Upload folder using huggingface_hub

.gitattributes

@@ -33,3 +33,4 @@ 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
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,117 @@
+---
+license: apache-2.0
+base_model:
+  - google/gemma-4-E4B-it
+base_model_relation: quantized
+pipeline_tag: image-text-to-text
+tags:
+- gemma4
+- conversational
+library_name: openvino
+---
+
+# gemma-4-E4B-it-int4-ov
+
+* Model creator: [google](https://huggingface.co/google)
+* Original model: [gemma-4-E4B-it](https://huggingface.co/google/gemma-4-E4B-it)
+
+## Description
+
+This is [gemma-4-E4B-it](https://huggingface.co/google/gemma-4-E4B-it) model converted to the [OpenVINO™ IR](https://docs.openvino.ai/2026/documentation/openvino-ir-format.html) (Intermediate Representation) format with weights compressed to INT4 by [NNCF](https://github.com/openvinotoolkit/nncf).
+
+## Quantization Parameters
+
+Weight compression was performed using `nncf.compress_weights` with the following parameters:
+
+* mode: **INT4_ASYM**
+* group_size: **128**
+* ratio: **1.0**
+
+For more information on quantization, check the [OpenVINO model optimization guide](https://docs.openvino.ai/2026/openvino-workflow/model-optimization-guide/weight-compression.html).
+
+## Compatibility
+
+The provided OpenVINO™ IR model is compatible with:
+
+* OpenVINO version 2026.1.0 and higher
+* Optimum Intel 1.27.0 and higher
+
+## Running Model Inference with [OpenVINO GenAI](https://github.com/openvinotoolkit/openvino.genai)
+
+1. Install packages required for using OpenVINO GenAI:
+
+```
+pip install openvino openvino-tokenizers openvino-genai
+
+pip install huggingface_hub
+```
+
+2. Download model from HuggingFace Hub:
+
+```python
+import huggingface_hub as hf_hub
+
+model_id = "OpenVINO/gemma-4-E4B-it-int4-ov"
+model_path = "gemma-4-E4B-it-int4-ov"
+
+hf_hub.snapshot_download(model_id, local_dir=model_path)
+```
+
+3. Run model inference:
+
+```python
+import openvino_genai as ov_genai
+import requests
+from PIL import Image
+from io import BytesIO
+import numpy as np
+import openvino as ov
+
+device = "CPU"
+pipe = ov_genai.VLMPipeline(model_path, device)
+
+def load_image(image_file):
+    if isinstance(image_file, str) and (image_file.startswith("http") or image_file.startswith("https")):
+        response = requests.get(image_file)
+        image = Image.open(BytesIO(response.content)).convert("RGB")
+    else:
+        image = Image.open(image_file).convert("RGB")
+    image_data = np.array(image.getdata()).reshape(1, image.size[1], image.size[0], 3).astype(np.uint8)
+    return ov.Tensor(image_data)
+
+prompt = "What is unusual in this picture?"
+
+url = "https://github.com/openvinotoolkit/openvino_notebooks/assets/29454499/d5fbbd1a-d484-415c-88cb-9986625b7b11"
+image_tensor = load_image(url)
+
+def streamer(subword: str) -> bool:
+    print(subword, end="", flush=True)
+    return False
+
+pipe.start_chat()
+output = pipe.generate(prompt, image=image_tensor, max_new_tokens=100, streamer=streamer)
+pipe.finish_chat()
+```
+
+More GenAI usage examples can be found in OpenVINO GenAI library [docs](https://docs.openvino.ai/2026/openvino-workflow-generative/inference-with-genai.html) and [samples](https://github.com/openvinotoolkit/openvino.genai?tab=readme-ov-file#openvino-genai-samples)
+
+You can find more detailed usage examples in OpenVINO Notebooks:
+
+- [VLM](https://openvinotoolkit.github.io/openvino_notebooks/?search=Visual+Language)
+- [LLM](https://openvinotoolkit.github.io/openvino_notebooks/?search=LLM)
+
+## Limitations
+
+Check the original [model card](https://huggingface.co/google/gemma-4-E4B-it) for limitations.
+
+## Legal information
+
+The original model is distributed under [Apache License Version 2.0](https://huggingface.co/google/gemma-4-E4B-it/blob/main/LICENSE) license. More details can be found in [gemma-4-E4B-it](https://huggingface.co/google/gemma-4-E4B-it).
+
+## Disclaimer
+
+Intel is committed to respecting human rights and avoiding causing or
+contributing to adverse impacts on human rights. See
+[Intel's Global Human Rights Principles](https://www.intel.com/content/dam/www/central-libraries/us/en/documents/policy-human-rights.pdf).
+Intel's products and software are intended only to be used in applications
+that do not cause or contribute to adverse impacts on human rights.

chat_template.jinja

@@ -0,0 +1,344 @@
+{%- 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 -%}
+        {%- set add_comma = false -%}
+        {%- if key not in standard_keys -%}
+            {%- if ns.found_first %},{% endif -%}
+            {%- set ns.found_first = true -%}
+            {{ key }}:{
+            {%- if value['description'] -%}
+                description:<|"|>{{ value['description'] }}<|"|>
+                {%- set add_comma = true -%}
+            {%- endif -%}
+            {%- if value['type'] | upper == 'STRING' -%}
+                {%- if value['enum'] -%}
+                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                    enum:{{ format_argument(value['enum']) }}
+                {%- endif -%}
+            {%- elif value['type'] | upper == 'ARRAY' -%}
+                {%- if value['items'] is mapping and value['items'] -%}
+                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                    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 -%}
+                                    <|"|>{{- req_item -}}<|"|>
+                                    {%- 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 -%}
+            {%- if value['nullable'] %}
+                {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                nullable:true
+            {%- endif -%}
+            {%- if value['type'] | upper == 'OBJECT' -%}
+                {%- if value['properties'] is defined and value['properties'] is mapping -%}
+                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                    properties:{
+                    {{- format_parameters(value['properties'], value['required'] | default([])) -}}
+                    }
+                {%- elif value is mapping -%}
+                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                    properties:{
+                    {{- format_parameters(value, value['required'] | default([])) -}}
+                    }
+                {%- endif -%}
+                {%- if value['required'] -%}
+                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                    required:[
+                    {%- for item in value['required'] | default([]) -%}
+                        <|"|>{{- item -}}<|"|>
+                        {%- if not loop.last %},{% endif -%}
+                    {%- endfor -%}
+                    ]
+                {%- endif -%}
+            {%- endif -%}
+            {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+            type:<|"|>{{ value['type'] | upper }}<|"|>}
+        {%- endif -%}
+    {%- endfor -%}
+{%- endmacro -%}
+{%- macro format_function_declaration(tool_data) -%}
+    declaration:{{- tool_data['function']['name'] -}}{description:<|"|>{{- tool_data['function']['description'] -}}<|"|>
+    {%- 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'] -%}
+                <|"|>{{- item -}}<|"|>
+                {{- ',' if not loop.last -}}
+            {%- endfor -%}
+            ],
+        {%- endif -%}
+        {%- if params['type'] -%}
+            type:<|"|>{{- params['type'] | upper -}}<|"|>}
+        {%- endif -%}
+    {%- endif -%}
+    {%- if 'response' in tool_data['function'] -%}
+        {%- set response_declaration = tool_data['function']['response'] -%}
+        ,response:{
+        {%- if response_declaration['description'] -%}
+            description:<|"|>{{- response_declaration['description'] -}}<|"|>,
+        {%- endif -%}
+        {%- if response_declaration['type'] | upper == 'OBJECT' -%}
+            type:<|"|>{{- response_declaration['type'] | upper -}}<|"|>}
+        {%- endif -%}
+    {%- endif -%}
+    }
+{%- endmacro -%}
+{%- macro format_argument(argument, escape_keys=True) -%}
+    {%- if argument is string -%}
+        {{- '<|"|>' + argument + '<|"|>' -}}
+    {%- elif argument is boolean -%}
+        {{- 'true' if argument else 'false' -}}
+    {%- 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 -%}
+                {{- '<|"|>' + key + '<|"|>' -}}
+            {%- 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 -%}
+{%- macro strip_thinking(text) -%}
+    {%- set ns = namespace(result='') -%}
+    {%- for part in text.split('<channel|>') -%}
+        {%- if '<|channel>' in part -%}
+            {%- set ns.result = ns.result + part.split('<|channel>')[0] -%}
+        {%- else -%}
+            {%- set ns.result = ns.result + part -%}
+        {%- endif -%}
+    {%- endfor -%}
+    {{- ns.result | trim -}}
+{%- endmacro -%}
+
+{%- macro format_tool_response_block(tool_name, response) -%}
+    {{- '<|tool_response>' -}}
+    {%- if response is mapping -%}
+        {{- 'response:' + tool_name + '{' -}}
+        {%- for key, value in response | dictsort -%}
+            {{- key -}}:{{- format_argument(value, escape_keys=False) -}}
+            {%- if not loop.last %},{% endif -%}
+        {%- endfor -%}
+        {{- '}' -}}
+    {%- else -%}
+        {{- 'response:' + tool_name + '{value:' + format_argument(response, escape_keys=False) + '}' -}}
+    {%- endif -%}
+    {{- '<tool_response|>' -}}
+{%- endmacro -%}
+
+{%- set ns = namespace(prev_message_type=None) -%}
+{%- set loop_messages = messages -%}
+{{- bos_token -}}
+{#- Handle System/Tool Definitions Block -#}
+{%- if (enable_thinking is defined and enable_thinking) or tools or messages[0]['role'] in ['system', 'developer'] -%}
+    {{- '<|turn>system\n' -}}
+
+    {#- Inject Thinking token at the very top of the FIRST system turn -#}
+    {%- if enable_thinking is defined and enable_thinking -%}
+        {{- '<|think|>\n' -}}
+        {%- set ns.prev_message_type = 'think' -%}
+    {%- endif -%}
+
+    {%- if messages[0]['role'] in ['system', 'developer'] -%}
+        {{- messages[0]['content'] | trim -}}
+        {%- set loop_messages = messages[1:] -%}
+    {%- endif -%}
+
+    {%- if tools -%}
+        {%- for tool in tools %}
+            {{- '<|tool>' -}}
+            {{- format_function_declaration(tool) | trim -}}
+            {{- '<tool|>' -}}
+        {%- endfor %}
+        {%- set ns.prev_message_type = 'tool' -%}
+    {%- endif -%}
+
+    {{- '<turn|>\n' -}}
+{%- endif %}
+
+{#- Pre-scan: find last user message index for reasoning guard -#}
+{%- set ns_turn = namespace(last_user_idx=-1) -%}
+{%- for i in range(loop_messages | length) -%}
+    {%- if loop_messages[i]['role'] == 'user' -%}
+        {%- set ns_turn.last_user_idx = i -%}
+    {%- endif -%}
+{%- endfor -%}
+
+{#- Loop through messages -#}
+{%- for message in loop_messages -%}
+    {%- if message['role'] != 'tool' -%}
+    {%- set ns.prev_message_type = None -%}
+    {%- set role = 'model' if message['role'] == 'assistant' else message['role'] -%}
+    {#- Detect continuation: suppress duplicate <|turn>model when previous non-tool message was also assistant -#}
+    {%- set prev_nt = namespace(role=None, found=false) -%}
+    {%- if loop.index0 > 0 -%}
+        {%- for j in range(loop.index0 - 1, -1, -1) -%}
+            {%- if not prev_nt.found -%}
+                {%- if loop_messages[j]['role'] != 'tool' -%}
+                    {%- set prev_nt.role = loop_messages[j]['role'] -%}
+                    {%- set prev_nt.found = true -%}
+                {%- endif -%}
+            {%- endif -%}
+        {%- endfor -%}
+    {%- endif -%}
+    {%- set continue_same_model_turn = (role == 'model' and prev_nt.role == 'assistant') -%}
+    {%- if not continue_same_model_turn -%}
+        {{- '<|turn>' + role + '\n' }}
+    {%- endif -%}
+
+    {#- Render reasoning/reasoning_content as thinking channel -#}
+    {%- set thinking_text = message.get('reasoning') or message.get('reasoning_content') -%}
+    {%- if thinking_text and loop.index0 > ns_turn.last_user_idx and message.get('tool_calls') -%}
+        {{- '<|channel>thought\n' + thinking_text + '\n<channel|>' -}}
+    {%- endif -%}
+
+            {%- if message['tool_calls'] -%}
+                {%- for tool_call in message['tool_calls'] -%}
+                    {%- set function = tool_call['function'] -%}
+                    {{- '<|tool_call>call:' + function['name'] + '{' -}}
+                    {%- if function['arguments'] is mapping -%}
+                        {%- set ns_args = namespace(found_first=false) -%}
+                        {%- for key, value in function['arguments'] | dictsort -%}
+                            {%- if ns_args.found_first %},{% endif -%}
+                            {%- set ns_args.found_first = true -%}
+                            {{- key -}}:{{- format_argument(value, escape_keys=False) -}}
+                        {%- endfor -%}
+                    {%- elif function['arguments'] is string -%}
+                        {{- function['arguments'] -}}
+                    {%- endif -%}
+                    {{- '}<tool_call|>' -}}
+                {%- endfor -%}
+                {%- set ns.prev_message_type = 'tool_call' -%}
+            {%- endif -%}
+
+            {%- set ns_tr_out = namespace(flag=false) -%}
+            {%- if message.get('tool_responses') -%}
+                {#- Legacy: tool_responses embedded on the assistant message (Google/Gemma native) -#}
+                {%- for tool_response in message['tool_responses'] -%}
+                    {{- format_tool_response_block(tool_response['name'] | default('unknown'), tool_response['response']) -}}
+                    {%- set ns_tr_out.flag = true -%}
+                    {%- set ns.prev_message_type = 'tool_response' -%}
+                {%- endfor -%}
+            {%- elif message.get('tool_calls') -%}
+                {#- OpenAI Chat Completions: forward-scan consecutive role:tool messages -#}
+                {%- set ns_tool_scan = namespace(stopped=false) -%}
+                {%- for k in range(loop.index0 + 1, loop_messages | length) -%}
+                    {%- if ns_tool_scan.stopped -%}
+                    {%- elif loop_messages[k]['role'] != 'tool' -%}
+                        {%- set ns_tool_scan.stopped = true -%}
+                    {%- else -%}
+                        {%- set follow = loop_messages[k] -%}
+                        {#- Resolve tool_call_id to function name -#}
+                        {%- set ns_tname = namespace(name=follow.get('name') | default('unknown')) -%}
+                        {%- for tc in message['tool_calls'] -%}
+                            {%- if tc.get('id') == follow.get('tool_call_id') -%}
+                                {%- set ns_tname.name = tc['function']['name'] -%}
+                            {%- endif -%}
+                        {%- endfor -%}
+                        {#- Handle content as string or content-parts array -#}
+                        {%- set tool_body = follow.get('content') -%}
+                        {%- if tool_body is string -%}
+                            {{- format_tool_response_block(ns_tname.name, tool_body) -}}
+                        {%- elif tool_body is sequence and tool_body is not string -%}
+                            {%- set ns_txt = namespace(s='') -%}
+                            {%- for part in tool_body -%}
+                                {%- if part.get('type') == 'text' -%}
+                                    {%- set ns_txt.s = ns_txt.s + (part.get('text') | default('')) -%}
+                                {%- endif -%}
+                            {%- endfor -%}
+                            {{- format_tool_response_block(ns_tname.name, ns_txt.s) -}}
+                        {%- else -%}
+                            {{- format_tool_response_block(ns_tname.name, tool_body) -}}
+                        {%- endif -%}
+                        {%- set ns_tr_out.flag = true -%}
+                        {%- set ns.prev_message_type = 'tool_response' -%}
+                    {%- endif -%}
+                {%- endfor -%}
+            {%- endif -%}
+
+            {%- if message['content'] is string -%}
+                {%- if role == 'model' -%}
+                    {{- strip_thinking(message['content']) -}}
+                {%- else -%}
+                    {{- message['content'] | trim -}}
+                {%- endif -%}
+            {%- elif message['content'] is sequence -%}
+                {%- for item in message['content'] -%}
+                    {%- if item['type'] == 'text' -%}
+                        {%- if role == 'model' -%}
+                            {{- strip_thinking(item['text']) -}}
+                        {%- else -%}
+                            {{- item['text'] | trim -}}
+                        {%- endif -%}
+                    {%- elif item['type'] == 'image' -%}
+                        {{- '<|image|>' -}}
+                        {%- set ns.prev_message_type = 'image' -%}
+                    {%- elif item['type'] == 'audio' -%}
+                        {{- '<|audio|>' -}}
+                        {%- set ns.prev_message_type = 'audio' -%}
+                    {%- elif item['type'] == 'video' -%}
+                        {{- '<|video|>' -}}
+                        {%- set ns.prev_message_type = 'video' -%}
+                    {%- endif -%}
+                {%- endfor -%}
+            {%- endif -%}
+
+        {%- if ns.prev_message_type == 'tool_call' and not ns_tr_out.flag -%}
+            {{- '<|tool_response>' -}}
+        {%- elif not (ns_tr_out.flag and not message.get('content')) -%}
+            {{- '<turn|>\n' -}}
+        {%- endif -%}
+    {%- endif -%}
+{%- endfor -%}
+
+{%- if add_generation_prompt -%}
+    {%- if ns.prev_message_type != 'tool_response' and ns.prev_message_type != 'tool_call' -%}
+        {{- '<|turn>model\n' -}}
+    {%- endif -%}
+{%- endif -%}

