Upload folder using huggingface_hub

.gitattributes

@@ -33,3 +33,8 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+shared/tokenizer/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+variants/cpu-int4/inference/frozen_base.onnx.data filter=lfs diff=lfs merge=lfs -text
+variants/cpu-int4/inference/model.onnx_data filter=lfs diff=lfs merge=lfs -text
+variants/cpu-int4/inference/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+variants/cpu-int4/train/checkpoint filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,40 @@
+---
+base_model: google/functiongemma-270m-it
+library_name: mobiletransformers
+pipeline_tag: text-generation
+tags:
+  - mobiletransformers
+  - onnx
+  - on-device
+  - android
+  - lora
+  - int4
+---
+
+# google/functiongemma-270m-it — MobileTransformers package
+
+On-device (Android) package exported from **google/functiongemma-270m-it** with MobileTransformers.
+
+## Provenance
+- Base model: `google/functiongemma-270m-it`
+- Selected task: `text-generation-with-past`
+- PEFT methods: lora
+- Quantization: int4
+- Toolchain: optimum-onnx 0.1.0, transformers 4.57.6, ort-training 1.23.0+cpu, ort-genai None
+
+## Licenses
+- Framework: not declared in this package — see the repository
+- Base model weights: see the base model above (this package redistributes an export of those weights, so their terms govern its contents)
+
+## Android runtime
+- Minimum API: 28
+- Recommended device memory (MB): None
+- Required ABIs: any
+
+## Variants
+
+| id | EP | quant | engines | features | min API | rec. RAM (MB) |
+| --- | --- | --- | --- | --- | --- | --- |
+| cpu-int4 | cpu | int4 | native | core, inference, train | 28 | None |
+
+Default variant: `cpu-int4`.

