Duplicate from onnx-community/functiongemma-270m-it-ONNX

Co-authored-by: Joshua <Xenova@users.noreply.huggingface.co>

.gitattributes

@@ -0,0 +1,40 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tar filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz 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
+onnx/model.onnx_data filter=lfs diff=lfs merge=lfs -text
+onnx/model_fp16.onnx_data filter=lfs diff=lfs merge=lfs -text
+onnx/model_q4.onnx_data filter=lfs diff=lfs merge=lfs -text
+onnx/model_q4f16.onnx_data filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,550 @@
+---
+license: gemma
+base_model:
+- google/functiongemma-270m-it
+library_name: transformers.js
+---
+
+
+# FunctionGemma model card
+
+**Model Page**: [FunctionGemma](https://ai.google.dev/gemma/docs/functiongemma)
+
+**Resources and Technical Documentation**:
+
+-   [Responsible Generative AI Toolkit](https://ai.google.dev/responsible)
+-   [FunctionGemma on Kaggle](https://www.kaggle.com/models/google/functiongemma/)
+-   [FunctionGemma on Vertex Model Garden](https://console.cloud.google.com/vertex-ai/publishers/google/model-garden/functiongemma)
+
+**Terms of Use**: [Terms](https://ai.google.dev/gemma/terms)\
+**Authors**: Google DeepMind
+
+## Model Information
+
+Summary description and brief definition of inputs and outputs.
+
+### Description
+
+> [!Note]
+> FunctionGemma is intended to be fine-tuned for your specific function-calling task, including multi-turn use cases.
+
+
+FunctionGemma is a lightweight, open model from Google, built as a foundation
+for creating your own specialized function calling models. FunctionGemma is not
+intended for use as a direct dialogue model, and is designed to be highly
+performant after further fine-tuning, as is typical of models this size. Built
+on the Gemma 3 270M model and with the same research and technology used to
+create the Gemini models, FunctionGemma has been trained specifically for
+function calling.  The model has the same architecture as Gemma 3, but uses a
+different chat format. The model is well suited for text-only function calling.
+The uniquely small size makes it possible to deploy in environments with limited
+resources such as laptops, desktops or your own cloud infrastructure,
+democratizing access to state of the art AI models and helping foster innovation
+for everyone. Furthermore, akin to the base Gemma 270M, the model has been
+optimized to be extremely versatile, performant on a variety of hardware in
+single turn scenarios, but should be finetuned on single turn or multiturn task
+specific data to achieve best accuracy in specific domains.
+To demonstrate how specializing the 270M parameter model can achieve high
+performance on specific agentic workflows, we have highlighted two use cases in
+the
+[Google AI Edge Gallery app](https://play.google.com/store/apps/details?id=com.google.ai.edge.gallery&pcampaignid=web_share).
+
+-   **Tiny Garden:** A model fine-tuned to power a voice-controlled
+    interactive game. It handles game logic to manage a virtual plot of land,
+    decomposing commands like "Plant sunflowers in the top row" and "Water the
+    flowers in plots 1 and 2" into app-specific functions (e.g., plant_seed,
+    water_plots) and coordinate targets. This demonstrates the model's capacity
+    to drive custom app mechanics without server connectivity.
+
+-   **Mobile Actions:** To empower developers to build their own expert
+    agents, we have published [a
+    dataset](https://huggingface.co/datasets/google/mobile-actions) and
+    [fine-tuning recipe](https://github.com/google-gemini/gemma-cookbook/blob/main/FunctionGemma/%5BFunctionGemma%5DFinetune_FunctionGemma_270M_for_Mobile_Actions_with_Hugging_Face.ipynb)
+    to demonstrate fine-tuning FunctionGemma. It translates user inputs (e.g.,
+    "Create a calendar event for lunch," "Turn on the flashlight") into
+    function calls that trigger Android OS system tools. This interactive
+    notebook demonstrates how to take the base FunctionGemma model and build a
+    "Mobile Actions" fine tune from scratch for use in the
+    [Google AI Edge gallery app](https://play.google.com/store/apps/details?id=com.google.ai.edge.gallery&pcampaignid=web_share).
+    This use case demonstrates the model's ability to act as an offline,
+    private agent for personal device tasks.
+
+### Inputs and outputs
+
+-   **Input:**
+    -   Text string, such as a question, a prompt, or a document to be
+        summarized
+    -   Total input context of  32K tokens
+-   **Output:**
+    -   Generated text in response to the input, such as an answer to a
+        question, or a summary of a document