config.json

@@ -0,0 +1,198 @@
+{
+  "architectures": [
+    "Gemma4ForConditionalGeneration"
+  ],
+  "audio_config": {
+    "_name_or_path": "",
+    "architectures": null,
+    "attention_chunk_size": 12,
+    "attention_context_left": 13,
+    "attention_context_right": 0,
+    "attention_invalid_logits_value": -1000000000.0,
+    "attention_logit_cap": 50.0,
+    "chunk_size_feed_forward": 0,
+    "conv_kernel_size": 5,
+    "dtype": "bfloat16",
+    "gradient_clipping": 10000000000.0,
+    "hidden_act": "silu",
+    "hidden_size": 1024,
+    "id2label": {
+      "0": "LABEL_0",
+      "1": "LABEL_1"
+    },
+    "initializer_range": 0.02,
+    "is_encoder_decoder": false,
+    "label2id": {
+      "LABEL_0": 0,
+      "LABEL_1": 1
+    },
+    "model_type": "gemma4_audio",
+    "num_attention_heads": 8,
+    "num_hidden_layers": 12,
+    "output_attentions": false,
+    "output_hidden_states": false,
+    "output_proj_dims": 1536,
+    "problem_type": null,
+    "residual_weight": 0.5,
+    "return_dict": true,
+    "rms_norm_eps": 1e-06,
+    "subsampling_conv_channels": [
+      128,
+      32
+    ],
+    "use_clipped_linears": true
+  },
+  "audio_token_id": 258881,
+  "boa_token_id": 256000,
+  "boi_token_id": 255999,
+  "dtype": "bfloat16",
+  "eoa_token_id": 258883,
+  "eoa_token_index": 258883,
+  "eoi_token_id": 258882,
+  "eos_token_id": [
+    1,
+    106
+  ],
+  "image_token_id": 258880,
+  "initializer_range": 0.02,
+  "model_type": "gemma4",
+  "text_config": {
+    "attention_bias": false,
+    "attention_dropout": 0.0,
+    "attention_k_eq_v": false,
+    "bos_token_id": 2,
+    "dtype": "bfloat16",
+    "enable_moe_block": false,
+    "eos_token_id": 1,
+    "expert_intermediate_size": null,
+    "final_logit_softcapping": 30.0,
+    "global_head_dim": 512,
+    "head_dim": 256,
+    "hidden_activation": "gelu_pytorch_tanh",
+    "hidden_size": 2560,
+    "hidden_size_per_layer_input": 256,
+    "initializer_range": 0.02,
+    "intermediate_size": 10240,
+    "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",
+      "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",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention"
+    ],
+    "max_position_embeddings": 131072,
+    "model_type": "gemma4_text",
+    "moe_intermediate_size": null,
+    "num_attention_heads": 8,
+    "num_experts": null,
+    "num_global_key_value_heads": null,
+    "num_hidden_layers": 42,
+    "num_key_value_heads": 2,
+    "num_kv_shared_layers": 18,
+    "pad_token_id": 0,
+    "rms_norm_eps": 1e-06,
+    "rope_parameters": {
+      "full_attention": {
+        "partial_rotary_factor": 0.25,
+        "rope_theta": 1000000.0,
+        "rope_type": "proportional"
+      },
+      "sliding_attention": {
+        "rope_theta": 10000.0,
+        "rope_type": "default"
+      }
+    },
+    "sliding_window": 512,
+    "tie_word_embeddings": true,
+    "top_k_experts": null,
+    "use_bidirectional_attention": null,
+    "use_cache": true,
+    "use_double_wide_mlp": false,
+    "vocab_size": 262144,
+    "vocab_size_per_layer_input": 262144
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "5.5.0",
+  "video_token_id": 258884,
+  "vision_config": {
+    "_name_or_path": "",
+    "architectures": null,
+    "attention_bias": false,
+    "attention_dropout": 0.0,
+    "chunk_size_feed_forward": 0,
+    "default_output_length": 280,
+    "dtype": "bfloat16",
+    "global_head_dim": 64,
+    "head_dim": 64,
+    "hidden_activation": "gelu_pytorch_tanh",
+    "hidden_size": 768,
+    "id2label": {
+      "0": "LABEL_0",
+      "1": "LABEL_1"
+    },
+    "initializer_range": 0.02,
+    "intermediate_size": 3072,
+    "is_encoder_decoder": false,
+    "label2id": {
+      "LABEL_0": 0,
+      "LABEL_1": 1
+    },
+    "max_position_embeddings": 131072,
+    "model_type": "gemma4_vision",
+    "num_attention_heads": 12,
+    "num_hidden_layers": 16,
+    "num_key_value_heads": 12,
+    "output_attentions": false,
+    "output_hidden_states": false,
+    "patch_size": 16,
+    "pooling_kernel_size": 3,
+    "position_embedding_size": 10240,
+    "problem_type": null,
+    "return_dict": true,
+    "rms_norm_eps": 1e-06,
+    "rope_parameters": {
+      "rope_theta": 100.0,
+      "rope_type": "default"
+    },
+    "standardize": false,
+    "use_clipped_linears": true
+  },
+  "vision_soft_tokens_per_image": 280
+}

generation_config.json

@@ -0,0 +1,14 @@
+{
+  "bos_token_id": 2,
+  "do_sample": true,
+  "eos_token_id": [
+    1,
+    106,
+    50
+  ],
+  "pad_token_id": 0,
+  "temperature": 1.0,
+  "top_k": 64,
+  "top_p": 0.95,
+  "transformers_version": "5.5.0"
+}

openvino_config.json

@@ -0,0 +1,44 @@
+{
+  "dtype": "int4",
+  "input_info": null,
+  "optimum_version": "2.1.0.dev0",
+  "output_attentions": false,
+  "quantization_config": {
+    "_dataset_kwargs": {},
+    "dataset": null,
+    "default_config": {
+      "quant_method": "default"
+    },
+    "ignored_scope": null,
+    "num_samples": null,
+    "processor": "google/gemma-4-E4B-it",
+    "quantization_configs": {
+      "lm_model": {
+        "_dataset_kwargs": {},
+        "all_layers": null,
+        "backup_precision": null,
+        "bits": 4,
+        "dataset": null,
+        "dq_group_size": null,
+        "dtype": "int4",
+        "gptq": null,
+        "group_size": 128,
+        "group_size_fallback": null,
+        "ignored_scope": null,
+        "lora_correction": null,
+        "num_samples": null,
+        "processor": "google/gemma-4-E4B-it",
+        "quant_method": "default",
+        "ratio": 1.0,
+        "scale_estimation": null,
+        "sensitivity_metric": null,
+        "statistics_path": null,
+        "sym": false,
+        "tokenizer": "google/gemma-4-E4B-it"
+      }
+    },
+    "tokenizer": "google/gemma-4-E4B-it"
+  },
+  "save_onnx_model": false,
+  "transformers_version": "5.5.0"
+}

openvino_detokenizer.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:461919bda7b546453e903052588c578e3e94e9e8215b7cec07cb00cb97459d13
+size 4413547