mobiletransformers_manifest.json

@@ -0,0 +1,293 @@
+{
+  "androidRuntime": {
+    "minimumAndroidApi": 28,
+    "recommendedDeviceMemoryMb": null,
+    "requiredAbis": []
+  },
+  "architectures": [
+    "gemma3_text"
+  ],
+  "artifactFormatVersion": 1,
+  "baseModelId": "google/functiongemma-270m-it",
+  "defaultVariant": "cpu-int4",
+  "downloadPlan": {
+    "cpu-int4": {
+      "checksums": [
+        "variants/cpu-int4/checksums.json"
+      ],
+      "core": [
+        "mobiletransformers_manifest.json",
+        "shared/tokenizer/**",
+        "shared/chat_template.jinja",
+        "shared/config.json",
+        "shared/generation_config.json"
+      ],
+      "genai": [],
+      "inference": [
+        "variants/cpu-int4/inference/**"
+      ],
+      "rag": [],
+      "train": [
+        "variants/cpu-int4/train/**"
+      ]
+    }
+  },
+  "exportedAt": "",
+  "fileSizes": {
+    "optimum/export_report.json": 822,
+    "optimum/supported_tasks.json": 34,
+    "shared/tokenizer/chat_template.jinja": 13792,
+    "shared/tokenizer/mobiletransformers_tokenizer_config.json": 356,
+    "shared/tokenizer/special_tokens_map.json": 706,
+    "shared/tokenizer/tokenizer.json": 33384899,
+    "shared/tokenizer/tokenizer_config.json": 1155714,
+    "variants/cpu-int4/checksums.json": 13424,
+    "variants/cpu-int4/inference/chat_template.jinja": 13792,
+    "variants/cpu-int4/inference/config.json": 1358,
+    "variants/cpu-int4/inference/frozen_base.onnx.data": 1684498944,
+    "variants/cpu-int4/inference/genai_config.json": 1044,
+    "variants/cpu-int4/inference/generation_config.json": 176,
+    "variants/cpu-int4/inference/merger_lora_fpin_fpout.onnx": 811,
+    "variants/cpu-int4/inference/model.layers.0.self_attn.q_proj.MatMul.weight.bin": 2621440,
+    "variants/cpu-int4/inference/model.layers.0.self_attn.q_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.0.self_attn.v_proj.MatMul.weight.bin": 655360,
+    "variants/cpu-int4/inference/model.layers.0.self_attn.v_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.1.self_attn.q_proj.MatMul.weight.bin": 2621440,
+    "variants/cpu-int4/inference/model.layers.1.self_attn.q_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.1.self_attn.v_proj.MatMul.weight.bin": 655360,
+    "variants/cpu-int4/inference/model.layers.1.self_attn.v_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.10.self_attn.q_proj.MatMul.weight.bin": 2621440,
+    "variants/cpu-int4/inference/model.layers.10.self_attn.q_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.10.self_attn.v_proj.MatMul.weight.bin": 655360,
+    "variants/cpu-int4/inference/model.layers.10.self_attn.v_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.11.self_attn.q_proj.MatMul.weight.bin": 2621440,
+    "variants/cpu-int4/inference/model.layers.11.self_attn.q_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.11.self_attn.v_proj.MatMul.weight.bin": 655360,
+    "variants/cpu-int4/inference/model.layers.11.self_attn.v_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.12.self_attn.q_proj.MatMul.weight.bin": 2621440,
+    "variants/cpu-int4/inference/model.layers.12.self_attn.q_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.12.self_attn.v_proj.MatMul.weight.bin": 655360,
+    "variants/cpu-int4/inference/model.layers.12.self_attn.v_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.13.self_attn.q_proj.MatMul.weight.bin": 2621440,
+    "variants/cpu-int4/inference/model.layers.13.self_attn.q_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.13.self_attn.v_proj.MatMul.weight.bin": 655360,
+    "variants/cpu-int4/inference/model.layers.13.self_attn.v_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.14.self_attn.q_proj.MatMul.weight.bin": 2621440,
+    "variants/cpu-int4/inference/model.layers.14.self_attn.q_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.14.self_attn.v_proj.MatMul.weight.bin": 655360,
+    "variants/cpu-int4/inference/model.layers.14.self_attn.v_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.15.self_attn.q_proj.MatMul.weight.bin": 2621440,
+    "variants/cpu-int4/inference/model.layers.15.self_attn.q_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.15.self_attn.v_proj.MatMul.weight.bin": 655360,
+    "variants/cpu-int4/inference/model.layers.15.self_attn.v_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.16.self_attn.q_proj.MatMul.weight.bin": 2621440,
+    "variants/cpu-int4/inference/model.layers.16.self_attn.q_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.16.self_attn.v_proj.MatMul.weight.bin": 655360,
+    "variants/cpu-int4/inference/model.layers.16.self_attn.v_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.17.self_attn.q_proj.MatMul.weight.bin": 2621440,
+    "variants/cpu-int4/inference/model.layers.17.self_attn.q_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.17.self_attn.v_proj.MatMul.weight.bin": 655360,
+    "variants/cpu-int4/inference/model.layers.17.self_attn.v_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.2.self_attn.q_proj.MatMul.weight.bin": 2621440,
+    "variants/cpu-int4/inference/model.layers.2.self_attn.q_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.2.self_attn.v_proj.MatMul.weight.bin": 655360,
+    "variants/cpu-int4/inference/model.layers.2.self_attn.v_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.3.self_attn.q_proj.MatMul.weight.bin": 2621440,
+    "variants/cpu-int4/inference/model.layers.3.self_attn.q_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.3.self_attn.v_proj.MatMul.weight.bin": 655360,
+    "variants/cpu-int4/inference/model.layers.3.self_attn.v_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.4.self_attn.q_proj.MatMul.weight.bin": 2621440,
+    "variants/cpu-int4/inference/model.layers.4.self_attn.q_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.4.self_attn.v_proj.MatMul.weight.bin": 655360,
+    "variants/cpu-int4/inference/model.layers.4.self_attn.v_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.5.self_attn.q_proj.MatMul.weight.bin": 2621440,
+    "variants/cpu-int4/inference/model.layers.5.self_attn.q_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.5.self_attn.v_proj.MatMul.weight.bin": 655360,
+    "variants/cpu-int4/inference/model.layers.5.self_attn.v_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.6.self_attn.q_proj.MatMul.weight.bin": 2621440,
+    "variants/cpu-int4/inference/model.layers.6.self_attn.q_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.6.self_attn.v_proj.MatMul.weight.bin": 655360,
+    "variants/cpu-int4/inference/model.layers.6.self_attn.v_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.7.self_attn.q_proj.MatMul.weight.bin": 2621440,
+    "variants/cpu-int4/inference/model.layers.7.self_attn.q_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.7.self_attn.v_proj.MatMul.weight.bin": 655360,
+    "variants/cpu-int4/inference/model.layers.7.self_attn.v_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.8.self_attn.q_proj.MatMul.weight.bin": 2621440,
+    "variants/cpu-int4/inference/model.layers.8.self_attn.q_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.8.self_attn.v_proj.MatMul.weight.bin": 655360,
+    "variants/cpu-int4/inference/model.layers.8.self_attn.v_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.9.self_attn.q_proj.MatMul.weight.bin": 2621440,
+    "variants/cpu-int4/inference/model.layers.9.self_attn.q_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.layers.9.self_attn.v_proj.MatMul.weight.bin": 655360,
+    "variants/cpu-int4/inference/model.layers.9.self_attn.v_proj.MatMul.weight.bin.sha256": 65,
+    "variants/cpu-int4/inference/model.onnx": 920639,
+    "variants/cpu-int4/inference/model.onnx_data": 1743481344,
+    "variants/cpu-int4/inference/optimum_config.json": 334,
+    "variants/cpu-int4/inference/special_tokens_map.json": 706,
+    "variants/cpu-int4/inference/tokenizer.json": 33384899,
+    "variants/cpu-int4/inference/tokenizer_config.json": 1155714,
+    "variants/cpu-int4/inference/weight_handoff_map.json": 53841,
+    "variants/cpu-int4/train/checkpoint": 314028424,
+    "variants/cpu-int4/train/eval_model.onnx": 1901828,
+    "variants/cpu-int4/train/optimizer_model.onnx": 538,
+    "variants/cpu-int4/train/trainable_parameters.json": 7011,
+    "variants/cpu-int4/train/training_config.json": 19841,
+    "variants/cpu-int4/train/training_model.onnx": 1688100
+  },
+  "license": {
+    "baseModelWeights": null,
+    "framework": null,
+    "noticeFile": null
+  },
+  "minReaderVersion": "1.0",
+  "mobiletransformersVersion": "0.1.0",
+  "onnxRuntimeGenAIVersion": null,
+  "onnxRuntimeTrainingVersion": "1.23.0+cpu",
+  "optimumOnnxVersion": "0.1.0",
+  "peftMethods": [
+    "lora"
+  ],
+  "quantization": [
+    "int4"
+  ],
+  "requiredFiles": [
+    "mobiletransformers_manifest.json",
+    "shared/tokenizer/tokenizer.json",
+    "variants/cpu-int4/inference/model.onnx"
+  ],
+  "schemaVersion": "1.0",
+  "selectedTask": "text-generation-with-past",
+  "sha256": {
+    "optimum/export_report.json": "edc6a33e6199cbc47846e57f770d37b56822796cda4df3ebfe51eeb23dbec046",
+    "optimum/supported_tasks.json": "2591d52d3133d5bd71e384d09e57f48d29e84f4e9a358697595bb4bcb56dcfdf",
+    "shared/tokenizer/chat_template.jinja": "db61fb01017bd82401d3ffca4f8e066cd56ff6d38d2a30c4058770c0bf7ab49b",
+    "shared/tokenizer/mobiletransformers_tokenizer_config.json": "2b8628a7ad20677fd280311678611f19cf97860938f18ee678083cc0e550c668",
+    "shared/tokenizer/special_tokens_map.json": "d39df6f8ddb132a622394f05a603b166ba5f22d485483e30d06c1821ef4ac9b4",
+    "shared/tokenizer/tokenizer.json": "b6b09a0b4a803ad453063ca4bb49a784540e8120004e2450e025df2b27d41fb2",
+    "shared/tokenizer/tokenizer_config.json": "c33878647bdc69ce85d4b9aeef953815afd8405f8d85001ab0bbb89aa14d2554",
+    "variants/cpu-int4/checksums.json": "2fb0f6eee9752e511aab5cea2529e31a219233d36635a9458d2f328d512515be",
+    "variants/cpu-int4/inference/chat_template.jinja": "db61fb01017bd82401d3ffca4f8e066cd56ff6d38d2a30c4058770c0bf7ab49b",
+    "variants/cpu-int4/inference/config.json": "88a795b85d50b1642e66193935a64c34326fa76efe107f9fceeafd31aa3e2af9",
+    "variants/cpu-int4/inference/frozen_base.onnx.data": "9c14b3237169fd6023199017e1949b02bbfe0ef1e80ac10c2d72bba670ebfa82",
+    "variants/cpu-int4/inference/genai_config.json": "acac04545d607a84f676049824d2e113f9367179741494a46279551b438c6efc",
+    "variants/cpu-int4/inference/generation_config.json": "90f3b346c07b98c000b6de297e805a83f504d6856b01c405a746d805c05444c2",
+    "variants/cpu-int4/inference/merger_lora_fpin_fpout.onnx": "953a8ad4704235c19ec06150788107921ba7f2578a5880a32cd59b473544712c",
+    "variants/cpu-int4/inference/model.layers.0.self_attn.q_proj.MatMul.weight.bin": "de5441ec8bb19c3a0038e7f3bd3cb25d9b81599d1e6977ceecdc1d894ede2fa1",
+    "variants/cpu-int4/inference/model.layers.0.self_attn.q_proj.MatMul.weight.bin.sha256": "b303f006c606265b707027dccaa1c22f9f19b7920a8d51e442cbcfde886d3b65",
+    "variants/cpu-int4/inference/model.layers.0.self_attn.v_proj.MatMul.weight.bin": "aced1f8faec6416a6cbee9227400021af9b1f4c0d25feddc54e255f1210575fb",
+    "variants/cpu-int4/inference/model.layers.0.self_attn.v_proj.MatMul.weight.bin.sha256": "e41d2d8d47eaea11a011971b193d1ea7f15222a23dc2c2d7342c983d5fad5d58",
+    "variants/cpu-int4/inference/model.layers.1.self_attn.q_proj.MatMul.weight.bin": "cc0c039a777068e664ea4916413b822ce497007dc31f22f7e7cc27f9ed0bd820",
+    "variants/cpu-int4/inference/model.layers.1.self_attn.q_proj.MatMul.weight.bin.sha256": "1e1651baa468edd1ec0a787271c362646bcf4669d06816750fb9907bda6161ed",
+    "variants/cpu-int4/inference/model.layers.1.self_attn.v_proj.MatMul.weight.bin": "e2cde258f00749809d3b5e6a3838dbccb866dfdc228c86e5375a1c3fb3be14b2",
+    "variants/cpu-int4/inference/model.layers.1.self_attn.v_proj.MatMul.weight.bin.sha256": "d670153828a24efe77701e8491f5bc182fd88c9f27b924c750b9c20ca4fbf89c",
+    "variants/cpu-int4/inference/model.layers.10.self_attn.q_proj.MatMul.weight.bin": "864537f8b063185411edf60cc77bb704f3ddd3d39b6f19f04bd0da2bf1ba5b5e",
+    "variants/cpu-int4/inference/model.layers.10.self_attn.q_proj.MatMul.weight.bin.sha256": "286bfb981d94d988671f4572b770728b9a285f400ec4edc6ad83c9ea499d2c40",
