From 8613aac482b504b185429e980842af61cf27c272 Mon Sep 17 00:00:00 2001
From: Yong Yue <yongyue@microsoft.com>
Date: Fri, 24 Jul 2026 16:15:38 +0800
Subject: [PATCH] recipe(wav2vec2): add
 jonatasgrosman/wav2vec2-large-xlsr-53-polish ASR recipes

Add CPU fp32 and fp16 recipes for this Wav2Vec2ForCTC Polish ASR model.
The recipe forces model_class=AutoModelForCTC to work around the default
AutoModelForSpeechSeq2Seq loader which rejects Wav2Vec2Config.
---
 ...omatic-speech-recognition_fp16_config.json | 53 +++++++++++++++++++
 ...omatic-speech-recognition_fp32_config.json | 34 ++++++++++++
 2 files changed, 87 insertions(+)
 create mode 100644 examples/recipes/jonatasgrosman_wav2vec2-large-xlsr-53-polish/cpu/cpu/automatic-speech-recognition_fp16_config.json
 create mode 100644 examples/recipes/jonatasgrosman_wav2vec2-large-xlsr-53-polish/cpu/cpu/automatic-speech-recognition_fp32_config.json

diff --git a/examples/recipes/jonatasgrosman_wav2vec2-large-xlsr-53-polish/cpu/cpu/automatic-speech-recognition_fp16_config.json b/examples/recipes/jonatasgrosman_wav2vec2-large-xlsr-53-polish/cpu/cpu/automatic-speech-recognition_fp16_config.json
new file mode 100644
index 000000000..a49f5fbfb
--- /dev/null
+++ b/examples/recipes/jonatasgrosman_wav2vec2-large-xlsr-53-polish/cpu/cpu/automatic-speech-recognition_fp16_config.json
@@ -0,0 +1,53 @@
+{
+  "export": {
+    "opset_version": 17,
+    "batch_size": 1,
+    "export_params": true,
+    "do_constant_folding": true,
+    "verbose": false,
+    "dynamo": false,
+    "enable_hierarchy_tags": true,
+    "clean_onnx": false,
+    "hierarchy_tag_format": "full",
+    "input_tensors": [
+      {
+        "name": "input_values",
+        "dtype": "float32",
+        "shape": [1, 16000],
+        "value_range": [-1, 1]
+      }
+    ],
+    "output_tensors": [
+      {
+        "name": "logits"
+      }
+    ]
+  },
+  "optim": {},
+  "quant": {
+    "mode": "fp16",
+    "samples": 10,
+    "calibration_method": "minmax",
+    "weight_type": "uint8",
+    "activation_type": "uint8",
+    "per_channel": false,
+    "symmetric": false,
+    "weight_symmetric": null,
+    "activation_symmetric": null,
+    "save_calibration": false,
+    "distribution": "uniform",
+    "seed": null,
+    "calibration_load_path": null,
+    "calibration_save_path": null,
+    "op_types_to_quantize": null,
+    "nodes_to_exclude": null,
+    "fp16_keep_io_types": true,
+    "fp16_op_block_list": null
+  },
+  "compile": null,
+  "loader": {
+    "task": "automatic-speech-recognition",
+    "model_class": "AutoModelForCTC",
+    "model_type": "wav2vec2"
+  }
+}
diff --git a/examples/recipes/jonatasgrosman_wav2vec2-large-xlsr-53-polish/cpu/cpu/automatic-speech-recognition_fp32_config.json b/examples/recipes/jonatasgrosman_wav2vec2-large-xlsr-53-polish/cpu/cpu/automatic-speech-recognition_fp32_config.json
new file mode 100644
index 000000000..66380c538
--- /dev/null
+++ b/examples/recipes/jonatasgrosman_wav2vec2-large-xlsr-53-polish/cpu/cpu/automatic-speech-recognition_fp32_config.json
@@ -0,0 +1,34 @@
+{
+  "export": {
+    "opset_version": 17,
+    "batch_size": 1,
+    "export_params": true,
+    "do_constant_folding": true,
+    "verbose": false,
+    "dynamo": false,
+    "enable_hierarchy_tags": true,
+    "clean_onnx": false,
+    "hierarchy_tag_format": "full",
+    "input_tensors": [
+      {
+        "name": "input_values",
+        "dtype": "float32",
+        "shape": [1, 16000],
+        "value_range": [-1, 1]
+      }
+    ],
+    "output_tensors": [
+      {
+        "name": "logits"
+      }
+    ]
+  },
+  "optim": {},
+  "quant": null,
+  "compile": null,
+  "loader": {
+    "task": "automatic-speech-recognition",
+    "model_class": "AutoModelForCTC",
+    "model_type": "wav2vec2"
+  }
+}