openvino_detokenizer.xml

@@ -0,0 +1,300 @@
+<?xml version="1.0"?>
+<net name="detokenizer" version="11">
+	<layers>
+		<layer id="0" name="Parameter_210440" type="Parameter" version="opset1">
+			<data shape="?,?" element_type="i64" />
+			<output>
+				<port id="0" precision="I64" names="Parameter_210440">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="1" name="Convert_210475" type="Convert" version="opset1">
+			<data destination_type="i32" />
+			<input>
+				<port id="0" precision="I64">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="1" precision="I32">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="2" name="Constant_210396" type="Const" version="opset1">
+			<data element_type="i32" shape="262144" offset="0" size="1048576" />
+			<output>
+				<port id="0" precision="I32">
+					<dim>262144</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="3" name="Constant_210398" type="Const" version="opset1">
+			<data element_type="i32" shape="262144" offset="1048576" size="1048576" />
+			<output>
+				<port id="0" precision="I32">
+					<dim>262144</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="4" name="Constant_210400" type="Const" version="opset1">
+			<data element_type="u8" shape="2316295" offset="2097152" size="2316295" />
+			<output>
+				<port id="0" precision="U8">
+					<dim>2316295</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="5" name="Slice_210445" type="Const" version="opset1">
+			<data element_type="i32" shape="24" offset="4413447" size="96" />
+			<output>
+				<port id="0" precision="I32">
+					<dim>24</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="6" name="VocabDecoder_210447" type="VocabDecoder" version="extension">
+			<data skip_tokens="" />
+			<input>
+				<port id="0" precision="I32">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+				<port id="1" precision="I32">
+					<dim>262144</dim>
+				</port>
+				<port id="2" precision="I32">
+					<dim>262144</dim>
+				</port>
+				<port id="3" precision="U8">
+					<dim>2316295</dim>
+				</port>
+				<port id="4" precision="I32">
+					<dim>24</dim>
+				</port>
+			</input>
+			<output>
+				<port id="5" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="6" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="7" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="8" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="9" precision="U8">
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="7" name="Constant_210449" type="Const" version="opset1">
+			<data element_type="u8" shape="3" offset="4413543" size="3" />
+			<output>
+				<port id="0" precision="U8">
+					<dim>3</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="8" name="Constant_210451" type="Const" version="opset1">
+			<data element_type="u8" shape="1" offset="4413546" size="1" />
+			<output>
+				<port id="0" precision="U8">
+					<dim>1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="9" name="RegexNormalization_210452" type="RegexNormalization" version="extension">
+			<data global_replace="true" />
+			<input>
+				<port id="0" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="1" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="2" precision="U8">
+					<dim>-1</dim>
+				</port>
+				<port id="3" precision="U8">
+					<dim>3</dim>
+				</port>
+				<port id="4" precision="U8">
+					<dim>1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="5" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="6" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="7" precision="U8">
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="10" name="ByteFallback_210453" type="ByteFallback" version="extension">
+			<input>
+				<port id="0" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="1" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="2" precision="U8">
+					<dim>-1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="3" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="4" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="5" precision="U8">
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="11" name="FuzeRagged_210454" type="FuzeRagged" version="extension">
+			<input>
+				<port id="0" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="1" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="2" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="3" precision="I32">
+					<dim>-1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="4" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="5" precision="I32">
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="12" name="UTF8Validate_210455" type="UTF8Validate" version="extension">
+			<data replace_mode="true" />
+			<input>
+				<port id="0" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="1" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="2" precision="U8">
+					<dim>-1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="3" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="4" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="5" precision="U8">
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="13" name="StringTensorPack_210456" type="StringTensorPack" version="opset15">
+			<input>
+				<port id="0" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="1" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="2" precision="U8">
+					<dim>-1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="3" precision="STRING" names="Result_210457,string_output">
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="14" name="Result_210457" type="Result" version="opset1" output_names="Result_210457,string_output">
+			<input>
+				<port id="0" precision="STRING">
+					<dim>-1</dim>
+				</port>
+			</input>
+		</layer>
+	</layers>
+	<edges>
+		<edge from-layer="0" from-port="0" to-layer="1" to-port="0" />
+		<edge from-layer="1" from-port="1" to-layer="6" to-port="0" />
+		<edge from-layer="2" from-port="0" to-layer="6" to-port="1" />
+		<edge from-layer="3" from-port="0" to-layer="6" to-port="2" />
+		<edge from-layer="4" from-port="0" to-layer="6" to-port="3" />
+		<edge from-layer="5" from-port="0" to-layer="6" to-port="4" />
+		<edge from-layer="6" from-port="7" to-layer="9" to-port="0" />
+		<edge from-layer="6" from-port="8" to-layer="9" to-port="1" />
+		<edge from-layer="6" from-port="9" to-layer="9" to-port="2" />
+		<edge from-layer="6" from-port="6" to-layer="11" to-port="1" />
+		<edge from-layer="6" from-port="5" to-layer="11" to-port="0" />
+		<edge from-layer="7" from-port="0" to-layer="9" to-port="3" />
+		<edge from-layer="8" from-port="0" to-layer="9" to-port="4" />
+		<edge from-layer="9" from-port="6" to-layer="10" to-port="1" />
+		<edge from-layer="9" from-port="7" to-layer="10" to-port="2" />
+		<edge from-layer="9" from-port="5" to-layer="10" to-port="0" />
+		<edge from-layer="10" from-port="3" to-layer="11" to-port="2" />
+		<edge from-layer="10" from-port="4" to-layer="11" to-port="3" />
+		<edge from-layer="10" from-port="5" to-layer="12" to-port="2" />
+		<edge from-layer="11" from-port="4" to-layer="12" to-port="0" />
+		<edge from-layer="11" from-port="5" to-layer="12" to-port="1" />
+		<edge from-layer="12" from-port="3" to-layer="13" to-port="0" />
+		<edge from-layer="12" from-port="4" to-layer="13" to-port="1" />
+		<edge from-layer="12" from-port="5" to-layer="13" to-port="2" />
+		<edge from-layer="13" from-port="3" to-layer="14" to-port="0" />
+	</edges>
+	<rt_info>
+		<info name="OpenVINO Runtime" value="2026.2.0-21622-fdad5b35a43" />
+		<add_attention_mask value="True" />
+		<add_prefix_space />
+		<add_special_tokens value="True" />
+		<bos_token_id value="2" />
+		<chat_template value="{%- macro format_parameters(properties, required) -%}&#10;    {%- set standard_keys = ['description', 'type', 'properties', 'required', 'nullable'] -%}&#10;    {%- set ns = namespace(found_first=false) -%}&#10;    {%- for key, value in properties | dictsort -%}&#10;        {%- set add_comma = false -%}&#10;        {%- if key not in standard_keys -%}&#10;            {%- if ns.found_first %},{% endif -%}&#10;            {%- set ns.found_first = true -%}&#10;            {{ key }}:{&#10;            {%- if value['description'] -%}&#10;                description:&lt;|&quot;|>{{ value['description'] }}&lt;|&quot;|>&#10;                {%- set add_comma = true -%}&#10;            {%- endif -%}&#10;            {%- if value['type'] | upper == 'STRING' -%}&#10;                {%- if value['enum'] -%}&#10;                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}&#10;                    enum:{{ format_argument(value['enum']) }}&#10;                {%- endif -%}&#10;            {%- elif value['type'] | upper == 'ARRAY' -%}&#10;                {%- if value['items'] is mapping and value['items'] -%}&#10;                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}&#10;                    items:{&#10;                    {%- set ns_items = namespace(found_first=false) -%}&#10;                    {%- for item_key, item_value in value['items'] | dictsort -%}&#10;                        {%- if item_value is not none -%}&#10;                            {%- if ns_items.found_first %},{% endif -%}&#10;                            {%- set ns_items.found_first = true -%}&#10;                            {%- if item_key == 'properties' -%}&#10;                                properties:{&#10;                                {%- if item_value is mapping -%}&#10;                                    {{- format_parameters(item_value, value['items']['required'] | default([])) -}}&#10;                                {%- endif -%}&#10;                                }&#10;                            {%- elif item_key == 'required' -%}&#10;                                required:[&#10;                                {%- for req_item in item_value -%}&#10;                                    &lt;|&quot;|>{{- req_item -}}&lt;|&quot;|>&#10;                                    {%- if not loop.last %},{% endif -%}&#10;                                {%- endfor -%}&#10;                                ]&#10;                            {%- elif item_key == 'type' -%}&#10;                                {%- if item_value is string -%}&#10;                                    type:{{ format_argument(item_value | upper) }}&#10;                                {%- else -%}&#10;                                    type:{{ format_argument(item_value | map('upper') | list) }}&#10;                                {%- endif -%}&#10;                            {%- else -%}&#10;                                {{ item_key }}:{{ format_argument(item_value) }}&#10;                            {%- endif -%}&#10;                        {%- endif -%}&#10;                    {%- endfor -%}&#10;                    }&#10;                {%- endif -%}&#10;            {%- endif -%}&#10;            {%- if value['nullable'] %}&#10;                {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}&#10;                nullable:true&#10;            {%- endif -%}&#10;            {%- if value['type'] | upper == 'OBJECT' -%}&#10;                {%- if value['properties'] is defined and value['properties'] is mapping -%}&#10;                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}&#10;                    properties:{&#10;                    {{- format_parameters(value['properties'], value['required'] | default([])) -}}&#10;                    }&#10;                {%- elif value is mapping -%}&#10;                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}&#10;                    properties:{&#10;                    {{- format_parameters(value, value['required'] | default([])) -}}&#10;                    }&#10;                {%- endif -%}&#10;                {%- if value['required'] -%}&#10;                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}&#10;                    required:[&#10;                    {%- for item in value['required'] | default([]) -%}&#10;                        &lt;|&quot;|>{{- item -}}&lt;|&quot;|>&#10;                        {%- if not loop.last %},{% endif -%}&#10;                    {%- endfor -%}&#10;                    ]&#10;                {%- endif -%}&#10;            {%- endif -%}&#10;            {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}&#10;            type:&lt;|&quot;|>{{ value['type'] | upper }}&lt;|&quot;|>}&#10;        {%- endif -%}&#10;    {%- endfor -%}&#10;{%- endmacro -%}&#10;{%- macro format_function_declaration(tool_data) -%}&#10;    declaration:{{- tool_data['function']['name'] -}}{description:&lt;|&quot;|>{{- tool_data['function']['description'] -}}&lt;|&quot;|>&#10;    {%- set params = tool_data['function']['parameters'] -%}&#10;    {%- if params -%}&#10;        ,parameters:{&#10;        {%- if params['properties'] -%}&#10;            properties:{ {{- format_parameters(params['properties'], params['required']) -}} },&#10;        {%- endif -%}&#10;        {%- if params['required'] -%}&#10;            required:[&#10;            {%- for item in params['required'] -%}&#10;                &lt;|&quot;|>{{- item -}}&lt;|&quot;|>&#10;                {{- ',' if not loop.last -}}&#10;            {%- endfor -%}&#10;            ],&#10;        {%- endif -%}&#10;        {%- if params['type'] -%}&#10;            type:&lt;|&quot;|>{{- params['type'] | upper -}}&lt;|&quot;|>}&#10;        {%- endif -%}&#10;    {%- endif -%}&#10;    {%- if 'response' in tool_data['function'] -%}&#10;        {%- set response_declaration = tool_data['function']['response'] -%}&#10;        ,response:{&#10;        {%- if response_declaration['description'] -%}&#10;            description:&lt;|&quot;|>{{- response_declaration['description'] -}}&lt;|&quot;|>,&#10;        {%- endif -%}&#10;        {%- if response_declaration['type'] | upper == 'OBJECT' -%}&#10;            type:&lt;|&quot;|>{{- response_declaration['type'] | upper -}}&lt;|&quot;|>}&#10;        {%- endif -%}&#10;    {%- endif -%}&#10;    }&#10;{%- endmacro -%}&#10;{%- macro format_argument(argument, escape_keys=True) -%}&#10;    {%- if argument is string -%}&#10;        {{- '&lt;|&quot;|>' + argument + '&lt;|&quot;|>' -}}&#10;    {%- elif argument is boolean -%}&#10;        {{- 'true' if argument else 'false' -}}&#10;    {%- elif argument is mapping -%}&#10;        {{- '{' -}}&#10;        {%- set ns = namespace(found_first=false) -%}&#10;        {%- for key, value in argument | dictsort -%}&#10;            {%- if ns.found_first %},{% endif -%}&#10;            {%- set ns.found_first = true -%}&#10;            {%- if escape_keys -%}&#10;                {{- '&lt;|&quot;|>' + key + '&lt;|&quot;|>' -}}&#10;            {%- else -%}&#10;                {{- key -}}&#10;            {%- endif -%}&#10;            :{{- format_argument(value, escape_keys=escape_keys) -}}&#10;        {%- endfor -%}&#10;        {{- '}' -}}&#10;    {%- elif argument is sequence -%}&#10;        {{- '[' -}}&#10;        {%- for item in argument -%}&#10;            {{- format_argument(item, escape_keys=escape_keys) -}}&#10;            {%- if not loop.last %},{% endif -%}&#10;        {%- endfor -%}&#10;        {{- ']' -}}&#10;    {%- else -%}&#10;        {{- argument -}}&#10;    {%- endif -%}&#10;{%- endmacro -%}&#10;{%- macro strip_thinking(text) -%}&#10;    {%- set ns = namespace(result='') -%}&#10;    {%- for part in text.split('&lt;channel|>') -%}&#10;        {%- if '&lt;|channel>' in part -%}&#10;            {%- set ns.result = ns.result + part.split('&lt;|channel>')[0] -%}&#10;        {%- else -%}&#10;            {%- set ns.result = ns.result + part -%}&#10;        {%- endif -%}&#10;    {%- endfor -%}&#10;    {{- ns.result | trim -}}&#10;{%- endmacro -%}&#10;&#10;{%- macro format_tool_response_block(tool_name, response) -%}&#10;    {{- '&lt;|tool_response>' -}}&#10;    {%- if response is mapping -%}&#10;        {{- 'response:' + tool_name + '{' -}}&#10;        {%- for key, value in response | dictsort -%}&#10;            {{- key -}}:{{- format_argument(value, escape_keys=False) -}}&#10;            {%- if not loop.last %},{% endif -%}&#10;        {%- endfor -%}&#10;        {{- '}' -}}&#10;    {%- else -%}&#10;        {{- 'response:' + tool_name + '{value:' + format_argument(response, escape_keys=False) + '}' -}}&#10;    {%- endif -%}&#10;    {{- '&lt;tool_response|>' -}}&#10;{%- endmacro -%}&#10;&#10;{%- set ns = namespace(prev_message_type=None) -%}&#10;{%- set loop_messages = messages -%}&#10;{{- bos_token -}}&#10;{#- Handle System/Tool Definitions Block -#}&#10;{%- if (enable_thinking is defined and enable_thinking) or tools or messages[0]['role'] in ['system', 'developer'] -%}&#10;    {{- '&lt;|turn>system\n' -}}&#10;&#10;    {#- Inject Thinking token at the very top of the FIRST system turn -#}&#10;    {%- if enable_thinking is defined and enable_thinking -%}&#10;        {{- '&lt;|think|>\n' -}}&#10;        {%- set ns.prev_message_type = 'think' -%}&#10;    {%- endif -%}&#10;&#10;    {%- if messages[0]['role'] in ['system', 'developer'] -%}&#10;        {{- messages[0]['content'] | trim -}}&#10;        {%- set loop_messages = messages[1:] -%}&#10;    {%- endif -%}&#10;&#10;    {%- if tools -%}&#10;        {%- for tool in tools %}&#10;            {{- '&lt;|tool>' -}}&#10;            {{- format_function_declaration(tool) | trim -}}&#10;            {{- '&lt;tool|>' -}}&#10;        {%- endfor %}&#10;        {%- set ns.prev_message_type = 'tool' -%}&#10;    {%- endif -%}&#10;&#10;    {{- '&lt;turn|>\n' -}}&#10;{%- endif %}&#10;&#10;{#- Pre-scan: find last user message index for reasoning guard -#}&#10;{%- set ns_turn = namespace(last_user_idx=-1) -%}&#10;{%- for i in range(loop_messages | length) -%}&#10;    {%- if loop_messages[i]['role'] == 'user' -%}&#10;        {%- set ns_turn.last_user_idx = i -%}&#10;    {%- endif -%}&#10;{%- endfor -%}&#10;&#10;{#- Loop through messages -#}&#10;{%- for message in loop_messages -%}&#10;    {%- if message['role'] != 'tool' -%}&#10;    {%- set ns.prev_message_type = None -%}&#10;    {%- set role = 'model' if message['role'] == 'assistant' else message['role'] -%}&#10;    {#- Detect continuation: suppress duplicate &lt;|turn>model when previous non-tool message was also assistant -#}&#10;    {%- set prev_nt = namespace(role=None, found=false) -%}&#10;    {%- if loop.index0 > 0 -%}&#10;        {%- for j in range(loop.index0 - 1, -1, -1) -%}&#10;            {%- if not prev_nt.found -%}&#10;                {%- if loop_messages[j]['role'] != 'tool' -%}&#10;                    {%- set prev_nt.role = loop_messages[j]['role'] -%}&#10;                    {%- set prev_nt.found = true -%}&#10;                {%- endif -%}&#10;            {%- endif -%}&#10;        {%- endfor -%}&#10;    {%- endif -%}&#10;    {%- set continue_same_model_turn = (role == 'model' and prev_nt.role == 'assistant') -%}&#10;    {%- if not continue_same_model_turn -%}&#10;        {{- '&lt;|turn>' + role + '\n' }}&#10;    {%- endif -%}&#10;&#10;    {#- Render reasoning/reasoning_content as thinking channel -#}&#10;    {%- set thinking_text = message.get('reasoning') or message.get('reasoning_content') -%}&#10;    {%- if thinking_text and loop.index0 > ns_turn.last_user_idx and message.get('tool_calls') -%}&#10;        {{- '&lt;|channel>thought\n' + thinking_text + '\n&lt;channel|>' -}}&#10;    {%- endif -%}&#10;&#10;            {%- if message['tool_calls'] -%}&#10;                {%- for tool_call in message['tool_calls'] -%}&#10;                    {%- set function = tool_call['function'] -%}&#10;                    {{- '&lt;|tool_call>call:' + function['name'] + '{' -}}&#10;                    {%- if function['arguments'] is mapping -%}&#10;                        {%- set ns_args = namespace(found_first=false) -%}&#10;                        {%- for key, value in function['arguments'] | dictsort -%}&#10;                            {%- if ns_args.found_first %},{% endif -%}&#10;                            {%- set ns_args.found_first = true -%}&#10;                            {{- key -}}:{{- format_argument(value, escape_keys=False) -}}&#10;                        {%- endfor -%}&#10;                    {%- elif function['arguments'] is string -%}&#10;                        {{- function['arguments'] -}}&#10;                    {%- endif -%}&#10;                    {{- '}&lt;tool_call|>' -}}&#10;                {%- endfor -%}&#10;                {%- set ns.prev_message_type = 'tool_call' -%}&#10;            {%- endif -%}&#10;&#10;            {%- set ns_tr_out = namespace(flag=false) -%}&#10;            {%- if message.get('tool_responses') -%}&#10;                {#- Legacy: tool_responses embedded on the assistant message (Google/Gemma native) -#}&#10;                {%- for tool_response in message['tool_responses'] -%}&#10;                    {{- format_tool_response_block(tool_response['name'] | default('unknown'), tool_response['response']) -}}&#10;                    {%- set ns_tr_out.flag = true -%}&#10;                    {%- set ns.prev_message_type = 'tool_response' -%}&#10;                {%- endfor -%}&#10;            {%- elif message.get('tool_calls') -%}&#10;                {#- OpenAI Chat Completions: forward-scan consecutive role:tool messages -#}&#10;                {%- set ns_tool_scan = namespace(stopped=false) -%}&#10;                {%- for k in range(loop.index0 + 1, loop_messages | length) -%}&#10;                    {%- if ns_tool_scan.stopped -%}&#10;                    {%- elif loop_messages[k]['role'] != 'tool' -%}&#10;                        {%- set ns_tool_scan.stopped = true -%}&#10;                    {%- else -%}&#10;                        {%- set follow = loop_messages[k] -%}&#10;                        {#- Resolve tool_call_id to function name -#}&#10;                        {%- set ns_tname = namespace(name=follow.get('name') | default('unknown')) -%}&#10;                        {%- for tc in message['tool_calls'] -%}&#10;                            {%- if tc.get('id') == follow.get('tool_call_id') -%}&#10;                                {%- set ns_tname.name = tc['function']['name'] -%}&#10;                            {%- endif -%}&#10;                        {%- endfor -%}&#10;                        {#- Handle content as string or content-parts array -#}&#10;                        {%- set tool_body = follow.get('content') -%}&#10;                        {%- if tool_body is string -%}&#10;                            {{- format_tool_response_block(ns_tname.name, tool_body) -}}&#10;                        {%- elif tool_body is sequence and tool_body is not string -%}&#10;                            {%- set ns_txt = namespace(s='') -%}&#10;                            {%- for part in tool_body -%}&#10;                                {%- if part.get('type') == 'text' -%}&#10;                                    {%- set ns_txt.s = ns_txt.s + (part.get('text') | default('')) -%}&#10;                                {%- endif -%}&#10;                            {%- endfor -%}&#10;                            {{- format_tool_response_block(ns_tname.name, ns_txt.s) -}}&#10;                        {%- else -%}&#10;                            {{- format_tool_response_block(ns_tname.name, tool_body) -}}&#10;                        {%- endif -%}&#10;                        {%- set ns_tr_out.flag = true -%}&#10;                        {%- set ns.prev_message_type = 'tool_response' -%}&#10;                    {%- endif -%}&#10;                {%- endfor -%}&#10;            {%- endif -%}&#10;&#10;            {%- if message['content'] is string -%}&#10;                {%- if role == 'model' -%}&#10;                    {{- strip_thinking(message['content']) -}}&#10;                {%- else -%}&#10;                    {{- message['content'] | trim -}}&#10;                {%- endif -%}&#10;            {%- elif message['content'] is sequence -%}&#10;                {%- for item in message['content'] -%}&#10;                    {%- if item['type'] == 'text' -%}&#10;                        {%- if role == 'model' -%}&#10;                            {{- strip_thinking(item['text']) -}}&#10;                        {%- else -%}&#10;                            {{- item['text'] | trim -}}&#10;                        {%- endif -%}&#10;                    {%- elif item['type'] == 'image' -%}&#10;                        {{- '&lt;|image|>' -}}&#10;                        {%- set ns.prev_message_type = 'image' -%}&#10;                    {%- elif item['type'] == 'audio' -%}&#10;                        {{- '&lt;|audio|>' -}}&#10;                        {%- set ns.prev_message_type = 'audio' -%}&#10;                    {%- elif item['type'] == 'video' -%}&#10;                        {{- '&lt;|video|>' -}}&#10;                        {%- set ns.prev_message_type = 'video' -%}&#10;                    {%- endif -%}&#10;                {%- endfor -%}&#10;            {%- endif -%}&#10;&#10;        {%- if ns.prev_message_type == 'tool_call' and not ns_tr_out.flag -%}&#10;            {{- '&lt;|tool_response>' -}}&#10;        {%- elif not (ns_tr_out.flag and not message.get('content')) -%}&#10;            {{- '&lt;turn|>\n' -}}&#10;        {%- endif -%}&#10;    {%- endif -%}&#10;{%- endfor -%}&#10;&#10;{%- if add_generation_prompt -%}&#10;    {%- if ns.prev_message_type != 'tool_response' and ns.prev_message_type != 'tool_call' -%}&#10;        {{- '&lt;|turn>model\n' -}}&#10;    {%- endif -%}&#10;{%- endif -%}" />
+		<clean_up_tokenization_spaces />
+		<detokenizer_input_type value="i64" />
+		<eos_token_id value="1" />
+		<handle_special_tokens_with_re />
+		<max_length />
+		<number_of_inputs value="1" />
+		<openvino_tokenizers_version value="2026.2.0.0-668-cb14978d71b" />
+		<openvino_version value="2026.2.0-21622-fdad5b35a43" />
+		<original_post_processor_template value="{&quot;type&quot;: &quot;TemplateProcessing&quot;, &quot;single&quot;: [{&quot;Sequence&quot;: {&quot;id&quot;: &quot;A&quot;, &quot;type_id&quot;: 0}}], &quot;pair&quot;: [{&quot;Sequence&quot;: {&quot;id&quot;: &quot;A&quot;, &quot;type_id&quot;: 0}}, {&quot;Sequence&quot;: {&quot;id&quot;: &quot;B&quot;, &quot;type_id&quot;: 1}}], &quot;special_tokens&quot;: {}}" />
+		<original_tokenizer_class value="&lt;class 'transformers.models.gemma.tokenization_gemma.GemmaTokenizer'>" />
+		<pad_token_id value="0" />
+		<processed_post_processor_template value="{&quot;single&quot;: {&quot;ids&quot;: [-1], &quot;type_ids&quot;: [0]}, &quot;pair&quot;: {&quot;ids&quot;: [-1, -2], &quot;type_ids&quot;: [0, 1]}}" />
+		<skip_special_tokens value="True" />
+		<streaming_detokenizer value="False" />
+		<tokenizer_output_type value="i64" />
+		<tokenizers_version value="0.22.2" />
+		<transformers_version value="5.5.0" />
+		<use_max_padding value="False" />
+		<use_sentencepiece_backend value="False" />
+		<utf8_replace_mode value="replace" />
+		<with_detokenizer value="True" />
+	</rt_info>
+</net>