+    "variants/cpu-int4/inference/model.layers.10.self_attn.v_proj.MatMul.weight.bin": "f8aedd8998857b74a4e93e3182911873cd9ad547f8f0ed8212856e8f0beaef00",
+    "variants/cpu-int4/inference/model.layers.10.self_attn.v_proj.MatMul.weight.bin.sha256": "528f124d7e9657be8b57303ec6248483bb0c6dde4c84084575e3c9afd0b8188b",
+    "variants/cpu-int4/inference/model.layers.11.self_attn.q_proj.MatMul.weight.bin": "534363c1244562ba5cc4043f27b4da97fa3a48dc2a60772532816b5f11865b91",
+    "variants/cpu-int4/inference/model.layers.11.self_attn.q_proj.MatMul.weight.bin.sha256": "19a221e71a3a79d359f1168ea1a3f88003340820f8ecffae5f5472d71be6700c",
+    "variants/cpu-int4/inference/model.layers.11.self_attn.v_proj.MatMul.weight.bin": "9b3cc8d43d80eae0f674534447d7845b17a4c779fa74adb1de07254f65de03f0",
+    "variants/cpu-int4/inference/model.layers.11.self_attn.v_proj.MatMul.weight.bin.sha256": "643bbdcb8d0381f20eaf58b1d8d9ff7c3f6435ec71d5ee7d4a93d7b60d541a83",
+    "variants/cpu-int4/inference/model.layers.12.self_attn.q_proj.MatMul.weight.bin": "5213011869521ddedba25e37944d311b30632ec3b2c061caeb6671d4f4b9596f",
+    "variants/cpu-int4/inference/model.layers.12.self_attn.q_proj.MatMul.weight.bin.sha256": "92e8c3fb1b76481b06c2ff2dd3ae3a99ad52477b274511b4ba3944dc9f376858",
+    "variants/cpu-int4/inference/model.layers.12.self_attn.v_proj.MatMul.weight.bin": "8dfd86b40b9e613538f07c8297cf3fb54cb3ebf99fe4918c30c8c51bea4da5c0",
+    "variants/cpu-int4/inference/model.layers.12.self_attn.v_proj.MatMul.weight.bin.sha256": "ca5d5b8d2723dcb8a31ed895ebd6f72d3111de86981d1f10a57fd64d393a5ceb",
+    "variants/cpu-int4/inference/model.layers.13.self_attn.q_proj.MatMul.weight.bin": "3e58dbbd1a83c01943979052519976f15dd2e13b4502fe72edf522f86960d618",
+    "variants/cpu-int4/inference/model.layers.13.self_attn.q_proj.MatMul.weight.bin.sha256": "8a75590cded9ed4eb8518b18d4f5694fd4131d8ef225bfff9d148412716fe7c6",
+    "variants/cpu-int4/inference/model.layers.13.self_attn.v_proj.MatMul.weight.bin": "21666b649c20492830d5f37db41e2ba54d270228c20a532eea2104fa3c61396e",
+    "variants/cpu-int4/inference/model.layers.13.self_attn.v_proj.MatMul.weight.bin.sha256": "327d2d9dd53b71a1662bb5751d554ef719aff537095da4bd8ac2aaaea47de555",
+    "variants/cpu-int4/inference/model.layers.14.self_attn.q_proj.MatMul.weight.bin": "ef8d82392a5ac5cf37f65c638a965c7c949111ec783b0c947ecfc8d261e1ad9f",
+    "variants/cpu-int4/inference/model.layers.14.self_attn.q_proj.MatMul.weight.bin.sha256": "c183ea62d344b6331a35070bdac7091347be27f629a00a3569acf40f01c87dd2",
+    "variants/cpu-int4/inference/model.layers.14.self_attn.v_proj.MatMul.weight.bin": "c1c727d66921e1dceec4142a2a2c36254189234bd90b09db4125927e70b8e458",
+    "variants/cpu-int4/inference/model.layers.14.self_attn.v_proj.MatMul.weight.bin.sha256": "75c2d4a4ac3606a197b259bb309ada6a55341a4c3c3e68ab2b5d29985699cd88",
+    "variants/cpu-int4/inference/model.layers.15.self_attn.q_proj.MatMul.weight.bin": "31f7849f39f1b1fe0d7ef163aae56279a8f48eac0ec81dceae7dd7fda77dd293",
+    "variants/cpu-int4/inference/model.layers.15.self_attn.q_proj.MatMul.weight.bin.sha256": "c314704b435abe1cd3ba672eb1a408e7c6f7c808a215bfb7938d27f5c4583dbe",
+    "variants/cpu-int4/inference/model.layers.15.self_attn.v_proj.MatMul.weight.bin": "8f5878be79b76139fc9cc5724e91074c742fffd021b2b7bd6f0406775ec69365",
+    "variants/cpu-int4/inference/model.layers.15.self_attn.v_proj.MatMul.weight.bin.sha256": "17e8852305a2dd0be3788629535015c3ba347bd8a84a318c1a2740846ea656d3",
+    "variants/cpu-int4/inference/model.layers.16.self_attn.q_proj.MatMul.weight.bin": "cc87b80c44df0b66e627944acd40e87a0b63c77a86d1cfa55085dad5c5610de7",
+    "variants/cpu-int4/inference/model.layers.16.self_attn.q_proj.MatMul.weight.bin.sha256": "331eece77b3bc5bea26c224ee395a6c864e2e98667cd6bd0ef8a20c48aeaa27c",
+    "variants/cpu-int4/inference/model.layers.16.self_attn.v_proj.MatMul.weight.bin": "0c1f91d46f8f885ca1970f61ada3e69fe7e2a44c4f103beb456e84fb0f39cb13",
+    "variants/cpu-int4/inference/model.layers.16.self_attn.v_proj.MatMul.weight.bin.sha256": "c25e491df56cb7902bc53dfe2fdc5f69cc6c94d56a73a1e354920ace0cdedf8c",
+    "variants/cpu-int4/inference/model.layers.17.self_attn.q_proj.MatMul.weight.bin": "75aa1f482c3a16420a092f912a073abc981fc584038eead4fc4dad219c4a18df",
+    "variants/cpu-int4/inference/model.layers.17.self_attn.q_proj.MatMul.weight.bin.sha256": "5dccc067c68edcfb2aafd0ecfe966965f274afc3455699e6add6a2f4b3c0151f",
+    "variants/cpu-int4/inference/model.layers.17.self_attn.v_proj.MatMul.weight.bin": "c331cf9885fc9a470111e2e8e59018a2ffa235ee1441aa005789e321aee80df3",
+    "variants/cpu-int4/inference/model.layers.17.self_attn.v_proj.MatMul.weight.bin.sha256": "b3058d47101dd69928aa1b6e78574dd47e8a9eedda003972accac1fcbc50c53f",
+    "variants/cpu-int4/inference/model.layers.2.self_attn.q_proj.MatMul.weight.bin": "0aab3b51d26366fda266a9a030b64fa0097a3c9fe0f2ac8b45cd455ed6d00844",
+    "variants/cpu-int4/inference/model.layers.2.self_attn.q_proj.MatMul.weight.bin.sha256": "0c44bd5f5b21e9b5292a13d4069c6b313c9b9aa80881c3b469ab0034c0bedd1e",
+    "variants/cpu-int4/inference/model.layers.2.self_attn.v_proj.MatMul.weight.bin": "f995a03e11e45e51c9038b610940e2388dea7f89b186c9c32bf0774c7ad56676",
+    "variants/cpu-int4/inference/model.layers.2.self_attn.v_proj.MatMul.weight.bin.sha256": "b7f5aa1ce6d2e1be556f65fb6ca619e7591e63dd975534dae239e71290ac5a7b",
+    "variants/cpu-int4/inference/model.layers.3.self_attn.q_proj.MatMul.weight.bin": "5f256353f99cd60325cd88345d8095a1d8bf3ec2b2ed8b7dab08368018fd8335",
+    "variants/cpu-int4/inference/model.layers.3.self_attn.q_proj.MatMul.weight.bin.sha256": "00990e7e4ec37977e4c3024989b9dedab6f8c94a76755aa59ff21bdde5eae41e",
+    "variants/cpu-int4/inference/model.layers.3.self_attn.v_proj.MatMul.weight.bin": "7412ad9549d423c2cfbe4922299f5759b2e30e6839ef89cf99f8880ea8872ee5",
+    "variants/cpu-int4/inference/model.layers.3.self_attn.v_proj.MatMul.weight.bin.sha256": "e832cd832df1e7c9249d968196ff8da050eb1f4bb733e6ed1f3df80988037bad",
+    "variants/cpu-int4/inference/model.layers.4.self_attn.q_proj.MatMul.weight.bin": "ad0a0d5c5db67a43ea9793d34d5ef09e081772e8e5fba2b771ee95983ecda45a",
+    "variants/cpu-int4/inference/model.layers.4.self_attn.q_proj.MatMul.weight.bin.sha256": "6baba6d67fea470291b3da65d1cf9834f6e103b9a6845c8d753c958a2d873b5e",
+    "variants/cpu-int4/inference/model.layers.4.self_attn.v_proj.MatMul.weight.bin": "f59967d4272b6bd14e86c0cdc9fed9f1bcb21f67faa8a1d769935f710247f1b3",
+    "variants/cpu-int4/inference/model.layers.4.self_attn.v_proj.MatMul.weight.bin.sha256": "294e36d47d7ea2796ec74012dba0d6e672eaade5d8fed5c4fc15885d735fd35a",
+    "variants/cpu-int4/inference/model.layers.5.self_attn.q_proj.MatMul.weight.bin": "ff1c5fe943f8959a8b9d09640aec44d881f035f771416ec88ac929e5d4fe6b4b",
+    "variants/cpu-int4/inference/model.layers.5.self_attn.q_proj.MatMul.weight.bin.sha256": "aa4e396ee86ecd0b1538e96230fc842659e7ec66c927493682fa914efce3047b",
+    "variants/cpu-int4/inference/model.layers.5.self_attn.v_proj.MatMul.weight.bin": "4ee8c2d65e018f46002157c99b0fd025cfc523e03b07ca6fe26da71659f51c70",
+    "variants/cpu-int4/inference/model.layers.5.self_attn.v_proj.MatMul.weight.bin.sha256": "31352189895dd2e1d92b88721c0950070cdd2caa8c68c8cae31233825ec5b667",
+    "variants/cpu-int4/inference/model.layers.6.self_attn.q_proj.MatMul.weight.bin": "353d2ac53c60dfce33c57e501010ec0dac799d9050617cf6f9f818d37e1a2296",
+    "variants/cpu-int4/inference/model.layers.6.self_attn.q_proj.MatMul.weight.bin.sha256": "e869d030595d5ed13ac046acfc08fd9487b3a9ba9e940edbb4724eba90b84167",
+    "variants/cpu-int4/inference/model.layers.6.self_attn.v_proj.MatMul.weight.bin": "e5d43600c7a68cc5889563120b46780f107945e26c9385b5fca04c3a22960843",
+    "variants/cpu-int4/inference/model.layers.6.self_attn.v_proj.MatMul.weight.bin.sha256": "fd3c65bcc2e5652073fdf480a8be666611a1e20d26662e648e5cdd7fb4e0340c",
+    "variants/cpu-int4/inference/model.layers.7.self_attn.q_proj.MatMul.weight.bin": "e9d722fef4ecc4e3dffddd5b1527bcca70af70978eb812df6d7c4452c4c10b90",
+    "variants/cpu-int4/inference/model.layers.7.self_attn.q_proj.MatMul.weight.bin.sha256": "5d11e119c112c0f2cd8a46a10856aa4e3a467048928ddd0de0891dec95164b18",
+    "variants/cpu-int4/inference/model.layers.7.self_attn.v_proj.MatMul.weight.bin": "d50fe1d3c9c265fc38d169bd63fa727062f145719419564a124a5c06fbe59b04",
+    "variants/cpu-int4/inference/model.layers.7.self_attn.v_proj.MatMul.weight.bin.sha256": "bb5995d820468db5e2c676bf0960061c85fff3ccaeebb0c76b89f905b400f0bc",
+    "variants/cpu-int4/inference/model.layers.8.self_attn.q_proj.MatMul.weight.bin": "eee1373ed2ee35607cda09ceeef4c92b7ab331a419ffb795c821a58c563e884b",
+    "variants/cpu-int4/inference/model.layers.8.self_attn.q_proj.MatMul.weight.bin.sha256": "ebf030acd3a19f40d679c3e8917c15a97e122f4f8076eff7043758fefbef1b4c",
+    "variants/cpu-int4/inference/model.layers.8.self_attn.v_proj.MatMul.weight.bin": "3b9e52ee458990d65fd339e4e544d5f379b1600dd4852fc3084a32e96bc00a0b",
+    "variants/cpu-int4/inference/model.layers.8.self_attn.v_proj.MatMul.weight.bin.sha256": "6142f9fd4e7ee9f118757cb582ce4d16c3105cb389343ee8aef7dbf4b83b939f",
+    "variants/cpu-int4/inference/model.layers.9.self_attn.q_proj.MatMul.weight.bin": "68c5c1b401f009fd33795a986edbef642ed6bb2e8cfdee72f8ef5d9930a608b4",
+    "variants/cpu-int4/inference/model.layers.9.self_attn.q_proj.MatMul.weight.bin.sha256": "ec94b226ebc6be2d0eff5dfd48edb112728ec2d7fb4968481f811a87b31a16ae",
+    "variants/cpu-int4/inference/model.layers.9.self_attn.v_proj.MatMul.weight.bin": "b029f98f39c8aef6ec3676ae87106eab90d9521ff05ab1c7812963f8a1929555",
+    "variants/cpu-int4/inference/model.layers.9.self_attn.v_proj.MatMul.weight.bin.sha256": "9004e7234ef3e8231241955e74f0cfd7880d2ba47a4b6f4f43b3e86759ea7793",
+    "variants/cpu-int4/inference/model.onnx": "7816d03ce0be987e340c51ab64916f13a0f70a68630c3a2de82370a32a6fabfa",
+    "variants/cpu-int4/inference/model.onnx_data": "984595b520feb718e9dde76cc6dcd2bcc590bb67987c4da01b15a6ea54cade05",
+    "variants/cpu-int4/inference/optimum_config.json": "78abd079b5f45b977fad47a63f173701f6681d555c1372a800221831ce811e4f",
+    "variants/cpu-int4/inference/special_tokens_map.json": "d39df6f8ddb132a622394f05a603b166ba5f22d485483e30d06c1821ef4ac9b4",
+    "variants/cpu-int4/inference/tokenizer.json": "b6b09a0b4a803ad453063ca4bb49a784540e8120004e2450e025df2b27d41fb2",
+    "variants/cpu-int4/inference/tokenizer_config.json": "c33878647bdc69ce85d4b9aeef953815afd8405f8d85001ab0bbb89aa14d2554",
+    "variants/cpu-int4/inference/weight_handoff_map.json": "2b809307b885be7b6bd2908831ab37e5f18ea6d867b6fa8be2d284138e94581f",
+    "variants/cpu-int4/train/checkpoint": "cd9ea126423c721891377e36136904e64aaa4c04db73f82b8f2732aec482b1be",
+    "variants/cpu-int4/train/eval_model.onnx": "b66f69dca13c1f8a43a760f9f5b9a4cad1a9518d4914e8d0e42bd025e5ce6b5c",
+    "variants/cpu-int4/train/optimizer_model.onnx": "8ed3b548bce52bf25c2154b1452314b3c48d079e28c52822dd348034f59dc054",
+    "variants/cpu-int4/train/trainable_parameters.json": "05fddff407edb77426860ce8637b046bffd277bb053b33aa6600a321ac993958",
+    "variants/cpu-int4/train/training_config.json": "4319ebadece693ac91fd1c0d5678e849521b9b3b54b70462df560653f5b1cb74",
+    "variants/cpu-int4/train/training_model.onnx": "8e6de37f3cccf0ccaa02815bbeaf774fa1fd0d0bc7e514102950937186a830cb"
+  },
+  "supportedTasks": [
+    "text-generation-with-past"
+  ],
+  "trainableParameterCount": 368640,
+  "trainingParameterCount": 268835639,
+  "transformersVersion": "4.57.6",
+  "trustRemoteCode": false,
+  "variants": [
+    {
+      "abi": null,
+      "executionProvider": "cpu",
+      "features": [
+        "core",
+        "inference",
+        "train"
+      ],
+      "id": "cpu-int4",
+      "minimumAndroidApi": 28,
+      "paths": {
+        "inference": "variants/cpu-int4/inference",
+        "tokenizer": "shared/tokenizer",
+        "train": "variants/cpu-int4/train"
+      },
+      "quantization": "int4",
+      "recommendedDeviceMemoryMb": null,
+      "supportedEngines": [
+        "native"
+      ],
+      "weightHandoff": "variants/cpu-int4/inference/weight_handoff_map.json"
+    }
+  ],
+  "weightHandoff": "variants/cpu-int4/inference/weight_handoff_map.json"
+}

