     DEFAULT_ONNX_OPSET = 13
     NORMALIZED_CONFIG_CLASS = NormalizedTextConfig.with_args(num_layers="n_layer", num_attention_heads="n_head")

diff --git a/optimum/exporters/tasks.py b/optimum/exporters/tasks.py
@@ -525,6 +525,13 @@ class TasksManager:
             "text2text-generation-with-past",
             onnx="EncoderDecoderOnnxConfig",
         ),
+        "esm": supported_tasks_mapping(
+            "feature-extraction",
+            "fill-mask",
+            "text-classification",
+            "token-classification",
+            onnx="EsmOnnxConfig",
+        ),
         "falcon": supported_tasks_mapping(
             "feature-extraction",
             "feature-extraction-with-past",

diff --git a/tests/exporters/exporters_utils.py b/tests/exporters/exporters_utils.py
@@ -74,6 +74,7 @@
         ],
         "mohitsha/tiny-random-testing-bert2gpt2": ["text2text-generation", "text2text-generation-with-past"],
     },
+    "esm": "hf-internal-testing/tiny-random-EsmModel",
     "falcon": {
         "fxmarty/really-tiny-falcon-testing": [
             "feature-extraction",
-Original file line number
+Diff line change
@@ Expand Up / @@ -42,6 +42,7 @@ Supported architectures: @@
     - Donut-Swin
     - Electra
     - Encoder Decoder
+    - ESM
     - Falcon
     - Flaubert
     - GPT-2
@@ Expand Down @@