openvino_language_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:59da3849b25737b90b1379e53924e8dcc6abe02778da996ecc0344f1dacd9be6
+size 2738063913

openvino_text_embeddings_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f4ba107d403f58e4190d0bfef396c9e4eda6762021e4f429c10a53bb4f87aee2
+size 671612936

openvino_text_embeddings_model.xml

@@ -0,0 +1,211 @@
+<?xml version="1.0"?>
+<net name="Model3" version="11">
+	<layers>
+		<layer id="0" name="input" type="Parameter" version="opset1">
+			<data shape="?,?" element_type="i64" />
+			<output>
+				<port id="0" precision="I64" names="input">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="1" name="self.model.model.language_model.embed_tokens.weight" type="Const" version="opset1">
+			<data element_type="i8" shape="262144, 2560" offset="0" size="671088640" />
+			<output>
+				<port id="0" precision="I8">
+					<dim>262144</dim>
+					<dim>2560</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="2" name="Convert_2054581" type="Convert" version="opset1">
+			<data destination_type="f16" />
+			<input>
+				<port id="0" precision="I8">
+					<dim>262144</dim>
+					<dim>2560</dim>
+				</port>
+			</input>
+			<output>
+				<port id="1" precision="FP16">
+					<dim>262144</dim>
+					<dim>2560</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="3" name="self.model.model.language_model.embed_tokens.weight/scale" type="Const" version="opset1">
+			<data element_type="f16" shape="262144, 1" offset="671088640" size="524288" />
+			<output>
+				<port id="0" precision="FP16">
+					<dim>262144</dim>
+					<dim>1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="4" name="self.model.model.language_model.embed_tokens.weight/fq_weights_0" type="Multiply" version="opset1">
+			<data auto_broadcast="numpy" />
+			<input>
+				<port id="0" precision="FP16">
+					<dim>262144</dim>
+					<dim>2560</dim>
+				</port>
+				<port id="1" precision="FP16">
+					<dim>262144</dim>
+					<dim>1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="2" precision="FP16">
+					<dim>262144</dim>
+					<dim>2560</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="5" name="self.model.model.language_model.embed_tokens.weight/fq_weights_0/convert" type="Convert" version="opset1">
+			<data destination_type="f32" />
+			<input>
+				<port id="0" precision="FP16">
+					<dim>262144</dim>
+					<dim>2560</dim>
+				</port>
+			</input>
+			<output>
+				<port id="1" precision="FP32">
+					<dim>262144</dim>
+					<dim>2560</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="6" name="__module.model.model.language_model.embed_tokens/aten::embedding/Convert" type="Convert" version="opset1">
+			<data destination_type="i32" />
+			<input>
+				<port id="0" precision="I64">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="1" precision="I32">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="7" name="__module.model.model.language_model.embed_tokens/aten::embedding/Constant" type="Const" version="opset1">
+			<data element_type="i32" shape="" offset="671612928" size="4" />
+			<output>
+				<port id="0" precision="I32" />
+			</output>
+		</layer>
+		<layer id="8" name="__module.model.model.language_model.embed_tokens/aten::embedding/Gather" type="Gather" version="opset8">
+			<data batch_dims="0" />
+			<input>
+				<port id="0" precision="FP32">
+					<dim>262144</dim>
+					<dim>2560</dim>
+				</port>
+				<port id="1" precision="I32">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+				<port id="2" precision="I32" />
+			</input>
+			<output>
+				<port id="3" precision="FP32" names="14">
+					<dim>-1</dim>
+					<dim>-1</dim>
+					<dim>2560</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="9" name="Constant_56637" type="Const" version="opset1">
+			<data element_type="f32" shape="1, 1, 1" offset="671612932" size="4" />
+			<output>
+				<port id="0" precision="FP32">
+					<dim>1</dim>
+					<dim>1</dim>
+					<dim>1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="10" name="__module.model.model.language_model.embed_tokens/aten::mul/Multiply" type="Multiply" version="opset1">
+			<data auto_broadcast="numpy" />
+			<input>
+				<port id="0" precision="FP32">
+					<dim>-1</dim>
+					<dim>-1</dim>
+					<dim>2560</dim>
+				</port>
+				<port id="1" precision="FP32">
+					<dim>1</dim>
+					<dim>1</dim>
+					<dim>1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="2" precision="FP32" names="inputs_embeds">
+					<dim>-1</dim>
+					<dim>-1</dim>
+					<dim>2560</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="11" name="Result_54561" type="Result" version="opset1" output_names="inputs_embeds">
+			<input>
+				<port id="0" precision="FP32">
+					<dim>-1</dim>
+					<dim>-1</dim>
+					<dim>2560</dim>
+				</port>
+			</input>
+		</layer>
+	</layers>
+	<edges>
+		<edge from-layer="0" from-port="0" to-layer="6" to-port="0" />
+		<edge from-layer="1" from-port="0" to-layer="2" to-port="0" />
+		<edge from-layer="2" from-port="1" to-layer="4" to-port="0" />
+		<edge from-layer="3" from-port="0" to-layer="4" to-port="1" />
+		<edge from-layer="4" from-port="2" to-layer="5" to-port="0" />
+		<edge from-layer="5" from-port="1" to-layer="8" to-port="0" />
+		<edge from-layer="6" from-port="1" to-layer="8" to-port="1" />
+		<edge from-layer="7" from-port="0" to-layer="8" to-port="2" />
+		<edge from-layer="8" from-port="3" to-layer="10" to-port="0" />
+		<edge from-layer="9" from-port="0" to-layer="10" to-port="1" />
+		<edge from-layer="10" from-port="2" to-layer="11" to-port="0" />
+	</edges>
+	<rt_info>
+		<info name="OpenVINO Runtime" value="2026.2.0-21622-fdad5b35a43" />
+		<Runtime_version value="2026.2.0-21622-fdad5b35a43" />
+		<conversion_parameters>
+			<framework value="pytorch" />
+			<is_python_object value="True" />
+		</conversion_parameters>
+		<nncf>
+			<friendly_names_were_updated value="True" />
+			<version value="3.1.0" />
+			<weight_compression>
+				<advanced_parameters value="{'statistics_path': None, 'lora_adapter_rank': 256, 'group_size_fallback_mode': 'error', 'min_adjusted_group_size': 32, 'awq_params': {'subset_size': 32, 'percent_to_apply': 0.002, 'alpha_min': 0.0, 'alpha_max': 1.0, 'steps': 100, 'prefer_data_aware_scaling': True}, 'scale_estimation_params': {'subset_size': 64, 'initial_steps': 5, 'scale_steps': 5, 'weight_penalty': -1.0}, 'gptq_params': {'damp_percent': 0.1, 'block_size': 128, 'subset_size': 128}, 'lora_correction_params': {'adapter_rank': 8, 'num_iterations': 3, 'apply_regularization': True, 'subset_size': 128, 'use_int8_adapters': True}, 'backend_params': {}, 'codebook': None, 'adaptive_codebook_params': {'value_type': 'f8e4m3', 'across_blocks': False, 'num_elements': 16}}" />
+				<all_layers value="False" />
+				<awq value="False" />
+				<backup_mode value="int8_asym" />
+				<compression_format value="dequantize" />
+				<gptq value="False" />
+				<group_size value="-1" />
+				<ignored_scope value="[]" />
+				<lora_correction value="False" />
+				<mode value="int8_sym" />
+				<ratio value="1.0" />
+				<scale_estimation value="False" />
+				<sensitivity_metric value="weight_quantization_error" />
+			</weight_compression>
+		</nncf>
+		<optimum>
+			<nncf_version value="3.1.0" />
+			<optimum_intel_version value="1.27.0.dev0+1fccd40" />
+			<optimum_version value="2.1.0.dev0" />
+			<pytorch_version value="2.11.0+cpu" />
+			<transformers_version value="5.5.0" />
+		</optimum>
+	</rt_info>
+</net>

openvino_text_embeddings_per_layer_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e666691fa271c8b6e26aa6eff33f3030148c0bfe43d3a1514b9b997a9a594c5e
+size 2819096656