optimum/export_report.json

@@ -0,0 +1,35 @@
+{
+  "androidRuntime": {
+    "minimumAndroidApi": 28,
+    "recommendedDeviceMemoryMb": null,
+    "requiredAbis": []
+  },
+  "architectures": [
+    "gemma3_text"
+  ],
+  "license": {
+    "baseModelWeights": null,
+    "framework": null,
+    "noticeFile": null
+  },
+  "mobiletransformersVersion": "0.1.0",
+  "onnxRuntimeTrainingVersion": "1.23.0+cpu",
+  "optimumOnnxVersion": "0.1.0",
+  "peftMethods": [
+    "lora"
+  ],
+  "quantization": [
+    "int4"
+  ],
+  "selectedTask": "text-generation-with-past",
+  "sourceParameterCount": 268098176,
+  "supportedTasks": [
+    "text-generation-with-past"
+  ],
+  "trainInferenceLossDeltaNats": 0.8837,
+  "trainableParameterCount": 368640,
+  "trainingParameterCount": 268835639,
+  "trainingQuantizedParameterCount": 253296731,
+  "transformersVersion": "4.57.6",
+  "trustRemoteCode": false
+}

optimum/supported_tasks.json

@@ -0,0 +1,3 @@
+[
+  "text-generation-with-past"
+]