openvino_text_embeddings_per_layer_model.xml

@@ -0,0 +1,595 @@
+<?xml version="1.0"?>
+<net name="Model9" version="11">
+	<layers>
+		<layer id="0" name="input_ids" type="Parameter" version="opset1">
+			<data shape="?,?" element_type="i64" />
+			<output>
+				<port id="0" precision="I64" names="input_ids">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="1" name="self.language_model.embed_tokens_per_layer.weight" type="Const" version="opset1">
+			<data element_type="i8" shape="262144, 10752" offset="0" size="2818572288" />
+			<output>
+				<port id="0" precision="I8">
+					<dim>262144</dim>
+					<dim>10752</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="2" name="Convert_2662272" type="Convert" version="opset1">
+			<data destination_type="f16" />
+			<input>
+				<port id="0" precision="I8">
+					<dim>262144</dim>
+					<dim>10752</dim>
+				</port>
+			</input>
+			<output>
+				<port id="1" precision="FP16">
+					<dim>262144</dim>
+					<dim>10752</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="3" name="self.language_model.embed_tokens_per_layer.weight/scale" type="Const" version="opset1">
+			<data element_type="f16" shape="262144, 1" offset="2818572288" size="524288" />
+			<output>
+				<port id="0" precision="FP16">
+					<dim>262144</dim>
+					<dim>1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="4" name="self.language_model.embed_tokens_per_layer.weight/fq_weights_0" type="Multiply" version="opset1">
+			<data auto_broadcast="numpy" />
+			<input>
+				<port id="0" precision="FP16">
+					<dim>262144</dim>
+					<dim>10752</dim>
+				</port>
+				<port id="1" precision="FP16">
+					<dim>262144</dim>
+					<dim>1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="2" precision="FP16">
+					<dim>262144</dim>
+					<dim>10752</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="5" name="self.language_model.embed_tokens_per_layer.weight/fq_weights_0/convert" type="Convert" version="opset1">
+			<data destination_type="f32" />
+			<input>
+				<port id="0" precision="FP16">
+					<dim>262144</dim>
+					<dim>10752</dim>
+				</port>
+			</input>
+			<output>
+				<port id="1" precision="FP32">
+					<dim>262144</dim>
+					<dim>10752</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="6" name="Constant_209976" type="Const" version="opset1">
+			<data element_type="i64" shape="1, 1" offset="2819096576" size="8" />
+			<output>
+				<port id="0" precision="I64">
+					<dim>1</dim>
+					<dim>1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="7" name="aten::eq/Equal" type="Equal" version="opset1">
+			<data auto_broadcast="numpy" />
+			<input>
+				<port id="0" precision="I64">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+				<port id="1" precision="I64">
+					<dim>1</dim>
+					<dim>1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="2" precision="BOOL" names="7">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="8" name="8" type="Const" version="opset1">
+			<data element_type="i64" shape="" offset="2819096584" size="8" />
+			<output>
+				<port id="0" precision="I64" names="8" />
+			</output>
+		</layer>
+		<layer id="9" name="aten::full_like/ShapeOf" type="ShapeOf" version="opset3">
+			<data output_type="i32" />
+			<input>
+				<port id="0" precision="I64">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="1" precision="I32">
+					<dim>2</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="10" name="aten::full_like/Broadcast" type="Broadcast" version="opset3">
+			<data mode="numpy" />
+			<input>
+				<port id="0" precision="I64" />
+				<port id="1" precision="I32">
+					<dim>2</dim>
+				</port>
+			</input>
+			<output>
+				<port id="2" precision="I64" names="14,24,34">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="11" name="aten::where/Select" type="Select" version="opset1">
+			<data auto_broadcast="numpy" />
+			<input>
+				<port id="0" precision="BOOL">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+				<port id="1" precision="I64">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+				<port id="2" precision="I64">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="3" precision="I64" names="15,per_layer_inputs_tokens.3">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="12" name="Constant_209977" type="Const" version="opset1">
+			<data element_type="i64" shape="1, 1" offset="2819096592" size="8" />
+			<output>
+				<port id="0" precision="I64">
+					<dim>1</dim>
+					<dim>1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="13" name="aten::eq/Equal_1" type="Equal" version="opset1">
+			<data auto_broadcast="numpy" />
+			<input>
+				<port id="0" precision="I64">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+				<port id="1" precision="I64">
+					<dim>1</dim>
+					<dim>1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="2" precision="BOOL" names="17">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="14" name="aten::where/Select_1" type="Select" version="opset1">
+			<data auto_broadcast="numpy" />
+			<input>
+				<port id="0" precision="BOOL">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+				<port id="1" precision="I64">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+				<port id="2" precision="I64">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="3" precision="I64" names="25,per_layer_inputs_tokens.5">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="15" name="Constant_209978" type="Const" version="opset1">
+			<data element_type="i64" shape="1, 1" offset="2819096600" size="8" />
+			<output>
+				<port id="0" precision="I64">
+					<dim>1</dim>
+					<dim>1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="16" name="aten::eq/Equal_2" type="Equal" version="opset1">
+			<data auto_broadcast="numpy" />
+			<input>
+				<port id="0" precision="I64">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+				<port id="1" precision="I64">
+					<dim>1</dim>
+					<dim>1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="2" precision="BOOL" names="27">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="17" name="aten::where/Select_2" type="Select" version="opset1">
+			<data auto_broadcast="numpy" />
+			<input>
+				<port id="0" precision="BOOL">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+				<port id="1" precision="I64">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+				<port id="2" precision="I64">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="3" precision="I64" names="35,per_layer_inputs_tokens">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="18" name="Constant_209979" type="Const" version="opset1">
+			<data element_type="i64" shape="1, 1" offset="2819096584" size="8" />
+			<output>
+				<port id="0" precision="I64">
+					<dim>1</dim>
+					<dim>1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="19" name="aten::ge/GreaterEqual" type="GreaterEqual" version="opset1">
+			<data auto_broadcast="numpy" />
+			<input>
+				<port id="0" precision="I64">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+				<port id="1" precision="I64">
+					<dim>1</dim>
+					<dim>1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="2" precision="BOOL" names="37">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="20" name="Constant_209980" type="Const" version="opset1">
+			<data element_type="i64" shape="1, 1" offset="2819096608" size="8" />
+			<output>
+				<port id="0" precision="I64">
+					<dim>1</dim>
+					<dim>1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="21" name="aten::lt/Less" type="Less" version="opset1">
+			<data auto_broadcast="numpy" />
+			<input>
+				<port id="0" precision="I64">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+				<port id="1" precision="I64">
+					<dim>1</dim>
+					<dim>1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="2" precision="BOOL" names="39">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="22" name="aten::logical_and/LogicalAnd" type="LogicalAnd" version="opset1">
+			<data auto_broadcast="numpy" />
+			<input>
+				<port id="0" precision="BOOL">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+				<port id="1" precision="BOOL">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="2" precision="BOOL" names="40,per_layer_inputs_mask">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="23" name="aten::zeros_like/Constant" type="Const" version="opset1">
+			<data element_type="f32" shape="" offset="2819096616" size="4" />
+			<output>
+				<port id="0" precision="FP32" />
+			</output>
+		</layer>
+		<layer id="24" name="aten::zeros_like/Broadcast" type="Broadcast" version="opset3">
+			<data mode="numpy" />
+			<input>
+				<port id="0" precision="FP32" />
+				<port id="1" precision="I32">
+					<dim>2</dim>
+				</port>
+			</input>
+			<output>
+				<port id="2" precision="FP32">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="25" name="aten::zeros_like/ConvertLike" type="Convert" version="opset1">
+			<data destination_type="i64" />
+			<input>
+				<port id="0" precision="FP32">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="1" precision="I64" names="46">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="26" name="aten::where/Select_3" type="Select" version="opset1">
+			<data auto_broadcast="numpy" />
+			<input>
+				<port id="0" precision="BOOL">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+				<port id="1" precision="I64">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+				<port id="2" precision="I64">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="3" precision="I64" names="47,input">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="27" name="__module.language_model.embed_tokens_per_layer/aten::embedding/Convert" type="Convert" version="opset1">
+			<data destination_type="i32" />
+			<input>
+				<port id="0" precision="I64">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="1" precision="I32">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="28" name="__module.language_model.embed_tokens_per_layer/aten::embedding/Constant" type="Const" version="opset1">
+			<data element_type="i32" shape="" offset="2819096616" size="4" />
+			<output>
+				<port id="0" precision="I32" />
+			</output>
+		</layer>
+		<layer id="29" name="__module.language_model.embed_tokens_per_layer/aten::embedding/Gather" type="Gather" version="opset8">
+			<data batch_dims="0" />
+			<input>
+				<port id="0" precision="FP32">
+					<dim>262144</dim>
+					<dim>10752</dim>
+				</port>
+				<port id="1" precision="I32">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+				<port id="2" precision="I32" />
+			</input>
+			<output>
+				<port id="3" precision="FP32" names="68">
+					<dim>-1</dim>
+					<dim>-1</dim>
+					<dim>10752</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="30" name="Constant_209981" type="Const" version="opset1">
+			<data element_type="f32" shape="1, 1, 1" offset="2819096620" size="4" />
+			<output>
+				<port id="0" precision="FP32">
+					<dim>1</dim>
+					<dim>1</dim>
+					<dim>1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="31" name="__module.language_model.embed_tokens_per_layer/aten::mul/Multiply" type="Multiply" version="opset1">
+			<data auto_broadcast="numpy" />
+			<input>
+				<port id="0" precision="FP32">
+					<dim>-1</dim>
+					<dim>-1</dim>
+					<dim>10752</dim>
+				</port>
+				<port id="1" precision="FP32">
+					<dim>1</dim>
+					<dim>1</dim>
+					<dim>1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="2" precision="FP32" names="70">
+					<dim>-1</dim>
+					<dim>-1</dim>
+					<dim>10752</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="32" name="Constant_210013" type="Const" version="opset1">
+			<data element_type="i64" shape="4" offset="2819096624" size="32" />
+			<output>
+				<port id="0" precision="I64">
+					<dim>4</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="33" name="aten::reshape/Reshape" type="Reshape" version="opset1">
+			<data special_zero="true" />
+			<input>
+				<port id="0" precision="FP32">
+					<dim>-1</dim>
+					<dim>-1</dim>
+					<dim>10752</dim>
+				</port>
+				<port id="1" precision="I64">
+					<dim>4</dim>
+				</port>
+			</input>
+			<output>
+				<port id="2" precision="FP32" names="text_embeds_per_layer">
+					<dim>-1</dim>
+					<dim>-1</dim>
+					<dim>42</dim>
+					<dim>256</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="34" name="Result_207699" type="Result" version="opset1" output_names="text_embeds_per_layer">
+			<input>
+				<port id="0" precision="FP32">
+					<dim>-1</dim>
+					<dim>-1</dim>
+					<dim>42</dim>
+					<dim>256</dim>
+				</port>
+			</input>
+		</layer>
+	</layers>
+	<edges>
+		<edge from-layer="0" from-port="0" to-layer="7" to-port="0" />
+		<edge from-layer="0" from-port="0" to-layer="9" to-port="0" />
+		<edge from-layer="0" from-port="0" to-layer="11" to-port="2" />
+		<edge from-layer="1" from-port="0" to-layer="2" to-port="0" />
+		<edge from-layer="2" from-port="1" to-layer="4" to-port="0" />
+		<edge from-layer="3" from-port="0" to-layer="4" to-port="1" />
+		<edge from-layer="4" from-port="2" to-layer="5" to-port="0" />
+		<edge from-layer="5" from-port="1" to-layer="29" to-port="0" />
+		<edge from-layer="6" from-port="0" to-layer="7" to-port="1" />
+		<edge from-layer="7" from-port="2" to-layer="11" to-port="0" />
+		<edge from-layer="8" from-port="0" to-layer="10" to-port="0" />
+		<edge from-layer="9" from-port="1" to-layer="10" to-port="1" />
+		<edge from-layer="9" from-port="1" to-layer="24" to-port="1" />
+		<edge from-layer="10" from-port="2" to-layer="17" to-port="1" />
+		<edge from-layer="10" from-port="2" to-layer="14" to-port="1" />
+		<edge from-layer="10" from-port="2" to-layer="11" to-port="1" />
+		<edge from-layer="11" from-port="3" to-layer="14" to-port="2" />
+		<edge from-layer="11" from-port="3" to-layer="13" to-port="0" />
+		<edge from-layer="12" from-port="0" to-layer="13" to-port="1" />
+		<edge from-layer="13" from-port="2" to-layer="14" to-port="0" />
+		<edge from-layer="14" from-port="3" to-layer="17" to-port="2" />
+		<edge from-layer="14" from-port="3" to-layer="16" to-port="0" />
+		<edge from-layer="15" from-port="0" to-layer="16" to-port="1" />
+		<edge from-layer="16" from-port="2" to-layer="17" to-port="0" />
+		<edge from-layer="17" from-port="3" to-layer="19" to-port="0" />
+		<edge from-layer="17" from-port="3" to-layer="26" to-port="1" />
+		<edge from-layer="17" from-port="3" to-layer="21" to-port="0" />
+		<edge from-layer="18" from-port="0" to-layer="19" to-port="1" />
+		<edge from-layer="19" from-port="2" to-layer="22" to-port="0" />
+		<edge from-layer="20" from-port="0" to-layer="21" to-port="1" />
+		<edge from-layer="21" from-port="2" to-layer="22" to-port="1" />
+		<edge from-layer="22" from-port="2" to-layer="26" to-port="0" />
+		<edge from-layer="23" from-port="0" to-layer="24" to-port="0" />
+		<edge from-layer="24" from-port="2" to-layer="25" to-port="0" />
+		<edge from-layer="25" from-port="1" to-layer="26" to-port="2" />
+		<edge from-layer="26" from-port="3" to-layer="27" to-port="0" />
+		<edge from-layer="27" from-port="1" to-layer="29" to-port="1" />
+		<edge from-layer="28" from-port="0" to-layer="29" to-port="2" />
+		<edge from-layer="29" from-port="3" to-layer="31" to-port="0" />
+		<edge from-layer="30" from-port="0" to-layer="31" to-port="1" />
+		<edge from-layer="31" from-port="2" to-layer="33" to-port="0" />
+		<edge from-layer="32" from-port="0" to-layer="33" to-port="1" />
+		<edge from-layer="33" from-port="2" to-layer="34" to-port="0" />
+	</edges>
+	<rt_info>
+		<info name="OpenVINO Runtime" value="2026.2.0-21622-fdad5b35a43" />
+		<Runtime_version value="2026.2.0-21622-fdad5b35a43" />
+		<conversion_parameters>
+			<framework value="pytorch" />
+			<is_python_object value="True" />
+		</conversion_parameters>
+		<nncf>
+			<friendly_names_were_updated value="True" />
+			<version value="3.1.0" />
+			<weight_compression>
+				<advanced_parameters value="{'statistics_path': None, 'lora_adapter_rank': 256, 'group_size_fallback_mode': 'error', 'min_adjusted_group_size': 32, 'awq_params': {'subset_size': 32, 'percent_to_apply': 0.002, 'alpha_min': 0.0, 'alpha_max': 1.0, 'steps': 100, 'prefer_data_aware_scaling': True}, 'scale_estimation_params': {'subset_size': 64, 'initial_steps': 5, 'scale_steps': 5, 'weight_penalty': -1.0}, 'gptq_params': {'damp_percent': 0.1, 'block_size': 128, 'subset_size': 128}, 'lora_correction_params': {'adapter_rank': 8, 'num_iterations': 3, 'apply_regularization': True, 'subset_size': 128, 'use_int8_adapters': True}, 'backend_params': {}, 'codebook': None, 'adaptive_codebook_params': {'value_type': 'f8e4m3', 'across_blocks': False, 'num_elements': 16}}" />
+				<all_layers value="False" />
+				<awq value="False" />
+				<backup_mode value="int8_asym" />
+				<compression_format value="dequantize" />
+				<gptq value="False" />
+				<group_size value="-1" />
+				<ignored_scope value="[]" />
+				<lora_correction value="False" />
+				<mode value="int8_sym" />
+				<ratio value="1.0" />
+				<scale_estimation value="False" />
+				<sensitivity_metric value="weight_quantization_error" />
+			</weight_compression>
+		</nncf>
+		<optimum>
+			<nncf_version value="3.1.0" />
+			<optimum_intel_version value="1.27.0.dev0+1fccd40" />
+			<optimum_version value="2.1.0.dev0" />
+			<pytorch_version value="2.11.0+cpu" />
+			<transformers_version value="5.5.0" />
+		</optimum>
+	</rt_info>
+</net>

openvino_tokenizer.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d151f48d29ba9d22fdf0f071cd16cb68e5f8fba51a5489c3d2df8e29030e4a13
+size 17270328

openvino_tokenizer.xml

@@ -0,0 +1,796 @@
+<?xml version="1.0"?>
+<net name="tokenizer" version="11">
+	<layers>
+		<layer id="0" name="Parameter_210308" type="Parameter" version="opset1">
+			<data shape="?" element_type="string" />
+			<output>
+				<port id="0" precision="STRING" names="Parameter_210308">
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="1" name="Constant_210314" type="Const" version="opset1">
+			<data element_type="i64" shape="" offset="0" size="8" />
+			<output>
+				<port id="0" precision="I64" />
+			</output>
+		</layer>
+		<layer id="2" name="StringTensorUnpack_210309" type="StringTensorUnpack" version="opset15">
+			<input>
+				<port id="0" precision="STRING">
+					<dim>-1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="1" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="2" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="3" precision="U8">
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="3" name="ShapeOf_210310" type="ShapeOf" version="opset3">
+			<data output_type="i64" />
+			<input>
+				<port id="0" precision="I32">
+					<dim>-1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="1" precision="I64">
+					<dim>1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="4" name="Constant_210311" type="Const" version="opset1">
+			<data element_type="i64" shape="" offset="0" size="8" />
+			<output>
+				<port id="0" precision="I64" />
+			</output>
+		</layer>
+		<layer id="5" name="Constant_210312" type="Const" version="opset1">
+			<data element_type="i64" shape="" offset="0" size="8" />
+			<output>
+				<port id="0" precision="I64" />
+			</output>
+		</layer>
+		<layer id="6" name="Gather_210313" type="Gather" version="opset8">
+			<data batch_dims="0" />
+			<input>
+				<port id="0" precision="I64">
+					<dim>1</dim>
+				</port>
+				<port id="1" precision="I64" />
+				<port id="2" precision="I64" />
+			</input>
+			<output>
+				<port id="3" precision="I64" />
+			</output>
+		</layer>
+		<layer id="7" name="Constant_210315" type="Const" version="opset1">
+			<data element_type="i64" shape="" offset="8" size="8" />
+			<output>
+				<port id="0" precision="I64" />
+			</output>
+		</layer>
+		<layer id="8" name="Range_210316" type="Range" version="opset4">
+			<data output_type="i32" />
+			<input>
+				<port id="0" precision="I64" />
+				<port id="1" precision="I64" />
+				<port id="2" precision="I64" />
+			</input>
+			<output>
+				<port id="3" precision="I32">
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="9" name="Constant_210317" type="Const" version="opset1">
+			<data element_type="i64" shape="" offset="8" size="8" />
+			<output>
+				<port id="0" precision="I64" />
+			</output>
+		</layer>
+		<layer id="10" name="Constant_210318" type="Const" version="opset1">
+			<data element_type="i64" shape="" offset="8" size="8" />
+			<output>
+				<port id="0" precision="I64" />
+			</output>
+		</layer>
+		<layer id="11" name="Add_210319" type="Add" version="opset1">
+			<data auto_broadcast="numpy" />
+			<input>
+				<port id="0" precision="I64" />
+				<port id="1" precision="I64" />
+			</input>
+			<output>
+				<port id="2" precision="I64" />
+			</output>
+		</layer>
+		<layer id="12" name="Constant_210320" type="Const" version="opset1">
+			<data element_type="i64" shape="" offset="8" size="8" />
+			<output>
+				<port id="0" precision="I64" />
+			</output>
+		</layer>
+		<layer id="13" name="Range_210321" type="Range" version="opset4">
+			<data output_type="i32" />
+			<input>
+				<port id="0" precision="I64" />
+				<port id="1" precision="I64" />
+				<port id="2" precision="I64" />
+			</input>
+			<output>
+				<port id="3" precision="I32">
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="14" name="Constant_210385" type="Const" version="opset1">
+			<data element_type="u8" shape="301" offset="16" size="301" />
+			<output>
+				<port id="0" precision="U8">
+					<dim>301</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="15" name="SpecialTokensSplit_210386" type="SpecialTokensSplit" version="extension">
+			<input>
+				<port id="0" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="1" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="2" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="3" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="4" precision="U8">
+					<dim>-1</dim>
+				</port>
+				<port id="5" precision="U8">
+					<dim>301</dim>
+				</port>
+			</input>
+			<output>
+				<port id="6" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="7" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="8" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="9" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="10" precision="U8">
+					<dim>-1</dim>
+				</port>
+				<port id="11" precision="BOOL">
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="16" name="Constant_210388" type="Const" version="opset1">
+			<data element_type="u8" shape="1" offset="317" size="1" />
+			<output>
+				<port id="0" precision="U8">
+					<dim>1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="17" name="Constant_210390" type="Const" version="opset1">
+			<data element_type="u8" shape="3" offset="318" size="3" />
+			<output>
+				<port id="0" precision="U8">
+					<dim>3</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="18" name="RegexNormalization_210391" type="RegexNormalization" version="extension">
+			<data global_replace="true" />
+			<input>
+				<port id="0" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="1" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="2" precision="U8">
+					<dim>-1</dim>
+				</port>
+				<port id="3" precision="BOOL">
+					<dim>-1</dim>
+				</port>
+				<port id="4" precision="U8">
+					<dim>1</dim>
+				</port>
+				<port id="5" precision="U8">
+					<dim>3</dim>
+				</port>
+			</input>
+			<output>
+				<port id="6" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="7" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="8" precision="U8">
+					<dim>-1</dim>
+				</port>
+				<port id="9" precision="BOOL">
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="19" name="Constant_210393" type="Const" version="opset1">
+			<data element_type="u8" shape="1" offset="317" size="1" />
+			<output>
+				<port id="0" precision="U8">
+					<dim>1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="20" name="RegexSplit_210394" type="RegexSplit" version="extension">
+			<data behaviour="mergedwithprevious" invert="false" max_splits="-1" />
+			<input>
+				<port id="0" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="1" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="2" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="3" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="4" precision="U8">
+					<dim>-1</dim>
+				</port>
+				<port id="5" precision="BOOL">
+					<dim>-1</dim>
+				</port>
+				<port id="6" precision="U8">
+					<dim>1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="7" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="8" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="9" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="10" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="11" precision="U8">
+					<dim>-1</dim>
+				</port>
+				<port id="12" precision="BOOL">
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="21" name="Constant_210396" type="Const" version="opset1">
+			<data element_type="i32" shape="262144" offset="321" size="1048576" />
+			<output>
+				<port id="0" precision="I32">
+					<dim>262144</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="22" name="Constant_210398" type="Const" version="opset1">
+			<data element_type="i32" shape="262144" offset="1048897" size="1048576" />
+			<output>
+				<port id="0" precision="I32">
+					<dim>262144</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="23" name="Constant_210400" type="Const" version="opset1">
+			<data element_type="u8" shape="2316295" offset="2097473" size="2316295" />
+			<output>
+				<port id="0" precision="U8">
+					<dim>2316295</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="24" name="Constant_210408" type="Const" version="opset1">
+			<data element_type="i32" shape="514906" offset="4413768" size="2059624" />
+			<output>
+				<port id="0" precision="I32">
+					<dim>514906</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="25" name="Constant_210410" type="Const" version="opset1">
+			<data element_type="i32" shape="514906" offset="6473392" size="2059624" />
+			<output>
+				<port id="0" precision="I32">
+					<dim>514906</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="26" name="Constant_210412" type="Const" version="opset1">
+			<data element_type="u8" shape="2714340" offset="8533016" size="2714340" />
+			<output>
+				<port id="0" precision="U8">
+					<dim>2714340</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="27" name="Constant_210414" type="Const" version="opset1">
+			<data element_type="i32" shape="514906" offset="11247356" size="2059624" />
+			<output>
+				<port id="0" precision="I32">
+					<dim>514906</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="28" name="Constant_210416" type="Const" version="opset1">
+			<data element_type="i32" shape="514906" offset="13306980" size="2059624" />
+			<output>
+				<port id="0" precision="I32">
+					<dim>514906</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="29" name="Constant_210418" type="Const" version="opset1">
+			<data element_type="u8" shape="1903225" offset="15366604" size="1903225" />
+			<output>
+				<port id="0" precision="U8">
+					<dim>1903225</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="30" name="Constant_210402" type="Const" version="opset1">
+			<data element_type="i32" shape="23" offset="17269829" size="92" />
+			<output>
+				<port id="0" precision="I32">
+					<dim>23</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="31" name="Constant_210404" type="Const" version="opset1">
+			<data element_type="i32" shape="23" offset="17269921" size="92" />
+			<output>
+				<port id="0" precision="I32">
+					<dim>23</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="32" name="Constant_210406" type="Const" version="opset1">
+			<data element_type="u8" shape="198" offset="17270013" size="198" />
+			<output>
+				<port id="0" precision="U8">
+					<dim>198</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="33" name="Constant_210419" type="Const" version="opset1">
+			<data element_type="i32" shape="23" offset="17270211" size="92" />
+			<output>
+				<port id="0" precision="I32">
+					<dim>23</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="34" name="BPETokenizer_210420" type="BPETokenizer" version="extension">
+			<data unk_token="&lt;unk>" fuse_unk="true" suffix_indicator="" end_suffix="" byte_fallback="true" cache_capacity="52428" />
+			<input>
+				<port id="0" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="1" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="2" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="3" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="4" precision="U8">
+					<dim>-1</dim>
+				</port>
+				<port id="5" precision="I32">
+					<dim>262144</dim>
+				</port>
+				<port id="6" precision="I32">
+					<dim>262144</dim>
+				</port>
+				<port id="7" precision="U8">
+					<dim>2316295</dim>
+				</port>
+				<port id="8" precision="I32">
+					<dim>514906</dim>
+				</port>
+				<port id="9" precision="I32">
+					<dim>514906</dim>
+				</port>
+				<port id="10" precision="U8">
+					<dim>2714340</dim>
+				</port>
+				<port id="11" precision="I32">
+					<dim>514906</dim>
+				</port>
+				<port id="12" precision="I32">
+					<dim>514906</dim>
+				</port>
+				<port id="13" precision="U8">
+					<dim>1903225</dim>
+				</port>
+				<port id="14" precision="I32">
+					<dim>23</dim>
+				</port>
+				<port id="15" precision="I32">
+					<dim>23</dim>
+				</port>
+				<port id="16" precision="U8">
+					<dim>198</dim>
+				</port>
+				<port id="17" precision="I32">
+					<dim>23</dim>
+				</port>
+			</input>
+			<output>
+				<port id="18" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="19" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="20" precision="I32">
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="35" name="Constant_210421" type="Const" version="opset1">
+			<data element_type="i32" shape="" offset="17270303" size="4" />
+			<output>
+				<port id="0" precision="I32" />
+			</output>
+		</layer>
+		<layer id="36" name="Constant_210423" type="Const" version="opset1">
+			<data element_type="u8" shape="4" offset="17270307" size="4" />
+			<output>
+				<port id="0" precision="U8">
+					<dim>4</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="37" name="Constant_210425" type="Const" version="opset1">
+			<data element_type="u8" shape="13" offset="17270311" size="13" />
+			<output>
+				<port id="0" precision="U8">
+					<dim>13</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="38" name="Truncate_210426" type="Truncate" version="extension">
+			<data m_num_inputs="1" />
+			<input>
+				<port id="0" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="1" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="2" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="3" precision="I32" />
+				<port id="4" precision="U8">
+					<dim>4</dim>
+				</port>
+				<port id="5" precision="U8">
+					<dim>13</dim>
+				</port>
+			</input>
+			<output>
+				<port id="6" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="7" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="8" precision="I32">
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="39" name="Constant_210427" type="Const" version="opset1">
+			<data element_type="i32" shape="1" offset="17270324" size="4" />
+			<output>
+				<port id="0" precision="I32">
+					<dim>1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="40" name="CombineSegments_210428" type="CombineSegments" version="extension">
+			<input>
+				<port id="0" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="1" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="2" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="3" precision="I32">
+					<dim>1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="4" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="5" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="6" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="7" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="8" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="9" precision="I32">
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="41" name="Subtract_210429" type="Subtract" version="opset1">
+			<data auto_broadcast="numpy" />
+			<input>
+				<port id="0" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="1" precision="I32">
+					<dim>-1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="2" precision="I32">
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="42" name="Constant_210430" type="Const" version="opset1">
+			<data element_type="i32" shape="" offset="17270324" size="4" />
+			<output>
+				<port id="0" precision="I32" />
+			</output>
+		</layer>
+		<layer id="43" name="ReduceMax_210431" type="ReduceMax" version="opset1">
+			<data keep_dims="false" />
+			<input>
+				<port id="0" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="1" precision="I32" />
+			</input>
+			<output>
+				<port id="2" precision="I32" />
+			</output>
+		</layer>
+		<layer id="44" name="Constant_210432" type="Const" version="opset1">
+			<data element_type="i32" shape="" offset="17270324" size="4" />
+			<output>
+				<port id="0" precision="I32" />
+			</output>
+		</layer>
+		<layer id="45" name="RaggedToDense_210433" type="RaggedToDense" version="extension">
+			<data pad_right="false" m_pad_max_length="false" />
+			<input>
+				<port id="0" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="1" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="2" precision="I32">
+					<dim>-1</dim>
+				</port>
+				<port id="3" precision="I32" />
+				<port id="4" precision="I32" />
+			</input>
+			<output>
+				<port id="5" precision="I32">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+				<port id="6" precision="BOOL">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="46" name="Convert_210434" type="Convert" version="opset1">
+			<data destination_type="i32" />
+			<input>
+				<port id="0" precision="BOOL">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="1" precision="I32">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="47" name="Convert_210434.0" type="Convert" version="opset1">
+			<data destination_type="i64" />
+			<input>
+				<port id="0" precision="I32">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="1" precision="I64" names="attention_mask">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="48" name="RaggedToDense_210433.0" type="Convert" version="opset1">
+			<data destination_type="i64" />
+			<input>
+				<port id="0" precision="I32">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</input>
+			<output>
+				<port id="1" precision="I64" names="input_ids">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</output>
+		</layer>
+		<layer id="49" name="Result_210437" type="Result" version="opset1" output_names="input_ids">
+			<input>
+				<port id="0" precision="I64">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</input>
+		</layer>
+		<layer id="50" name="Result_210439" type="Result" version="opset1" output_names="attention_mask">
+			<input>
+				<port id="0" precision="I64">
+					<dim>-1</dim>
+					<dim>-1</dim>
+				</port>
+			</input>
+		</layer>
+	</layers>
+	<edges>
+		<edge from-layer="0" from-port="0" to-layer="2" to-port="0" />
+		<edge from-layer="1" from-port="0" to-layer="8" to-port="0" />
+		<edge from-layer="2" from-port="1" to-layer="3" to-port="0" />
+		<edge from-layer="2" from-port="3" to-layer="15" to-port="4" />
+		<edge from-layer="2" from-port="2" to-layer="15" to-port="3" />
+		<edge from-layer="2" from-port="1" to-layer="15" to-port="2" />
+		<edge from-layer="3" from-port="1" to-layer="6" to-port="0" />
+		<edge from-layer="4" from-port="0" to-layer="6" to-port="1" />
+		<edge from-layer="5" from-port="0" to-layer="6" to-port="2" />
+		<edge from-layer="6" from-port="3" to-layer="11" to-port="0" />
+		<edge from-layer="6" from-port="3" to-layer="8" to-port="1" />
+		<edge from-layer="7" from-port="0" to-layer="8" to-port="2" />
+		<edge from-layer="8" from-port="3" to-layer="15" to-port="0" />
+		<edge from-layer="9" from-port="0" to-layer="13" to-port="0" />
+		<edge from-layer="10" from-port="0" to-layer="11" to-port="1" />
+		<edge from-layer="11" from-port="2" to-layer="13" to-port="1" />
+		<edge from-layer="12" from-port="0" to-layer="13" to-port="2" />
+		<edge from-layer="13" from-port="3" to-layer="15" to-port="1" />
+		<edge from-layer="14" from-port="0" to-layer="15" to-port="5" />
+		<edge from-layer="15" from-port="11" to-layer="18" to-port="3" />
+		<edge from-layer="15" from-port="7" to-layer="20" to-port="1" />
+		<edge from-layer="15" from-port="6" to-layer="20" to-port="0" />
+		<edge from-layer="15" from-port="10" to-layer="18" to-port="2" />
+		<edge from-layer="15" from-port="9" to-layer="18" to-port="1" />
+		<edge from-layer="15" from-port="8" to-layer="18" to-port="0" />
+		<edge from-layer="16" from-port="0" to-layer="18" to-port="4" />
+		<edge from-layer="17" from-port="0" to-layer="18" to-port="5" />
+		<edge from-layer="18" from-port="6" to-layer="20" to-port="2" />
+		<edge from-layer="18" from-port="7" to-layer="20" to-port="3" />
+		<edge from-layer="18" from-port="8" to-layer="20" to-port="4" />
+		<edge from-layer="18" from-port="9" to-layer="20" to-port="5" />
+		<edge from-layer="19" from-port="0" to-layer="20" to-port="6" />
+		<edge from-layer="20" from-port="7" to-layer="34" to-port="0" />
+		<edge from-layer="20" from-port="8" to-layer="34" to-port="1" />
+		<edge from-layer="20" from-port="9" to-layer="34" to-port="2" />
+		<edge from-layer="20" from-port="10" to-layer="34" to-port="3" />
+		<edge from-layer="20" from-port="11" to-layer="34" to-port="4" />
+		<edge from-layer="21" from-port="0" to-layer="34" to-port="5" />
+		<edge from-layer="22" from-port="0" to-layer="34" to-port="6" />
+		<edge from-layer="23" from-port="0" to-layer="34" to-port="7" />
+		<edge from-layer="24" from-port="0" to-layer="34" to-port="8" />
+		<edge from-layer="25" from-port="0" to-layer="34" to-port="9" />
+		<edge from-layer="26" from-port="0" to-layer="34" to-port="10" />
+		<edge from-layer="27" from-port="0" to-layer="34" to-port="11" />
+		<edge from-layer="28" from-port="0" to-layer="34" to-port="12" />
+		<edge from-layer="29" from-port="0" to-layer="34" to-port="13" />
+		<edge from-layer="30" from-port="0" to-layer="34" to-port="14" />
+		<edge from-layer="31" from-port="0" to-layer="34" to-port="15" />
+		<edge from-layer="32" from-port="0" to-layer="34" to-port="16" />
+		<edge from-layer="33" from-port="0" to-layer="34" to-port="17" />