shared/tokenizer/chat_template.jinja

@@ -0,0 +1,279 @@
+{%- macro format_parameters(properties, required) -%}
+    {%- set standard_keys = ['description', 'type', 'properties', 'required', 'nullable'] -%}
+    {%- set ns = namespace(found_first=false) -%}
+    {%- for key, value in properties | dictsort -%}
+        {%- if key not in standard_keys -%}
+            {%- if ns.found_first %},{% endif -%}
+            {%- set ns.found_first = true -%}
+            {{- key }}:{description:<escape>{{ value['description'] }}<escape>
+            {%- if value['type'] | upper == 'STRING' -%}
+                {%- if value['enum'] -%}
+                    ,enum:{{ format_argument(value['enum']) }}
+                {%- endif -%}
+            {%- elif value['type'] | upper == 'OBJECT' -%}
+                ,properties:{
+                {%- if value['properties'] is defined and value['properties'] is mapping -%}
+                    {{- format_parameters(value['properties'], value['required'] | default([])) -}}
+                {%- elif value is mapping -%}
+                    {{- format_parameters(value, value['required'] | default([])) -}}
+                {%- endif -%}
+                }
+                {%- if value['required'] -%}
+                    ,required:[
+                    {%- for item in value['required'] | default([]) -%}
+                        <escape>{{- item -}}<escape>
+                        {%- if not loop.last %},{% endif -%}
+                    {%- endfor -%}
+                    ]
+                {%- endif -%}
+            {%- elif value['type'] | upper == 'ARRAY' -%}
+                {%- if value['items'] is mapping and value['items'] -%}
+                    ,items:{
+                    {%- set ns_items = namespace(found_first=false) -%}
+                    {%- for item_key, item_value in value['items'] | dictsort -%}
+                        {%- if item_value is not none -%}
+                            {%- if ns_items.found_first %},{% endif -%}
+                            {%- set ns_items.found_first = true -%}
+                            {%- if item_key == 'properties' -%}
+                                properties:{
+                                {%- if item_value is mapping -%}
+                                    {{- format_parameters(item_value, value['items']['required'] | default([])) -}}
+                                {%- endif -%}
+                                }
+                            {%- elif item_key == 'required' -%}
+                                required:[
+                                {%- for req_item in item_value -%}
+                                    <escape>{{- req_item -}}<escape>
+                                    {%- if not loop.last %},{% endif -%}
+                                {%- endfor -%}
+                                ]
+                            {%- elif item_key == 'type' -%}
+                                {%- if item_value is string -%}
+                                    type:{{ format_argument(item_value | upper) }}
+                                {%- else -%}
+                                    type:{{ format_argument(item_value | map('upper') | list) }}
+                                {%- endif -%}
+                            {%- else -%}
+                                {{ item_key }}:{{ format_argument(item_value) }}
+                            {%- endif -%}
+                        {%- endif -%}
+                    {%- endfor -%}
+                    }
+                {%- endif -%}
+            {%- endif -%}
+            ,type:<escape>{{ value['type'] | upper }}<escape>}
+        {%- endif -%}
+    {%- endfor -%}
+{%- endmacro -%}
+{% macro format_function_declaration(tool_data) -%}
+declaration:{{- tool_data['function']['name'] -}}
+{description:<escape>{{- tool_data['function']['description'] -}}<escape>
+{%- set params = tool_data['function']['parameters'] -%}
+{%- if params -%}
+    ,parameters:{
+    {%- if params['properties'] -%}
+        properties:{ {{- format_parameters(params['properties'], params['required']) -}} },
+    {%- endif -%}
+    {%- if params['required'] -%}
+        required:[
+        {%- for item in params['required'] -%}
+            <escape>{{- item -}}<escape>
+            {{- ',' if not loop.last -}}
+        {%- endfor -%}
+        ],
+    {%- endif -%}
+    {%- if params['type'] -%}
+        type:<escape>{{- params['type'] | upper -}}<escape>}
+    {%- endif -%}
+{%- endif -%}
+}
+{%- endmacro -%}
+{% macro format_argument(argument, escape_keys=True) -%}
+{%- if argument is string -%}
+    {{- '<escape>' + argument + '<escape>' -}}
+{%- elif argument is boolean -%}
+    {%- if argument -%}
+        {{- 'true' -}}
+    {%- else -%}
+        {{- 'false' -}}
+    {%- endif -%}
+{%- elif argument is mapping -%}
+    {{- '{' -}}
+    {%- set ns = namespace(found_first=false) -%}
+    {%- for key, value in argument | dictsort -%}
+        {%- if ns.found_first %},{% endif -%}
+        {%- set ns.found_first = true -%}
+        {%- if escape_keys -%}
+            {{- '<escape>' + key + '<escape>' -}}
+        {%- else -%}
+            {{- key -}}
+        {%- endif -%}
+        :{{- format_argument(value, escape_keys=escape_keys) -}}
+    {%- endfor -%}
+    {{- '}' -}}
+{%- elif argument is sequence -%}
+    {{- '[' -}}
+    {%- for item in argument -%}
+        {{- format_argument(item, escape_keys=escape_keys) -}}
+        {%- if not loop.last %},{% endif -%}
+    {%- endfor -%}
+    {{- ']' -}}
+{%- else -%}
+    {{- argument -}}
+{%- endif -%}
+{%- endmacro -%}
+{{ bos_token }}
+{%- set ns = namespace(prev_message_type=None) -%}
+{#- Tool Declarations -#}
+{%- set loop_messages = messages -%}
+{%- if tools or messages[0]['role'] == 'system' or messages[0]['role'] == 'developer' -%}
+    {{- '<start_of_turn>developer\n' -}}
+    {%- if messages[0]['role'] == 'system' or messages[0]['role'] == 'developer' -%}
+        {%- if messages[0]['content'] is string -%}
+            {{- messages[0]['content'] | trim -}}
+        {%- elif messages[0]['content'] is sequence -%}
+            {%- for item in messages[0]['content'] -%}
+                {%- if item['type'] == 'text' -%}
+                    {{- item['text'] | trim -}}
+                {%- endif -%}
+            {%- endfor -%}
+        {%- endif -%}
+        {%- set loop_messages = messages[1:] -%}
+    {%- endif -%}
+    {%- if tools -%}
+        {%- for tool in tools %}
+            {{- '<start_function_declaration>' -}}
+            {{- format_function_declaration(tool) | trim }}
+            {{- '<end_function_declaration>' -}}
+        {%- endfor %}
+    {%- endif -%}
+    {{- '<end_of_turn>\n' }}
+{%- endif %}
+{#- Loop through messages. -#}
+{%- for message in loop_messages -%}
+    {%- if (message['role'] == 'assistant') -%}
+        {#- Rename "assistant" to "model". -#}
+        {%- set role = "model" -%}
+    {%- else -%}
+        {%- set role = message['role'] -%}
+    {%- endif -%}
+    {%- if role != 'tool' -%}
+        {%- if ns.prev_message_type != 'tool_response' -%}
+            {{- '<start_of_turn>' + role + '\n' }}
+        {%- endif -%}
+        {%- set ns.prev_message_type = None -%}
+        {%- if 'content' in message and message['content'] is not none -%}
+            {%- if message['content'] is string -%}
+                {{ message['content'] | trim }}
+            {%- elif message['content'] is sequence -%}
+                {%- for item in message['content'] -%}
+                    {%- if item['type'] == 'image' -%}
+                        {{ '<start_of_image>' }}
+                    {%- elif item['type'] == 'text' -%}
+                        {{ item['text'] | trim }}
+                    {%- endif -%}
+                {%- endfor -%}
+            {%- else -%}
+                {{ raise_exception("Invalid content type in user/assistant message") }}
+            {%- endif -%}
+            {%- set ns.prev_message_type = 'content' -%}
+        {%- endif -%}
+        {%- if 'tool_calls' in message and message['tool_calls'] and message['tool_calls'] is iterable -%}
+            {#- Tool Calls -#}
+            {%- for tool_call in message['tool_calls'] -%}
+                {% set function = tool_call['function'] %}
+                {{-  '<start_function_call>call:' + function['name'] + '{' -}}
+                {%- if 'arguments' in function -%}
+                    {%- if function['arguments'] is mapping -%}
+                        {%- set ns = namespace(found_first=false) -%}
+                        {%- for key, value in function['arguments'] | dictsort -%}
+                            {%- if ns.found_first %},{% endif -%}
+                            {%- set ns.found_first = true -%}
+                            {{- key -}}:{{- format_argument(value, escape_keys=False) -}}
+                        {%- endfor -%}
+                    {%- elif function['arguments'] is string -%}
+                        {# This handles string-JSON, just in case #}
+                    {{ function['arguments'] }}
+                    {%- endif %}
+                {%- endif -%}
+                {{- '}<end_function_call>' -}}
+            {%- endfor -%}
+            {%- if loop.last -%}
+                {{ '<start_function_response>' }}
+            {%- endif -%}
+            {%- set ns.prev_message_type = 'tool_call' -%}
+        {%- endif -%}
+    {%- else -%}
+        {#- Tool Responses -#}
+        {%- if 'content' in message and message['content'] -%}
+            {%- if message['content'] is mapping -%}
+                {%- if 'name' in message['content'] and 'response' in message['content'] -%}
+                    {{ '<start_function_response>response:' + message['content']['name'] | trim + '{' }}
+                    {%- set response_ns = namespace(found_first=false) -%}
+                    {%- for key, value in message['content']['response'] | dictsort -%}
+                        {%- if response_ns.found_first %},{% endif -%}
+                        {%- set response_ns.found_first = true -%}
+                        {{- key -}}:{{- format_argument(value, escape_keys=False) -}}
+                    {%- endfor -%}
+                    {{- '}<end_function_response>' -}}
+                {%- elif 'name' in message -%}
+                    {{ '<start_function_response>response:' + message['name'] | trim + '{' }}
+                    {%- set response_ns = namespace(found_first=false) -%}
+                    {%- for key, value in message['content'] | dictsort -%}
+                        {%- if response_ns.found_first %},{% endif -%}
+                        {%- set response_ns.found_first = true -%}
+                        {{- key -}}:{{- format_argument(value, escape_keys=False) -}}
+                    {%- endfor -%}
+                    {{- '}<end_function_response>' -}}
+                {%- else -%}
+                    {{ raise_exception("Invalid tool response mapping: must contain 'name' and 'response' keys, or 'name' must be in the message.") }}
+                {%- endif -%}
+            {%- elif message['content'] is string -%}
+                {%- if 'name' in message -%}
+                     {{ '<start_function_response>response:' + message['name'] | trim + '{value:' + format_argument(message['content'], escape_keys=False) + '}<end_function_response>' }}
+                {%- else -%}
+                     {{ raise_exception("Invalid tool response: 'name' must be provided.") }}
+                {%- endif -%}
+            {%- elif message['content'] is sequence -%}
+                {%- for item in message['content'] -%}
+                    {%- if item is mapping -%}
+                        {%- if 'name' in item and 'response' in item -%}
+                            {{ '<start_function_response>response:' + item['name'] | trim + '{' }}
+                            {%- set response_ns = namespace(found_first=false) -%}
+                            {%- for key, value in item['response'] | dictsort -%}
+                                {%- if response_ns.found_first %},{% endif -%}
+                                {%- set response_ns.found_first = true -%}
+                                {{- key -}}:{{- format_argument(value, escape_keys=False) -}}
+                            {%- endfor -%}
+                            {{- '}<end_function_response>' -}}
+                        {%- elif 'name' in message -%}
+                            {{ '<start_function_response>response:' + message['name'] | trim + '{' }}
+                            {%- set response_ns = namespace(found_first=false) -%}
+                            {%- for key, value in item | dictsort -%}
+                                {%- if response_ns.found_first %},{% endif -%}
+                                {%- set response_ns.found_first = true -%}
+                                {{- key -}}:{{- format_argument(value, escape_keys=False) -}}
+                            {%- endfor -%}
+                            {{- '}<end_function_response>' -}}
+                        {%- else -%}
+                            {{ raise_exception("Invalid tool response mapping: must contain 'name' and 'response' keys, or 'name' must be in the message.") }}
+                        {%- endif -%}
+                    {%- else -%}
+                        {{ raise_exception("Invalid tool response message: multiple responses must all be mappings") }}
+                    {%- endif -%}
+                {%- endfor -%}
+            {%- else -%}
+                {{ raise_exception("Invalid content type in tool message: must be mapping, sequence of mappings, or string.") }}
+            {%- endif -%}
+        {%- endif -%}
+        {%- set ns.prev_message_type = 'tool_response' -%}
+    {%- endif -%}
+    {%- if ns.prev_message_type not in ['tool_call', 'tool_response'] -%}
+        {{ '<end_of_turn>\n' }}
+    {%- endif -%}
+{%- endfor -%}
+{%- if add_generation_prompt -%}
+    {%- if ns.prev_message_type != 'tool_response' -%}
+        {{- '<start_of_turn>model\n' -}}
+    {%- endif -%}
+{%- endif -%}