+		<edge from-layer="34" from-port="18" to-layer="38" to-port="0" />
+		<edge from-layer="34" from-port="19" to-layer="38" to-port="1" />
+		<edge from-layer="34" from-port="20" to-layer="38" to-port="2" />
+		<edge from-layer="35" from-port="0" to-layer="38" to-port="3" />
+		<edge from-layer="36" from-port="0" to-layer="38" to-port="4" />
+		<edge from-layer="37" from-port="0" to-layer="38" to-port="5" />
+		<edge from-layer="38" from-port="6" to-layer="40" to-port="0" />
+		<edge from-layer="38" from-port="8" to-layer="40" to-port="2" />
+		<edge from-layer="38" from-port="7" to-layer="40" to-port="1" />
+		<edge from-layer="39" from-port="0" to-layer="40" to-port="3" />
+		<edge from-layer="40" from-port="5" to-layer="41" to-port="0" />
+		<edge from-layer="40" from-port="4" to-layer="41" to-port="1" />
+		<edge from-layer="40" from-port="4" to-layer="45" to-port="0" />
+		<edge from-layer="40" from-port="5" to-layer="45" to-port="1" />
+		<edge from-layer="40" from-port="6" to-layer="45" to-port="2" />
+		<edge from-layer="41" from-port="2" to-layer="43" to-port="0" />
+		<edge from-layer="42" from-port="0" to-layer="43" to-port="1" />
+		<edge from-layer="43" from-port="2" to-layer="45" to-port="3" />
+		<edge from-layer="44" from-port="0" to-layer="45" to-port="4" />
+		<edge from-layer="45" from-port="6" to-layer="46" to-port="0" />
+		<edge from-layer="45" from-port="5" to-layer="48" to-port="0" />
+		<edge from-layer="46" from-port="1" to-layer="47" to-port="0" />
+		<edge from-layer="47" from-port="1" to-layer="50" to-port="0" />
+		<edge from-layer="48" from-port="1" to-layer="49" to-port="0" />
+	</edges>
+	<rt_info>
+		<info name="OpenVINO Runtime" value="2026.2.0-21622-fdad5b35a43" />
+		<add_attention_mask value="True" />
+		<add_prefix_space />
+		<add_special_tokens value="True" />
+		<bos_token_id value="2" />
+		<chat_template value="{%- macro format_parameters(properties, required) -%}&#10;    {%- set standard_keys = ['description', 'type', 'properties', 'required', 'nullable'] -%}&#10;    {%- set ns = namespace(found_first=false) -%}&#10;    {%- for key, value in properties | dictsort -%}&#10;        {%- set add_comma = false -%}&#10;        {%- if key not in standard_keys -%}&#10;            {%- if ns.found_first %},{% endif -%}&#10;            {%- set ns.found_first = true -%}&#10;            {{ key }}:{&#10;            {%- if value['description'] -%}&#10;                description:&lt;|&quot;|>{{ value['description'] }}&lt;|&quot;|>&#10;                {%- set add_comma = true -%}&#10;            {%- endif -%}&#10;            {%- if value['type'] | upper == 'STRING' -%}&#10;                {%- if value['enum'] -%}&#10;                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}&#10;                    enum:{{ format_argument(value['enum']) }}&#10;                {%- endif -%}&#10;            {%- elif value['type'] | upper == 'ARRAY' -%}&#10;                {%- if value['items'] is mapping and value['items'] -%}&#10;                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}&#10;                    items:{&#10;                    {%- set ns_items = namespace(found_first=false) -%}&#10;                    {%- for item_key, item_value in value['items'] | dictsort -%}&#10;                        {%- if item_value is not none -%}&#10;                            {%- if ns_items.found_first %},{% endif -%}&#10;                            {%- set ns_items.found_first = true -%}&#10;                            {%- if item_key == 'properties' -%}&#10;                                properties:{&#10;                                {%- if item_value is mapping -%}&#10;                                    {{- format_parameters(item_value, value['items']['required'] | default([])) -}}&#10;                                {%- endif -%}&#10;                                }&#10;                            {%- elif item_key == 'required' -%}&#10;                                required:[&#10;                                {%- for req_item in item_value -%}&#10;                                    &lt;|&quot;|>{{- req_item -}}&lt;|&quot;|>&#10;                                    {%- if not loop.last %},{% endif -%}&#10;                                {%- endfor -%}&#10;                                ]&#10;                            {%- elif item_key == 'type' -%}&#10;                                {%- if item_value is string -%}&#10;                                    type:{{ format_argument(item_value | upper) }}&#10;                                {%- else -%}&#10;                                    type:{{ format_argument(item_value | map('upper') | list) }}&#10;                                {%- endif -%}&#10;                            {%- else -%}&#10;                                {{ item_key }}:{{ format_argument(item_value) }}&#10;                            {%- endif -%}&#10;                        {%- endif -%}&#10;                    {%- endfor -%}&#10;                    }&#10;                {%- endif -%}&#10;            {%- endif -%}&#10;            {%- if value['nullable'] %}&#10;                {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}&#10;                nullable:true&#10;            {%- endif -%}&#10;            {%- if value['type'] | upper == 'OBJECT' -%}&#10;                {%- if value['properties'] is defined and value['properties'] is mapping -%}&#10;                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}&#10;                    properties:{&#10;                    {{- format_parameters(value['properties'], value['required'] | default([])) -}}&#10;                    }&#10;                {%- elif value is mapping -%}&#10;                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}&#10;                    properties:{&#10;                    {{- format_parameters(value, value['required'] | default([])) -}}&#10;                    }&#10;                {%- endif -%}&#10;                {%- if value['required'] -%}&#10;                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}&#10;                    required:[&#10;                    {%- for item in value['required'] | default([]) -%}&#10;                        &lt;|&quot;|>{{- item -}}&lt;|&quot;|>&#10;                        {%- if not loop.last %},{% endif -%}&#10;                    {%- endfor -%}&#10;                    ]&#10;                {%- endif -%}&#10;            {%- endif -%}&#10;            {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}&#10;            type:&lt;|&quot;|>{{ value['type'] | upper }}&lt;|&quot;|>}&#10;        {%- endif -%}&#10;    {%- endfor -%}&#10;{%- endmacro -%}&#10;{%- macro format_function_declaration(tool_data) -%}&#10;    declaration:{{- tool_data['function']['name'] -}}{description:&lt;|&quot;|>{{- tool_data['function']['description'] -}}&lt;|&quot;|>&#10;    {%- set params = tool_data['function']['parameters'] -%}&#10;    {%- if params -%}&#10;        ,parameters:{&#10;        {%- if params['properties'] -%}&#10;            properties:{ {{- format_parameters(params['properties'], params['required']) -}} },&#10;        {%- endif -%}&#10;        {%- if params['required'] -%}&#10;            required:[&#10;            {%- for item in params['required'] -%}&#10;                &lt;|&quot;|>{{- item -}}&lt;|&quot;|>&#10;                {{- ',' if not loop.last -}}&#10;            {%- endfor -%}&#10;            ],&#10;        {%- endif -%}&#10;        {%- if params['type'] -%}&#10;            type:&lt;|&quot;|>{{- params['type'] | upper -}}&lt;|&quot;|>}&#10;        {%- endif -%}&#10;    {%- endif -%}&#10;    {%- if 'response' in tool_data['function'] -%}&#10;        {%- set response_declaration = tool_data['function']['response'] -%}&#10;        ,response:{&#10;        {%- if response_declaration['description'] -%}&#10;            description:&lt;|&quot;|>{{- response_declaration['description'] -}}&lt;|&quot;|>,&#10;        {%- endif -%}&#10;        {%- if response_declaration['type'] | upper == 'OBJECT' -%}&#10;            type:&lt;|&quot;|>{{- response_declaration['type'] | upper -}}&lt;|&quot;|>}&#10;        {%- endif -%}&#10;    {%- endif -%}&#10;    }&#10;{%- endmacro -%}&#10;{%- macro format_argument(argument, escape_keys=True) -%}&#10;    {%- if argument is string -%}&#10;        {{- '&lt;|&quot;|>' + argument + '&lt;|&quot;|>' -}}&#10;    {%- elif argument is boolean -%}&#10;        {{- 'true' if argument else 'false' -}}&#10;    {%- elif argument is mapping -%}&#10;        {{- '{' -}}&#10;        {%- set ns = namespace(found_first=false) -%}&#10;        {%- for key, value in argument | dictsort -%}&#10;            {%- if ns.found_first %},{% endif -%}&#10;            {%- set ns.found_first = true -%}&#10;            {%- if escape_keys -%}&#10;                {{- '&lt;|&quot;|>' + key + '&lt;|&quot;|>' -}}&#10;            {%- else -%}&#10;                {{- key -}}&#10;            {%- endif -%}&#10;            :{{- format_argument(value, escape_keys=escape_keys) -}}&#10;        {%- endfor -%}&#10;        {{- '}' -}}&#10;    {%- elif argument is sequence -%}&#10;        {{- '[' -}}&#10;        {%- for item in argument -%}&#10;            {{- format_argument(item, escape_keys=escape_keys) -}}&#10;            {%- if not loop.last %},{% endif -%}&#10;        {%- endfor -%}&#10;        {{- ']' -}}&#10;    {%- else -%}&#10;        {{- argument -}}&#10;    {%- endif -%}&#10;{%- endmacro -%}&#10;{%- macro strip_thinking(text) -%}&#10;    {%- set ns = namespace(result='') -%}&#10;    {%- for part in text.split('&lt;channel|>') -%}&#10;        {%- if '&lt;|channel>' in part -%}&#10;            {%- set ns.result = ns.result + part.split('&lt;|channel>')[0] -%}&#10;        {%- else -%}&#10;            {%- set ns.result = ns.result + part -%}&#10;        {%- endif -%}&#10;    {%- endfor -%}&#10;    {{- ns.result | trim -}}&#10;{%- endmacro -%}&#10;&#10;{%- macro format_tool_response_block(tool_name, response) -%}&#10;    {{- '&lt;|tool_response>' -}}&#10;    {%- if response is mapping -%}&#10;        {{- 'response:' + tool_name + '{' -}}&#10;        {%- for key, value in response | dictsort -%}&#10;            {{- key -}}:{{- format_argument(value, escape_keys=False) -}}&#10;            {%- if not loop.last %},{% endif -%}&#10;        {%- endfor -%}&#10;        {{- '}' -}}&#10;    {%- else -%}&#10;        {{- 'response:' + tool_name + '{value:' + format_argument(response, escape_keys=False) + '}' -}}&#10;    {%- endif -%}&#10;    {{- '&lt;tool_response|>' -}}&#10;{%- endmacro -%}&#10;&#10;{%- set ns = namespace(prev_message_type=None) -%}&#10;{%- set loop_messages = messages -%}&#10;{{- bos_token -}}&#10;{#- Handle System/Tool Definitions Block -#}&#10;{%- if (enable_thinking is defined and enable_thinking) or tools or messages[0]['role'] in ['system', 'developer'] -%}&#10;    {{- '&lt;|turn>system\n' -}}&#10;&#10;    {#- Inject Thinking token at the very top of the FIRST system turn -#}&#10;    {%- if enable_thinking is defined and enable_thinking -%}&#10;        {{- '&lt;|think|>\n' -}}&#10;        {%- set ns.prev_message_type = 'think' -%}&#10;    {%- endif -%}&#10;&#10;    {%- if messages[0]['role'] in ['system', 'developer'] -%}&#10;        {{- messages[0]['content'] | trim -}}&#10;        {%- set loop_messages = messages[1:] -%}&#10;    {%- endif -%}&#10;&#10;    {%- if tools -%}&#10;        {%- for tool in tools %}&#10;            {{- '&lt;|tool>' -}}&#10;            {{- format_function_declaration(tool) | trim -}}&#10;            {{- '&lt;tool|>' -}}&#10;        {%- endfor %}&#10;        {%- set ns.prev_message_type = 'tool' -%}&#10;    {%- endif -%}&#10;&#10;    {{- '&lt;turn|>\n' -}}&#10;{%- endif %}&#10;&#10;{#- Pre-scan: find last user message index for reasoning guard -#}&#10;{%- set ns_turn = namespace(last_user_idx=-1) -%}&#10;{%- for i in range(loop_messages | length) -%}&#10;    {%- if loop_messages[i]['role'] == 'user' -%}&#10;        {%- set ns_turn.last_user_idx = i -%}&#10;    {%- endif -%}&#10;{%- endfor -%}&#10;&#10;{#- Loop through messages -#}&#10;{%- for message in loop_messages -%}&#10;    {%- if message['role'] != 'tool' -%}&#10;    {%- set ns.prev_message_type = None -%}&#10;    {%- set role = 'model' if message['role'] == 'assistant' else message['role'] -%}&#10;    {#- Detect continuation: suppress duplicate &lt;|turn>model when previous non-tool message was also assistant -#}&#10;    {%- set prev_nt = namespace(role=None, found=false) -%}&#10;    {%- if loop.index0 > 0 -%}&#10;        {%- for j in range(loop.index0 - 1, -1, -1) -%}&#10;            {%- if not prev_nt.found -%}&#10;                {%- if loop_messages[j]['role'] != 'tool' -%}&#10;                    {%- set prev_nt.role = loop_messages[j]['role'] -%}&#10;                    {%- set prev_nt.found = true -%}&#10;                {%- endif -%}&#10;            {%- endif -%}&#10;        {%- endfor -%}&#10;    {%- endif -%}&#10;    {%- set continue_same_model_turn = (role == 'model' and prev_nt.role == 'assistant') -%}&#10;    {%- if not continue_same_model_turn -%}&#10;        {{- '&lt;|turn>' + role + '\n' }}&#10;    {%- endif -%}&#10;&#10;    {#- Render reasoning/reasoning_content as thinking channel -#}&#10;    {%- set thinking_text = message.get('reasoning') or message.get('reasoning_content') -%}&#10;    {%- if thinking_text and loop.index0 > ns_turn.last_user_idx and message.get('tool_calls') -%}&#10;        {{- '&lt;|channel>thought\n' + thinking_text + '\n&lt;channel|>' -}}&#10;    {%- endif -%}&#10;&#10;            {%- if message['tool_calls'] -%}&#10;                {%- for tool_call in message['tool_calls'] -%}&#10;                    {%- set function = tool_call['function'] -%}&#10;                    {{- '&lt;|tool_call>call:' + function['name'] + '{' -}}&#10;                    {%- if function['arguments'] is mapping -%}&#10;                        {%- set ns_args = namespace(found_first=false) -%}&#10;                        {%- for key, value in function['arguments'] | dictsort -%}&#10;                            {%- if ns_args.found_first %},{% endif -%}&#10;                            {%- set ns_args.found_first = true -%}&#10;                            {{- key -}}:{{- format_argument(value, escape_keys=False) -}}&#10;                        {%- endfor -%}&#10;                    {%- elif function['arguments'] is string -%}&#10;                        {{- function['arguments'] -}}&#10;                    {%- endif -%}&#10;                    {{- '}&lt;tool_call|>' -}}&#10;                {%- endfor -%}&#10;                {%- set ns.prev_message_type = 'tool_call' -%}&#10;            {%- endif -%}&#10;&#10;            {%- set ns_tr_out = namespace(flag=false) -%}&#10;            {%- if message.get('tool_responses') -%}&#10;                {#- Legacy: tool_responses embedded on the assistant message (Google/Gemma native) -#}&#10;                {%- for tool_response in message['tool_responses'] -%}&#10;                    {{- format_tool_response_block(tool_response['name'] | default('unknown'), tool_response['response']) -}}&#10;                    {%- set ns_tr_out.flag = true -%}&#10;                    {%- set ns.prev_message_type = 'tool_response' -%}&#10;                {%- endfor -%}&#10;            {%- elif message.get('tool_calls') -%}&#10;                {#- OpenAI Chat Completions: forward-scan consecutive role:tool messages -#}&#10;                {%- set ns_tool_scan = namespace(stopped=false) -%}&#10;                {%- for k in range(loop.index0 + 1, loop_messages | length) -%}&#10;                    {%- if ns_tool_scan.stopped -%}&#10;                    {%- elif loop_messages[k]['role'] != 'tool' -%}&#10;                        {%- set ns_tool_scan.stopped = true -%}&#10;                    {%- else -%}&#10;                        {%- set follow = loop_messages[k] -%}&#10;                        {#- Resolve tool_call_id to function name -#}&#10;                        {%- set ns_tname = namespace(name=follow.get('name') | default('unknown')) -%}&#10;                        {%- for tc in message['tool_calls'] -%}&#10;                            {%- if tc.get('id') == follow.get('tool_call_id') -%}&#10;                                {%- set ns_tname.name = tc['function']['name'] -%}&#10;                            {%- endif -%}&#10;                        {%- endfor -%}&#10;                        {#- Handle content as string or content-parts array -#}&#10;                        {%- set tool_body = follow.get('content') -%}&#10;                        {%- if tool_body is string -%}&#10;                            {{- format_tool_response_block(ns_tname.name, tool_body) -}}&#10;                        {%- elif tool_body is sequence and tool_body is not string -%}&#10;                            {%- set ns_txt = namespace(s='') -%}&#10;                            {%- for part in tool_body -%}&#10;                                {%- if part.get('type') == 'text' -%}&#10;                                    {%- set ns_txt.s = ns_txt.s + (part.get('text') | default('')) -%}&#10;                                {%- endif -%}&#10;                            {%- endfor -%}&#10;                            {{- format_tool_response_block(ns_tname.name, ns_txt.s) -}}&#10;                        {%- else -%}&#10;                            {{- format_tool_response_block(ns_tname.name, tool_body) -}}&#10;                        {%- endif -%}&#10;                        {%- set ns_tr_out.flag = true -%}&#10;                        {%- set ns.prev_message_type = 'tool_response' -%}&#10;                    {%- endif -%}&#10;                {%- endfor -%}&#10;            {%- endif -%}&#10;&#10;            {%- if message['content'] is string -%}&#10;                {%- if role == 'model' -%}&#10;                    {{- strip_thinking(message['content']) -}}&#10;                {%- else -%}&#10;                    {{- message['content'] | trim -}}&#10;                {%- endif -%}&#10;            {%- elif message['content'] is sequence -%}&#10;                {%- for item in message['content'] -%}&#10;                    {%- if item['type'] == 'text' -%}&#10;                        {%- if role == 'model' -%}&#10;                            {{- strip_thinking(item['text']) -}}&#10;                        {%- else -%}&#10;                            {{- item['text'] | trim -}}&#10;                        {%- endif -%}&#10;                    {%- elif item['type'] == 'image' -%}&#10;                        {{- '&lt;|image|>' -}}&#10;                        {%- set ns.prev_message_type = 'image' -%}&#10;                    {%- elif item['type'] == 'audio' -%}&#10;                        {{- '&lt;|audio|>' -}}&#10;                        {%- set ns.prev_message_type = 'audio' -%}&#10;                    {%- elif item['type'] == 'video' -%}&#10;                        {{- '&lt;|video|>' -}}&#10;                        {%- set ns.prev_message_type = 'video' -%}&#10;                    {%- endif -%}&#10;                {%- endfor -%}&#10;            {%- endif -%}&#10;&#10;        {%- if ns.prev_message_type == 'tool_call' and not ns_tr_out.flag -%}&#10;            {{- '&lt;|tool_response>' -}}&#10;        {%- elif not (ns_tr_out.flag and not message.get('content')) -%}&#10;            {{- '&lt;turn|>\n' -}}&#10;        {%- endif -%}&#10;    {%- endif -%}&#10;{%- endfor -%}&#10;&#10;{%- if add_generation_prompt -%}&#10;    {%- if ns.prev_message_type != 'tool_response' and ns.prev_message_type != 'tool_call' -%}&#10;        {{- '&lt;|turn>model\n' -}}&#10;    {%- endif -%}&#10;{%- endif -%}" />
+		<clean_up_tokenization_spaces />
+		<detokenizer_input_type value="i64" />
+		<eos_token_id value="1" />
+		<handle_special_tokens_with_re />
+		<max_length />
+		<number_of_inputs value="1" />
+		<openvino_tokenizers_version value="2026.2.0.0-668-cb14978d71b" />
+		<openvino_version value="2026.2.0-21622-fdad5b35a43" />
+		<original_post_processor_template value="{&quot;type&quot;: &quot;TemplateProcessing&quot;, &quot;single&quot;: [{&quot;Sequence&quot;: {&quot;id&quot;: &quot;A&quot;, &quot;type_id&quot;: 0}}], &quot;pair&quot;: [{&quot;Sequence&quot;: {&quot;id&quot;: &quot;A&quot;, &quot;type_id&quot;: 0}}, {&quot;Sequence&quot;: {&quot;id&quot;: &quot;B&quot;, &quot;type_id&quot;: 1}}], &quot;special_tokens&quot;: {}}" />
+		<original_tokenizer_class value="&lt;class 'transformers.models.gemma.tokenization_gemma.GemmaTokenizer'>" />
+		<pad_token_id value="0" />
+		<processed_post_processor_template value="{&quot;single&quot;: {&quot;ids&quot;: [-1], &quot;type_ids&quot;: [0]}, &quot;pair&quot;: {&quot;ids&quot;: [-1, -2], &quot;type_ids&quot;: [0, 1]}}" />
+		<skip_special_tokens value="True" />
+		<streaming_detokenizer value="False" />
+		<tokenizer_output_type value="i64" />
+		<tokenizers_version value="0.22.2" />
+		<transformers_version value="5.5.0" />
+		<use_max_padding value="False" />
+		<use_sentencepiece_backend value="False" />
+		<utf8_replace_mode value="replace" />
+		<with_detokenizer value="True" />
+	</rt_info>
+</net>

openvino_vision_embeddings_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6c7609d984db75c39461f74b418cf3eac65b90da8678a08bc4e158e2f4524c3d
+size 169812532

preprocessor_config.json

@@ -0,0 +1,23 @@
+{
+  "do_convert_rgb": true,
+  "do_normalize": false,
+  "do_rescale": true,
+  "do_resize": true,
+  "image_mean": [
+    0.0,
+    0.0,
+    0.0
+  ],
+  "image_processor_type": "Gemma4ImageProcessor",
+  "image_seq_length": 280,
+  "image_std": [
+    1.0,
+    1.0,
+    1.0
+  ],
+  "max_soft_tokens": 280,
+  "patch_size": 16,
+  "pooling_kernel_size": 3,
+  "resample": 3,
+  "rescale_factor": 0.00392156862745098
+}

processor_config.json

@@ -0,0 +1,75 @@
+{
+  "audio_ms_per_token": 40,
+  "audio_seq_length": 750,
+  "feature_extractor": {
+    "dither": 0.0,
+    "feature_extractor_type": "Gemma4AudioFeatureExtractor",
+    "feature_size": 128,
+    "fft_length": 512,
+    "fft_overdrive": false,
+    "frame_length": 320,
+    "hop_length": 160,
+    "input_scale_factor": 1.0,
+    "max_frequency": 8000.0,
+    "mel_floor": 0.001,
+    "min_frequency": 0.0,
+    "padding_side": "right",
+    "padding_value": 0.0,
+    "per_bin_mean": null,
+    "per_bin_stddev": null,
+    "preemphasis": 0.0,
+    "preemphasis_htk_flavor": true,
+    "return_attention_mask": true,
+    "sampling_rate": 16000
+  },
+  "image_processor": {
+    "do_convert_rgb": true,
+    "do_normalize": false,
+    "do_rescale": true,
+    "do_resize": true,
+    "image_mean": [
+      0.0,
+      0.0,
+      0.0
+    ],
+    "image_processor_type": "Gemma4ImageProcessor",
+    "image_seq_length": 280,
+    "image_std": [
+      1.0,
+      1.0,
+      1.0
+    ],
+    "max_soft_tokens": 280,
+    "patch_size": 16,
+    "pooling_kernel_size": 3,
+    "resample": 3,
+    "rescale_factor": 0.00392156862745098
+  },
+  "image_seq_length": 280,
+  "processor_class": "Gemma4Processor",
+  "video_processor": {
+    "do_convert_rgb": true,
+    "do_normalize": true,
+    "do_rescale": true,
+    "do_resize": true,
+    "do_sample_frames": true,
+    "image_mean": [
+      0.0,
+      0.0,
+      0.0
+    ],
+    "image_std": [
+      1.0,
+      1.0,
+      1.0
+    ],
+    "max_soft_tokens": 70,
+    "num_frames": 32,
+    "patch_size": 16,
+    "pooling_kernel_size": 3,
+    "resample": 3,
+    "rescale_factor": 0.00392156862745098,
+    "return_metadata": false,
+    "video_processor_type": "Gemma4VideoProcessor"
+  }
+}

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cc8d3a0ce36466ccc1278bf987df5f71db1719b9ca6b4118264f45cb627bfe0f
+size 32169626

tokenizer_config.json

@@ -0,0 +1,95 @@
+{
+  "audio_token": "<|audio|>",
+  "backend": "tokenizers",
+  "boa_token": "<|audio>",
+  "boi_token": "<|image>",
+  "bos_token": "<bos>",
+  "eoa_token": "<audio|>",
+  "eoc_token": "<channel|>",
+  "eoi_token": "<image|>",
+  "eos_token": "<eos>",
+  "eot_token": "<turn|>",
+  "escape_token": "<|\"|>",
+  "etc_token": "<tool_call|>",
+  "etd_token": "<tool|>",
+  "etr_token": "<tool_response|>",
+  "extra_special_tokens": [
+    "<|video|>"
+  ],
+  "image_token": "<|image|>",
+  "is_local": false,
+  "mask_token": "<mask>",
+  "model_max_length": 1000000000000000019884624838656,
+  "model_specific_special_tokens": {
+    "audio_token": "<|audio|>",
+    "boa_token": "<|audio>",
+    "boi_token": "<|image>",
+    "eoa_token": "<audio|>",
+    "eoc_token": "<channel|>",
+    "eoi_token": "<image|>",
+    "eot_token": "<turn|>",
+    "escape_token": "<|\"|>",
+    "etc_token": "<tool_call|>",
+    "etd_token": "<tool|>",
+    "etr_token": "<tool_response|>",
+    "image_token": "<|image|>",
+    "soc_token": "<|channel>",
+    "sot_token": "<|turn>",
+    "stc_token": "<|tool_call>",
+    "std_token": "<|tool>",
+    "str_token": "<|tool_response>",
+    "think_token": "<|think|>"
+  },
+  "pad_token": "<pad>",
+  "padding_side": "left",
+  "processor_class": "Gemma4Processor",
+  "response_schema": {
+    "properties": {
+      "content": {
+        "type": "string"
+      },
+      "role": {
+        "const": "assistant"
+      },
+      "thinking": {
+        "type": "string"
+      },
+      "tool_calls": {
+        "items": {
+          "properties": {
+            "function": {
+              "properties": {
+                "arguments": {
+                  "additionalProperties": {},
+                  "type": "object",
+                  "x-parser": "gemma4-tool-call"
+                },
+                "name": {
+                  "type": "string"
+                }
+              },
+              "type": "object",
+              "x-regex": "call\\:(?P<name>\\w+)(?P<arguments>\\{.*\\})"
+            },
+            "type": {
+              "const": "function"
+            }
+          },
+          "type": "object"
+        },
+        "type": "array",
+        "x-regex-iterator": "<\\|tool_call>(.*?)<tool_call\\|>"
+      }
+    },
+    "type": "object",
+    "x-regex": "(\\<\\|channel\\>thought\\n(?P<thinking>.*?)\\<channel\\|\\>)?(?P<tool_calls>\\<\\|tool_call\\>.*\\<tool_call\\|\\>)?(?P<content>(?:(?!\\<turn\\|\\>)(?!\\<\\|tool_response\\>).)+)?(?:\\<turn\\|\\>|\\<\\|tool_response\\>)?"
+  },
+  "soc_token": "<|channel>",
+  "sot_token": "<|turn>",
+  "stc_token": "<|tool_call>",
+  "std_token": "<|tool>",
+  "str_token": "<|tool_response>",
+  "think_token": "<|think|>",
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>"
+}