shared/tokenizer/mobiletransformers_tokenizer_config.json

@@ -0,0 +1,17 @@
+{
+    "model": {
+        "bos_token_id": null,
+        "context_length": 2048,
+        "num_attention_heads": 12,
+        "num_hidden_layers": 12,
+        "num_key_value_heads": 12,
+        "eos_token_id": [
+            1,
+            50,
+            106
+        ],
+        "pad_token_id": 1,
+        "type": "unknown",
+        "vocab_size": 262146
+    }
+}

shared/tokenizer/special_tokens_map.json

@@ -0,0 +1,34 @@
+{
+  "boi_token": "<start_of_image>",
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eoi_token": "<end_of_image>",
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "image_token": "<image_soft_token>",
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "sfr_token": "<start_function_response>",
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

shared/tokenizer/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b6b09a0b4a803ad453063ca4bb49a784540e8120004e2450e025df2b27d41fb2
+size 33384899

variants/cpu-int4/checksums.json

@@ -0,0 +1,93 @@
+{
+  "variants/cpu-int4/inference/chat_template.jinja": "db61fb01017bd82401d3ffca4f8e066cd56ff6d38d2a30c4058770c0bf7ab49b",
+  "variants/cpu-int4/inference/config.json": "88a795b85d50b1642e66193935a64c34326fa76efe107f9fceeafd31aa3e2af9",
+  "variants/cpu-int4/inference/frozen_base.onnx.data": "9c14b3237169fd6023199017e1949b02bbfe0ef1e80ac10c2d72bba670ebfa82",
+  "variants/cpu-int4/inference/genai_config.json": "acac04545d607a84f676049824d2e113f9367179741494a46279551b438c6efc",
+  "variants/cpu-int4/inference/generation_config.json": "90f3b346c07b98c000b6de297e805a83f504d6856b01c405a746d805c05444c2",
+  "variants/cpu-int4/inference/merger_lora_fpin_fpout.onnx": "953a8ad4704235c19ec06150788107921ba7f2578a5880a32cd59b473544712c",
+  "variants/cpu-int4/inference/model.layers.0.self_attn.q_proj.MatMul.weight.bin": "de5441ec8bb19c3a0038e7f3bd3cb25d9b81599d1e6977ceecdc1d894ede2fa1",
+  "variants/cpu-int4/inference/model.layers.0.self_attn.q_proj.MatMul.weight.bin.sha256": "b303f006c606265b707027dccaa1c22f9f19b7920a8d51e442cbcfde886d3b65",
+  "variants/cpu-int4/inference/model.layers.0.self_attn.v_proj.MatMul.weight.bin": "aced1f8faec6416a6cbee9227400021af9b1f4c0d25feddc54e255f1210575fb",
+  "variants/cpu-int4/inference/model.layers.0.self_attn.v_proj.MatMul.weight.bin.sha256": "e41d2d8d47eaea11a011971b193d1ea7f15222a23dc2c2d7342c983d5fad5d58",
+  "variants/cpu-int4/inference/model.layers.1.self_attn.q_proj.MatMul.weight.bin": "cc0c039a777068e664ea4916413b822ce497007dc31f22f7e7cc27f9ed0bd820",
+  "variants/cpu-int4/inference/model.layers.1.self_attn.q_proj.MatMul.weight.bin.sha256": "1e1651baa468edd1ec0a787271c362646bcf4669d06816750fb9907bda6161ed",
+  "variants/cpu-int4/inference/model.layers.1.self_attn.v_proj.MatMul.weight.bin": "e2cde258f00749809d3b5e6a3838dbccb866dfdc228c86e5375a1c3fb3be14b2",
+  "variants/cpu-int4/inference/model.layers.1.self_attn.v_proj.MatMul.weight.bin.sha256": "d670153828a24efe77701e8491f5bc182fd88c9f27b924c750b9c20ca4fbf89c",
+  "variants/cpu-int4/inference/model.layers.10.self_attn.q_proj.MatMul.weight.bin": "864537f8b063185411edf60cc77bb704f3ddd3d39b6f19f04bd0da2bf1ba5b5e",
+  "variants/cpu-int4/inference/model.layers.10.self_attn.q_proj.MatMul.weight.bin.sha256": "286bfb981d94d988671f4572b770728b9a285f400ec4edc6ad83c9ea499d2c40",
+  "variants/cpu-int4/inference/model.layers.10.self_attn.v_proj.MatMul.weight.bin": "f8aedd8998857b74a4e93e3182911873cd9ad547f8f0ed8212856e8f0beaef00",
+  "variants/cpu-int4/inference/model.layers.10.self_attn.v_proj.MatMul.weight.bin.sha256": "528f124d7e9657be8b57303ec6248483bb0c6dde4c84084575e3c9afd0b8188b",
+  "variants/cpu-int4/inference/model.layers.11.self_attn.q_proj.MatMul.weight.bin": "534363c1244562ba5cc4043f27b4da97fa3a48dc2a60772532816b5f11865b91",
+  "variants/cpu-int4/inference/model.layers.11.self_attn.q_proj.MatMul.weight.bin.sha256": "19a221e71a3a79d359f1168ea1a3f88003340820f8ecffae5f5472d71be6700c",
+  "variants/cpu-int4/inference/model.layers.11.self_attn.v_proj.MatMul.weight.bin": "9b3cc8d43d80eae0f674534447d7845b17a4c779fa74adb1de07254f65de03f0",
+  "variants/cpu-int4/inference/model.layers.11.self_attn.v_proj.MatMul.weight.bin.sha256": "643bbdcb8d0381f20eaf58b1d8d9ff7c3f6435ec71d5ee7d4a93d7b60d541a83",
+  "variants/cpu-int4/inference/model.layers.12.self_attn.q_proj.MatMul.weight.bin": "5213011869521ddedba25e37944d311b30632ec3b2c061caeb6671d4f4b9596f",
+  "variants/cpu-int4/inference/model.layers.12.self_attn.q_proj.MatMul.weight.bin.sha256": "92e8c3fb1b76481b06c2ff2dd3ae3a99ad52477b274511b4ba3944dc9f376858",
+  "variants/cpu-int4/inference/model.layers.12.self_attn.v_proj.MatMul.weight.bin": "8dfd86b40b9e613538f07c8297cf3fb54cb3ebf99fe4918c30c8c51bea4da5c0",
+  "variants/cpu-int4/inference/model.layers.12.self_attn.v_proj.MatMul.weight.bin.sha256": "ca5d5b8d2723dcb8a31ed895ebd6f72d3111de86981d1f10a57fd64d393a5ceb",
+  "variants/cpu-int4/inference/model.layers.13.self_attn.q_proj.MatMul.weight.bin": "3e58dbbd1a83c01943979052519976f15dd2e13b4502fe72edf522f86960d618",
+  "variants/cpu-int4/inference/model.layers.13.self_attn.q_proj.MatMul.weight.bin.sha256": "8a75590cded9ed4eb8518b18d4f5694fd4131d8ef225bfff9d148412716fe7c6",
+  "variants/cpu-int4/inference/model.layers.13.self_attn.v_proj.MatMul.weight.bin": "21666b649c20492830d5f37db41e2ba54d270228c20a532eea2104fa3c61396e",
+  "variants/cpu-int4/inference/model.layers.13.self_attn.v_proj.MatMul.weight.bin.sha256": "327d2d9dd53b71a1662bb5751d554ef719aff537095da4bd8ac2aaaea47de555",
+  "variants/cpu-int4/inference/model.layers.14.self_attn.q_proj.MatMul.weight.bin": "ef8d82392a5ac5cf37f65c638a965c7c949111ec783b0c947ecfc8d261e1ad9f",
+  "variants/cpu-int4/inference/model.layers.14.self_attn.q_proj.MatMul.weight.bin.sha256": "c183ea62d344b6331a35070bdac7091347be27f629a00a3569acf40f01c87dd2",
+  "variants/cpu-int4/inference/model.layers.14.self_attn.v_proj.MatMul.weight.bin": "c1c727d66921e1dceec4142a2a2c36254189234bd90b09db4125927e70b8e458",
+  "variants/cpu-int4/inference/model.layers.14.self_attn.v_proj.MatMul.weight.bin.sha256": "75c2d4a4ac3606a197b259bb309ada6a55341a4c3c3e68ab2b5d29985699cd88",
+  "variants/cpu-int4/inference/model.layers.15.self_attn.q_proj.MatMul.weight.bin": "31f7849f39f1b1fe0d7ef163aae56279a8f48eac0ec81dceae7dd7fda77dd293",
+  "variants/cpu-int4/inference/model.layers.15.self_attn.q_proj.MatMul.weight.bin.sha256": "c314704b435abe1cd3ba672eb1a408e7c6f7c808a215bfb7938d27f5c4583dbe",
+  "variants/cpu-int4/inference/model.layers.15.self_attn.v_proj.MatMul.weight.bin": "8f5878be79b76139fc9cc5724e91074c742fffd021b2b7bd6f0406775ec69365",
+  "variants/cpu-int4/inference/model.layers.15.self_attn.v_proj.MatMul.weight.bin.sha256": "17e8852305a2dd0be3788629535015c3ba347bd8a84a318c1a2740846ea656d3",
+  "variants/cpu-int4/inference/model.layers.16.self_attn.q_proj.MatMul.weight.bin": "cc87b80c44df0b66e627944acd40e87a0b63c77a86d1cfa55085dad5c5610de7",
+  "variants/cpu-int4/inference/model.layers.16.self_attn.q_proj.MatMul.weight.bin.sha256": "331eece77b3bc5bea26c224ee395a6c864e2e98667cd6bd0ef8a20c48aeaa27c",
+  "variants/cpu-int4/inference/model.layers.16.self_attn.v_proj.MatMul.weight.bin": "0c1f91d46f8f885ca1970f61ada3e69fe7e2a44c4f103beb456e84fb0f39cb13",
+  "variants/cpu-int4/inference/model.layers.16.self_attn.v_proj.MatMul.weight.bin.sha256": "c25e491df56cb7902bc53dfe2fdc5f69cc6c94d56a73a1e354920ace0cdedf8c",
+  "variants/cpu-int4/inference/model.layers.17.self_attn.q_proj.MatMul.weight.bin": "75aa1f482c3a16420a092f912a073abc981fc584038eead4fc4dad219c4a18df",
+  "variants/cpu-int4/inference/model.layers.17.self_attn.q_proj.MatMul.weight.bin.sha256": "5dccc067c68edcfb2aafd0ecfe966965f274afc3455699e6add6a2f4b3c0151f",
+  "variants/cpu-int4/inference/model.layers.17.self_attn.v_proj.MatMul.weight.bin": "c331cf9885fc9a470111e2e8e59018a2ffa235ee1441aa005789e321aee80df3",
+  "variants/cpu-int4/inference/model.layers.17.self_attn.v_proj.MatMul.weight.bin.sha256": "b3058d47101dd69928aa1b6e78574dd47e8a9eedda003972accac1fcbc50c53f",
+  "variants/cpu-int4/inference/model.layers.2.self_attn.q_proj.MatMul.weight.bin": "0aab3b51d26366fda266a9a030b64fa0097a3c9fe0f2ac8b45cd455ed6d00844",
+  "variants/cpu-int4/inference/model.layers.2.self_attn.q_proj.MatMul.weight.bin.sha256": "0c44bd5f5b21e9b5292a13d4069c6b313c9b9aa80881c3b469ab0034c0bedd1e",
+  "variants/cpu-int4/inference/model.layers.2.self_attn.v_proj.MatMul.weight.bin": "f995a03e11e45e51c9038b610940e2388dea7f89b186c9c32bf0774c7ad56676",
+  "variants/cpu-int4/inference/model.layers.2.self_attn.v_proj.MatMul.weight.bin.sha256": "b7f5aa1ce6d2e1be556f65fb6ca619e7591e63dd975534dae239e71290ac5a7b",
+  "variants/cpu-int4/inference/model.layers.3.self_attn.q_proj.MatMul.weight.bin": "5f256353f99cd60325cd88345d8095a1d8bf3ec2b2ed8b7dab08368018fd8335",
+  "variants/cpu-int4/inference/model.layers.3.self_attn.q_proj.MatMul.weight.bin.sha256": "00990e7e4ec37977e4c3024989b9dedab6f8c94a76755aa59ff21bdde5eae41e",
+  "variants/cpu-int4/inference/model.layers.3.self_attn.v_proj.MatMul.weight.bin": "7412ad9549d423c2cfbe4922299f5759b2e30e6839ef89cf99f8880ea8872ee5",
+  "variants/cpu-int4/inference/model.layers.3.self_attn.v_proj.MatMul.weight.bin.sha256": "e832cd832df1e7c9249d968196ff8da050eb1f4bb733e6ed1f3df80988037bad",
+  "variants/cpu-int4/inference/model.layers.4.self_attn.q_proj.MatMul.weight.bin": "ad0a0d5c5db67a43ea9793d34d5ef09e081772e8e5fba2b771ee95983ecda45a",
+  "variants/cpu-int4/inference/model.layers.4.self_attn.q_proj.MatMul.weight.bin.sha256": "6baba6d67fea470291b3da65d1cf9834f6e103b9a6845c8d753c958a2d873b5e",
+  "variants/cpu-int4/inference/model.layers.4.self_attn.v_proj.MatMul.weight.bin": "f59967d4272b6bd14e86c0cdc9fed9f1bcb21f67faa8a1d769935f710247f1b3",
+  "variants/cpu-int4/inference/model.layers.4.self_attn.v_proj.MatMul.weight.bin.sha256": "294e36d47d7ea2796ec74012dba0d6e672eaade5d8fed5c4fc15885d735fd35a",
+  "variants/cpu-int4/inference/model.layers.5.self_attn.q_proj.MatMul.weight.bin": "ff1c5fe943f8959a8b9d09640aec44d881f035f771416ec88ac929e5d4fe6b4b",
+  "variants/cpu-int4/inference/model.layers.5.self_attn.q_proj.MatMul.weight.bin.sha256": "aa4e396ee86ecd0b1538e96230fc842659e7ec66c927493682fa914efce3047b",
+  "variants/cpu-int4/inference/model.layers.5.self_attn.v_proj.MatMul.weight.bin": "4ee8c2d65e018f46002157c99b0fd025cfc523e03b07ca6fe26da71659f51c70",
+  "variants/cpu-int4/inference/model.layers.5.self_attn.v_proj.MatMul.weight.bin.sha256": "31352189895dd2e1d92b88721c0950070cdd2caa8c68c8cae31233825ec5b667",
+  "variants/cpu-int4/inference/model.layers.6.self_attn.q_proj.MatMul.weight.bin": "353d2ac53c60dfce33c57e501010ec0dac799d9050617cf6f9f818d37e1a2296",
+  "variants/cpu-int4/inference/model.layers.6.self_attn.q_proj.MatMul.weight.bin.sha256": "e869d030595d5ed13ac046acfc08fd9487b3a9ba9e940edbb4724eba90b84167",
+  "variants/cpu-int4/inference/model.layers.6.self_attn.v_proj.MatMul.weight.bin": "e5d43600c7a68cc5889563120b46780f107945e26c9385b5fca04c3a22960843",
+  "variants/cpu-int4/inference/model.layers.6.self_attn.v_proj.MatMul.weight.bin.sha256": "fd3c65bcc2e5652073fdf480a8be666611a1e20d26662e648e5cdd7fb4e0340c",
+  "variants/cpu-int4/inference/model.layers.7.self_attn.q_proj.MatMul.weight.bin": "e9d722fef4ecc4e3dffddd5b1527bcca70af70978eb812df6d7c4452c4c10b90",
+  "variants/cpu-int4/inference/model.layers.7.self_attn.q_proj.MatMul.weight.bin.sha256": "5d11e119c112c0f2cd8a46a10856aa4e3a467048928ddd0de0891dec95164b18",
+  "variants/cpu-int4/inference/model.layers.7.self_attn.v_proj.MatMul.weight.bin": "d50fe1d3c9c265fc38d169bd63fa727062f145719419564a124a5c06fbe59b04",
+  "variants/cpu-int4/inference/model.layers.7.self_attn.v_proj.MatMul.weight.bin.sha256": "bb5995d820468db5e2c676bf0960061c85fff3ccaeebb0c76b89f905b400f0bc",
+  "variants/cpu-int4/inference/model.layers.8.self_attn.q_proj.MatMul.weight.bin": "eee1373ed2ee35607cda09ceeef4c92b7ab331a419ffb795c821a58c563e884b",
+  "variants/cpu-int4/inference/model.layers.8.self_attn.q_proj.MatMul.weight.bin.sha256": "ebf030acd3a19f40d679c3e8917c15a97e122f4f8076eff7043758fefbef1b4c",
+  "variants/cpu-int4/inference/model.layers.8.self_attn.v_proj.MatMul.weight.bin": "3b9e52ee458990d65fd339e4e544d5f379b1600dd4852fc3084a32e96bc00a0b",
+  "variants/cpu-int4/inference/model.layers.8.self_attn.v_proj.MatMul.weight.bin.sha256": "6142f9fd4e7ee9f118757cb582ce4d16c3105cb389343ee8aef7dbf4b83b939f",
+  "variants/cpu-int4/inference/model.layers.9.self_attn.q_proj.MatMul.weight.bin": "68c5c1b401f009fd33795a986edbef642ed6bb2e8cfdee72f8ef5d9930a608b4",
+  "variants/cpu-int4/inference/model.layers.9.self_attn.q_proj.MatMul.weight.bin.sha256": "ec94b226ebc6be2d0eff5dfd48edb112728ec2d7fb4968481f811a87b31a16ae",
+  "variants/cpu-int4/inference/model.layers.9.self_attn.v_proj.MatMul.weight.bin": "b029f98f39c8aef6ec3676ae87106eab90d9521ff05ab1c7812963f8a1929555",
+  "variants/cpu-int4/inference/model.layers.9.self_attn.v_proj.MatMul.weight.bin.sha256": "9004e7234ef3e8231241955e74f0cfd7880d2ba47a4b6f4f43b3e86759ea7793",
+  "variants/cpu-int4/inference/model.onnx": "7816d03ce0be987e340c51ab64916f13a0f70a68630c3a2de82370a32a6fabfa",
+  "variants/cpu-int4/inference/model.onnx_data": "984595b520feb718e9dde76cc6dcd2bcc590bb67987c4da01b15a6ea54cade05",
+  "variants/cpu-int4/inference/optimum_config.json": "78abd079b5f45b977fad47a63f173701f6681d555c1372a800221831ce811e4f",
+  "variants/cpu-int4/inference/special_tokens_map.json": "d39df6f8ddb132a622394f05a603b166ba5f22d485483e30d06c1821ef4ac9b4",
+  "variants/cpu-int4/inference/tokenizer.json": "b6b09a0b4a803ad453063ca4bb49a784540e8120004e2450e025df2b27d41fb2",
+  "variants/cpu-int4/inference/tokenizer_config.json": "c33878647bdc69ce85d4b9aeef953815afd8405f8d85001ab0bbb89aa14d2554",
+  "variants/cpu-int4/inference/weight_handoff_map.json": "2b809307b885be7b6bd2908831ab37e5f18ea6d867b6fa8be2d284138e94581f",
+  "variants/cpu-int4/train/checkpoint": "cd9ea126423c721891377e36136904e64aaa4c04db73f82b8f2732aec482b1be",
+  "variants/cpu-int4/train/eval_model.onnx": "b66f69dca13c1f8a43a760f9f5b9a4cad1a9518d4914e8d0e42bd025e5ce6b5c",
+  "variants/cpu-int4/train/optimizer_model.onnx": "8ed3b548bce52bf25c2154b1452314b3c48d079e28c52822dd348034f59dc054",
+  "variants/cpu-int4/train/trainable_parameters.json": "05fddff407edb77426860ce8637b046bffd277bb053b33aa6600a321ac993958",
+  "variants/cpu-int4/train/training_config.json": "4319ebadece693ac91fd1c0d5678e849521b9b3b54b70462df560653f5b1cb74",
+  "variants/cpu-int4/train/training_model.onnx": "8e6de37f3cccf0ccaa02815bbeaf774fa1fd0d0bc7e514102950937186a830cb"
+}

variants/cpu-int4/inference/chat_template.jinja

@@ -0,0 +1,279 @@
+{%- macro format_parameters(properties, required) -%}
+    {%- set standard_keys = ['description', 'type', 'properties', 'required', 'nullable'] -%}
+    {%- set ns = namespace(found_first=false) -%}
+    {%- for key, value in properties | dictsort -%}
+        {%- if key not in standard_keys -%}
+            {%- if ns.found_first %},{% endif -%}
+            {%- set ns.found_first = true -%}
+            {{- key }}:{description:<escape>{{ value['description'] }}<escape>
+            {%- if value['type'] | upper == 'STRING' -%}
+                {%- if value['enum'] -%}
+                    ,enum:{{ format_argument(value['enum']) }}
+                {%- endif -%}
+            {%- elif value['type'] | upper == 'OBJECT' -%}
+                ,properties:{
+                {%- if value['properties'] is defined and value['properties'] is mapping -%}
+                    {{- format_parameters(value['properties'], value['required'] | default([])) -}}
+                {%- elif value is mapping -%}
+                    {{- format_parameters(value, value['required'] | default([])) -}}
+                {%- endif -%}
+                }
+                {%- if value['required'] -%}
+                    ,required:[
+                    {%- for item in value['required'] | default([]) -%}
+                        <escape>{{- item -}}<escape>
+                        {%- if not loop.last %},{% endif -%}
+                    {%- endfor -%}
+                    ]
+                {%- endif -%}
+            {%- elif value['type'] | upper == 'ARRAY' -%}
+                {%- if value['items'] is mapping and value['items'] -%}
+                    ,items:{
+                    {%- set ns_items = namespace(found_first=false) -%}
+                    {%- for item_key, item_value in value['items'] | dictsort -%}
+                        {%- if item_value is not none -%}
+                            {%- if ns_items.found_first %},{% endif -%}
+                            {%- set ns_items.found_first = true -%}
+                            {%- if item_key == 'properties' -%}
+                                properties:{
+                                {%- if item_value is mapping -%}
+                                    {{- format_parameters(item_value, value['items']['required'] | default([])) -}}
+                                {%- endif -%}
+                                }
+                            {%- elif item_key == 'required' -%}
+                                required:[
+                                {%- for req_item in item_value -%}
+                                    <escape>{{- req_item -}}<escape>
+                                    {%- if not loop.last %},{% endif -%}
+                                {%- endfor -%}
+                                ]
+                            {%- elif item_key == 'type' -%}
+                                {%- if item_value is string -%}
+                                    type:{{ format_argument(item_value | upper) }}
+                                {%- else -%}
+                                    type:{{ format_argument(item_value | map('upper') | list) }}
+                                {%- endif -%}
+                            {%- else -%}
+                                {{ item_key }}:{{ format_argument(item_value) }}
+                            {%- endif -%}
+                        {%- endif -%}
+                    {%- endfor -%}
+                    }
+                {%- endif -%}
+            {%- endif -%}
+            ,type:<escape>{{ value['type'] | upper }}<escape>}
+        {%- endif -%}
+    {%- endfor -%}
+{%- endmacro -%}
+{% macro format_function_declaration(tool_data) -%}
+declaration:{{- tool_data['function']['name'] -}}
+{description:<escape>{{- tool_data['function']['description'] -}}<escape>
+{%- set params = tool_data['function']['parameters'] -%}
+{%- if params -%}
+    ,parameters:{
+    {%- if params['properties'] -%}
+        properties:{ {{- format_parameters(params['properties'], params['required']) -}} },
+    {%- endif -%}
+    {%- if params['required'] -%}
+        required:[
+        {%- for item in params['required'] -%}
+            <escape>{{- item -}}<escape>
+            {{- ',' if not loop.last -}}
+        {%- endfor -%}
+        ],
+    {%- endif -%}
+    {%- if params['type'] -%}
+        type:<escape>{{- params['type'] | upper -}}<escape>}
+    {%- endif -%}
+{%- endif -%}
+}
+{%- endmacro -%}
+{% macro format_argument(argument, escape_keys=True) -%}
+{%- if argument is string -%}
+    {{- '<escape>' + argument + '<escape>' -}}
+{%- elif argument is boolean -%}
+    {%- if argument -%}
+        {{- 'true' -}}
+    {%- else -%}
+        {{- 'false' -}}
+    {%- endif -%}
+{%- elif argument is mapping -%}
+    {{- '{' -}}
+    {%- set ns = namespace(found_first=false) -%}
+    {%- for key, value in argument | dictsort -%}
+        {%- if ns.found_first %},{% endif -%}
+        {%- set ns.found_first = true -%}
+        {%- if escape_keys -%}
+            {{- '<escape>' + key + '<escape>' -}}
+        {%- else -%}
+            {{- key -}}
+        {%- endif -%}
+        :{{- format_argument(value, escape_keys=escape_keys) -}}
+    {%- endfor -%}
+    {{- '}' -}}
+{%- elif argument is sequence -%}
+    {{- '[' -}}
+    {%- for item in argument -%}
+        {{- format_argument(item, escape_keys=escape_keys) -}}
+        {%- if not loop.last %},{% endif -%}
+    {%- endfor -%}
+    {{- ']' -}}
+{%- else -%}
+    {{- argument -}}
+{%- endif -%}
+{%- endmacro -%}
+{{ bos_token }}
+{%- set ns = namespace(prev_message_type=None) -%}
+{#- Tool Declarations -#}
+{%- set loop_messages = messages -%}
+{%- if tools or messages[0]['role'] == 'system' or messages[0]['role'] == 'developer' -%}
+    {{- '<start_of_turn>developer\n' -}}
+    {%- if messages[0]['role'] == 'system' or messages[0]['role'] == 'developer' -%}
+        {%- if messages[0]['content'] is string -%}
+            {{- messages[0]['content'] | trim -}}
+        {%- elif messages[0]['content'] is sequence -%}
+            {%- for item in messages[0]['content'] -%}
+                {%- if item['type'] == 'text' -%}
+                    {{- item['text'] | trim -}}
+                {%- endif -%}
+            {%- endfor -%}
+        {%- endif -%}
+        {%- set loop_messages = messages[1:] -%}
+    {%- endif -%}
+    {%- if tools -%}
+        {%- for tool in tools %}
+            {{- '<start_function_declaration>' -}}
+            {{- format_function_declaration(tool) | trim }}
+            {{- '<end_function_declaration>' -}}
+        {%- endfor %}
+    {%- endif -%}
+    {{- '<end_of_turn>\n' }}
+{%- endif %}
+{#- Loop through messages. -#}
+{%- for message in loop_messages -%}
+    {%- if (message['role'] == 'assistant') -%}
+        {#- Rename "assistant" to "model". -#}
+        {%- set role = "model" -%}
+    {%- else -%}
+        {%- set role = message['role'] -%}
+    {%- endif -%}
+    {%- if role != 'tool' -%}
+        {%- if ns.prev_message_type != 'tool_response' -%}
+            {{- '<start_of_turn>' + role + '\n' }}
+        {%- endif -%}
+        {%- set ns.prev_message_type = None -%}
+        {%- if 'content' in message and message['content'] is not none -%}
+            {%- if message['content'] is string -%}
+                {{ message['content'] | trim }}
+            {%- elif message['content'] is sequence -%}
+                {%- for item in message['content'] -%}
+                    {%- if item['type'] == 'image' -%}
+                        {{ '<start_of_image>' }}
+                    {%- elif item['type'] == 'text' -%}
+                        {{ item['text'] | trim }}
+                    {%- endif -%}
+                {%- endfor -%}
+            {%- else -%}
+                {{ raise_exception("Invalid content type in user/assistant message") }}
+            {%- endif -%}
+            {%- set ns.prev_message_type = 'content' -%}
+        {%- endif -%}
+        {%- if 'tool_calls' in message and message['tool_calls'] and message['tool_calls'] is iterable -%}
+            {#- Tool Calls -#}
+            {%- for tool_call in message['tool_calls'] -%}
+                {% set function = tool_call['function'] %}
+                {{-  '<start_function_call>call:' + function['name'] + '{' -}}
+                {%- if 'arguments' in function -%}
+                    {%- if function['arguments'] is mapping -%}
+                        {%- set ns = namespace(found_first=false) -%}
+                        {%- for key, value in function['arguments'] | dictsort -%}
+                            {%- if ns.found_first %},{% endif -%}
+                            {%- set ns.found_first = true -%}
+                            {{- key -}}:{{- format_argument(value, escape_keys=False) -}}
+                        {%- endfor -%}
+                    {%- elif function['arguments'] is string -%}
+                        {# This handles string-JSON, just in case #}
+                    {{ function['arguments'] }}
+                    {%- endif %}
+                {%- endif -%}
+                {{- '}<end_function_call>' -}}
+            {%- endfor -%}
+            {%- if loop.last -%}
+                {{ '<start_function_response>' }}
+            {%- endif -%}
+            {%- set ns.prev_message_type = 'tool_call' -%}
+        {%- endif -%}
+    {%- else -%}
+        {#- Tool Responses -#}
+        {%- if 'content' in message and message['content'] -%}
+            {%- if message['content'] is mapping -%}
+                {%- if 'name' in message['content'] and 'response' in message['content'] -%}
+                    {{ '<start_function_response>response:' + message['content']['name'] | trim + '{' }}
+                    {%- set response_ns = namespace(found_first=false) -%}
+                    {%- for key, value in message['content']['response'] | dictsort -%}
+                        {%- if response_ns.found_first %},{% endif -%}
+                        {%- set response_ns.found_first = true -%}
+                        {{- key -}}:{{- format_argument(value, escape_keys=False) -}}
+                    {%- endfor -%}
+                    {{- '}<end_function_response>' -}}
+                {%- elif 'name' in message -%}
+                    {{ '<start_function_response>response:' + message['name'] | trim + '{' }}
+                    {%- set response_ns = namespace(found_first=false) -%}
+                    {%- for key, value in message['content'] | dictsort -%}
+                        {%- if response_ns.found_first %},{% endif -%}
+                        {%- set response_ns.found_first = true -%}
+                        {{- key -}}:{{- format_argument(value, escape_keys=False) -}}
+                    {%- endfor -%}
+                    {{- '}<end_function_response>' -}}
+                {%- else -%}
+                    {{ raise_exception("Invalid tool response mapping: must contain 'name' and 'response' keys, or 'name' must be in the message.") }}
+                {%- endif -%}
+            {%- elif message['content'] is string -%}
+                {%- if 'name' in message -%}
+                     {{ '<start_function_response>response:' + message['name'] | trim + '{value:' + format_argument(message['content'], escape_keys=False) + '}<end_function_response>' }}
+                {%- else -%}
+                     {{ raise_exception("Invalid tool response: 'name' must be provided.") }}
+                {%- endif -%}
+            {%- elif message['content'] is sequence -%}
+                {%- for item in message['content'] -%}
+                    {%- if item is mapping -%}
+                        {%- if 'name' in item and 'response' in item -%}
+                            {{ '<start_function_response>response:' + item['name'] | trim + '{' }}
+                            {%- set response_ns = namespace(found_first=false) -%}
+                            {%- for key, value in item['response'] | dictsort -%}
+                                {%- if response_ns.found_first %},{% endif -%}
+                                {%- set response_ns.found_first = true -%}
+                                {{- key -}}:{{- format_argument(value, escape_keys=False) -}}
+                            {%- endfor -%}
+                            {{- '}<end_function_response>' -}}
+                        {%- elif 'name' in message -%}
+                            {{ '<start_function_response>response:' + message['name'] | trim + '{' }}
+                            {%- set response_ns = namespace(found_first=false) -%}
+                            {%- for key, value in item | dictsort -%}
+                                {%- if response_ns.found_first %},{% endif -%}
+                                {%- set response_ns.found_first = true -%}
+                                {{- key -}}:{{- format_argument(value, escape_keys=False) -}}
+                            {%- endfor -%}
+                            {{- '}<end_function_response>' -}}
+                        {%- else -%}
+                            {{ raise_exception("Invalid tool response mapping: must contain 'name' and 'response' keys, or 'name' must be in the message.") }}
+                        {%- endif -%}
+                    {%- else -%}
+                        {{ raise_exception("Invalid tool response message: multiple responses must all be mappings") }}
+                    {%- endif -%}
+                {%- endfor -%}
+            {%- else -%}
+                {{ raise_exception("Invalid content type in tool message: must be mapping, sequence of mappings, or string.") }}
+            {%- endif -%}
+        {%- endif -%}
+        {%- set ns.prev_message_type = 'tool_response' -%}
+    {%- endif -%}
+    {%- if ns.prev_message_type not in ['tool_call', 'tool_response'] -%}
+        {{ '<end_of_turn>\n' }}
+    {%- endif -%}
+{%- endfor -%}
+{%- if add_generation_prompt -%}
+    {%- if ns.prev_message_type != 'tool_response' -%}
+        {{- '<start_of_turn>model\n' -}}
+    {%- endif -%}
+{%- endif -%}

variants/cpu-int4/inference/config.json

@@ -0,0 +1,57 @@
+{
+  "_sliding_window_pattern": 6,
+  "architectures": [
+    "Gemma3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_logit_softcapping": null,
+  "bos_token_id": 2,
+  "dtype": "float32",
+  "eos_token_id": [
+    1,
+    50
+  ],
+  "final_logit_softcapping": null,
+  "head_dim": 256,
+  "hidden_activation": "gelu_pytorch_tanh",
+  "hidden_size": 640,
+  "initializer_range": 0.02,
+  "intermediate_size": 2048,
+  "layer_types": [
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "model_type": "gemma3_text",
+  "num_attention_heads": 4,
+  "num_hidden_layers": 18,
+  "num_key_value_heads": 1,
+  "pad_token_id": 0,
+  "query_pre_attn_scalar": 256,
+  "rms_norm_eps": 1e-06,
+  "rope_local_base_freq": 10000.0,
+  "rope_scaling": null,
+  "rope_theta": 1000000.0,
+  "sliding_window": 512,
+  "transformers_version": "4.57.6",
+  "use_bidirectional_attention": false,
+  "use_cache": true,
+  "vocab_size": 262144
+}

variants/cpu-int4/inference/frozen_base.onnx.data

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9c14b3237169fd6023199017e1949b02bbfe0ef1e80ac10c2d72bba670ebfa82
+size 1684498944

variants/cpu-int4/inference/genai_config.json

@@ -0,0 +1,45 @@
+{
+  "model": {
+    "bos_token_id": 2,
+    "eos_token_id": [
+      1,
+      50
+    ],
+    "pad_token_id": 0,
+    "context_length": 32768,
+    "type": "gemma3_text",
+    "vocab_size": 262144,
+    "decoder": {
+      "session_options": {
+        "provider_options": []
+      },
+      "filename": "model.onnx",
+      "head_size": 256,
+      "hidden_size": 640,
+      "num_attention_heads": 4,
+      "num_key_value_heads": 1,
+      "num_hidden_layers": 18,
+      "inputs": {
+        "input_ids": "input_ids",
+        "attention_mask": "attention_mask",
+        "position_ids": "position_ids",
+        "past_key_names": "past_key_values.%d.key",
+        "past_value_names": "past_key_values.%d.value"
+      },
+      "outputs": {
+        "logits": "logits",
+        "present_key_names": "present.%d.key",
+        "present_value_names": "present.%d.value"
+      }
+    }
+  },
+  "search": {
+    "max_length": 32768,
+    "min_length": 0,
+    "do_sample": false,
+    "top_k": 1,
+    "top_p": 1.0,
+    "temperature": 1.0,
+    "repetition_penalty": 1.0
+  }
+}

variants/cpu-int4/inference/generation_config.json

@@ -0,0 +1,12 @@
+{
+  "cache_implementation": "hybrid",
+  "do_sample": true,
+  "eos_token_id": [
+    1,
+    50,
+    106
+  ],
+  "top_k": 64,
+  "top_p": 0.95,
+  "transformers_version": "4.57.6"
+}

variants/cpu-int4/inference/merger_lora_fpin_fpout.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:953a8ad4704235c19ec06150788107921ba7f2578a5880a32cd59b473544712c
+size 811

variants/cpu-int4/inference/model.layers.0.self_attn.q_proj.MatMul.weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:de5441ec8bb19c3a0038e7f3bd3cb25d9b81599d1e6977ceecdc1d894ede2fa1
+size 2621440

variants/cpu-int4/inference/model.layers.0.self_attn.q_proj.MatMul.weight.bin.sha256

@@ -0,0 +1 @@
+de5441ec8bb19c3a0038e7f3bd3cb25d9b81599d1e6977ceecdc1d894ede2fa1

variants/cpu-int4/inference/model.layers.0.self_attn.v_proj.MatMul.weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aced1f8faec6416a6cbee9227400021af9b1f4c0d25feddc54e255f1210575fb
+size 655360

variants/cpu-int4/inference/model.layers.0.self_attn.v_proj.MatMul.weight.bin.sha256

@@ -0,0 +1 @@
+aced1f8faec6416a6cbee9227400021af9b1f4c0d25feddc54e255f1210575fb

variants/cpu-int4/inference/model.layers.1.self_attn.q_proj.MatMul.weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cc0c039a777068e664ea4916413b822ce497007dc31f22f7e7cc27f9ed0bd820
+size 2621440

variants/cpu-int4/inference/model.layers.1.self_attn.q_proj.MatMul.weight.bin.sha256

@@ -0,0 +1 @@
+cc0c039a777068e664ea4916413b822ce497007dc31f22f7e7cc27f9ed0bd820

variants/cpu-int4/inference/model.layers.1.self_attn.v_proj.MatMul.weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e2cde258f00749809d3b5e6a3838dbccb866dfdc228c86e5375a1c3fb3be14b2
+size 655360

variants/cpu-int4/inference/model.layers.1.self_attn.v_proj.MatMul.weight.bin.sha256

@@ -0,0 +1 @@
+e2cde258f00749809d3b5e6a3838dbccb866dfdc228c86e5375a1c3fb3be14b2

variants/cpu-int4/inference/model.layers.10.self_attn.q_proj.MatMul.weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:864537f8b063185411edf60cc77bb704f3ddd3d39b6f19f04bd0da2bf1ba5b5e
+size 2621440

variants/cpu-int4/inference/model.layers.10.self_attn.q_proj.MatMul.weight.bin.sha256

@@ -0,0 +1 @@
+864537f8b063185411edf60cc77bb704f3ddd3d39b6f19f04bd0da2bf1ba5b5e

variants/cpu-int4/inference/model.layers.10.self_attn.v_proj.MatMul.weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f8aedd8998857b74a4e93e3182911873cd9ad547f8f0ed8212856e8f0beaef00
+size 655360

variants/cpu-int4/inference/model.layers.10.self_attn.v_proj.MatMul.weight.bin.sha256

@@ -0,0 +1 @@
+f8aedd8998857b74a4e93e3182911873cd9ad547f8f0ed8212856e8f0beaef00

variants/cpu-int4/inference/model.layers.11.self_attn.q_proj.MatMul.weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:534363c1244562ba5cc4043f27b4da97fa3a48dc2a60772532816b5f11865b91
+size 2621440

variants/cpu-int4/inference/model.layers.11.self_attn.q_proj.MatMul.weight.bin.sha256

@@ -0,0 +1 @@
+534363c1244562ba5cc4043f27b4da97fa3a48dc2a60772532816b5f11865b91

variants/cpu-int4/inference/model.layers.11.self_attn.v_proj.MatMul.weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9b3cc8d43d80eae0f674534447d7845b17a4c779fa74adb1de07254f65de03f0
+size 655360

variants/cpu-int4/inference/model.layers.11.self_attn.v_proj.MatMul.weight.bin.sha256

@@ -0,0 +1 @@
+9b3cc8d43d80eae0f674534447d7845b17a4c779fa74adb1de07254f65de03f0

variants/cpu-int4/inference/model.layers.12.self_attn.q_proj.MatMul.weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5213011869521ddedba25e37944d311b30632ec3b2c061caeb6671d4f4b9596f
+size 2621440

variants/cpu-int4/inference/model.layers.12.self_attn.q_proj.MatMul.weight.bin.sha256

@@ -0,0 +1 @@
+5213011869521ddedba25e37944d311b30632ec3b2c061caeb6671d4f4b9596f

variants/cpu-int4/inference/model.layers.12.self_attn.v_proj.MatMul.weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8dfd86b40b9e613538f07c8297cf3fb54cb3ebf99fe4918c30c8c51bea4da5c0
+size 655360

variants/cpu-int4/inference/model.layers.12.self_attn.v_proj.MatMul.weight.bin.sha256

@@ -0,0 +1 @@
+8dfd86b40b9e613538f07c8297cf3fb54cb3ebf99fe4918c30c8c51bea4da5c0

variants/cpu-int4/inference/model.layers.13.self_attn.q_proj.MatMul.weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3e58dbbd1a83c01943979052519976f15dd2e13b4502fe72edf522f86960d618
+size 2621440

variants/cpu-int4/inference/model.layers.13.self_attn.q_proj.MatMul.weight.bin.sha256

@@ -0,0 +1 @@
+3e58dbbd1a83c01943979052519976f15dd2e13b4502fe72edf522f86960d618

variants/cpu-int4/inference/model.layers.13.self_attn.v_proj.MatMul.weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:21666b649c20492830d5f37db41e2ba54d270228c20a532eea2104fa3c61396e
+size 655360

variants/cpu-int4/inference/model.layers.13.self_attn.v_proj.MatMul.weight.bin.sha256

@@ -0,0 +1 @@
+21666b649c20492830d5f37db41e2ba54d270228c20a532eea2104fa3c61396e

variants/cpu-int4/inference/model.layers.14.self_attn.q_proj.MatMul.weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ef8d82392a5ac5cf37f65c638a965c7c949111ec783b0c947ecfc8d261e1ad9f
+size 2621440

variants/cpu-int4/inference/model.layers.14.self_attn.q_proj.MatMul.weight.bin.sha256

@@ -0,0 +1 @@
+ef8d82392a5ac5cf37f65c638a965c7c949111ec783b0c947ecfc8d261e1ad9f

variants/cpu-int4/inference/model.layers.14.self_attn.v_proj.MatMul.weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c1c727d66921e1dceec4142a2a2c36254189234bd90b09db4125927e70b8e458
+size 655360

variants/cpu-int4/inference/model.layers.14.self_attn.v_proj.MatMul.weight.bin.sha256

@@ -0,0 +1 @@
+c1c727d66921e1dceec4142a2a2c36254189234bd90b09db4125927e70b8e458

variants/cpu-int4/inference/model.layers.15.self_attn.q_proj.MatMul.weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:31f7849f39f1b1fe0d7ef163aae56279a8f48eac0ec81dceae7dd7fda77dd293
+size 2621440

variants/cpu-int4/inference/model.layers.15.self_attn.q_proj.MatMul.weight.bin.sha256

@@ -0,0 +1 @@
+31f7849f39f1b1fe0d7ef163aae56279a8f48eac0ec81dceae7dd7fda77dd293

variants/cpu-int4/inference/model.layers.15.self_attn.v_proj.MatMul.weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8f5878be79b76139fc9cc5724e91074c742fffd021b2b7bd6f0406775ec69365
+size 655360

variants/cpu-int4/inference/model.layers.15.self_attn.v_proj.MatMul.weight.bin.sha256

@@ -0,0 +1 @@
+8f5878be79b76139fc9cc5724e91074c742fffd021b2b7bd6f0406775ec69365

variants/cpu-int4/inference/model.layers.16.self_attn.q_proj.MatMul.weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cc87b80c44df0b66e627944acd40e87a0b63c77a86d1cfa55085dad5c5610de7
+size 2621440