+    -   Total output context up to 32K tokens per request, subtracting
+        the request input tokens
+
+### Basic Usage
+
+The following is a code example of how to use FunctionGemma to generate a function call from a JSON definition using the Hugging Face Transformers.js library. 
+
+If you haven't already, you can install the [Transformers.js](https://huggingface.co/docs/transformers.js) JavaScript library from [NPM](https://www.npmjs.com/package/@huggingface/transformers) using:
+```bash
+npm i @huggingface/transformers
+```
+
+You can then use the model as follows:
+
+```js
+import { AutoModelForCausalLM, AutoTokenizer } from "@huggingface/transformers";
+
+// Load the model and tokenizer
+const model_id = "onnx-community/functiongemma-270m-it-ONNX";
+const tokenizer = await AutoTokenizer.from_pretrained(model_id);
+const model = await AutoModelForCausalLM.from_pretrained(model_id);
+
+const weather_function_schema = {
+  type: "function",
+  function: {
+    name: "get_current_temperature",
+    description: "Gets the current temperature for a given location.",
+    parameters: {
+      type: "object",
+      properties: {
+        location: {
+          type: "string",
+          description: "The city name, e.g. San Francisco",
+        },
+      },
+      required: ["location"],
+    },
+  },
+};
+
+const messages = [
+  {
+    role: "developer",
+    content: "You are a model that can do function calling with the following functions",
+  },
+  {
+    role: "user",
+    content: "What's the temperature in London?",
+  },
+];
+
+const inputs = tokenizer.apply_chat_template(messages, {
+  tools: [weather_function_schema],
+  tokenize: true,
+  add_generation_prompt: true,
+  return_dict: true,
+});
+
+const output = await model.generate({ ...inputs, max_new_tokens: 512 });
+const decoded = tokenizer.decode(output.slice(0, [inputs.input_ids.dims[1], null]), { skip_special_tokens: false });
+console.log(decoded);
+// <start_function_call>call:get_current_temperature{location:<escape>London<escape>}<end_function_call><start_function_response>
+```
+
+For more detailed examples see the [Gemma documentation](https://ai.google.dev/gemma/docs/functiongemma).
+
+## Model Data
+
+Data used for model training and how the data was processed.
+
+### Training Dataset
+
+These models were trained on a dataset of text data that includes a wide
+variety of sources. The model was trained with 6T tokens. The knowledge cutoff
+date for the training data was August 2024. There are the key components:
+
+-   Public Tool Definitions - Common APIs found on the web
+-   Tool Use Interactions - These are a mix of prompts, function calls,
+    function responses, and natural language responses from the model to
+    summarise the function call response, or request clarifications when the
+    prompt is ambiguous or incomplete.
+
+### Data Preprocessing
+
+Here are the key data cleaning and filtering methods applied to the training
+data:
+
+-   CSAM Filtering: Rigorous CSAM (Child Sexual Abuse Material) filtering
+    was applied at multiple stages in the data preparation process to ensure
+    the exclusion of harmful and illegal content.
+-   Sensitive Data Filtering: As part of making Gemma pre-trained models
+    safe and reliable, automated techniques were used to filter out certain
+    personal information and other sensitive data from training sets.
+-   Additional methods: Filtering based on content quality and safety in
+    line with
+    [our policies](https://ai.google/static/documents/ai-responsibility-update-published-february-2025.pdf).
+
+## Implementation Information
+
+Details about the model internals.
+
+### Hardware
+
+Gemma was trained using [Tensor Processing Unit
+(TPU)](https://cloud.google.com/tpu/docs/intro-to-tpu) hardware (TPUv4p, TPUv5p
+and TPUv5e). Training vision-language models (VLMs) requires significant
+computational power. TPUs, designed specifically for matrix operations common in
+machine learning, offer several advantages in this domain:
+
+-   Performance: TPUs are specifically designed to handle the massive
+    computations involved in training VLMs. They can speed up training
+    considerably compared to CPUs.
+-   Memory: TPUs often come with large amounts of high-bandwidth memory,
+    allowing for the handling of large models and batch sizes during training.
+    This can lead to better model quality.
+-   Scalability: TPU Pods (large clusters of TPUs) provide a scalable
+    solution for handling the growing complexity of large foundation models.
+    You can distribute training across multiple TPU devices for faster and more
+    efficient processing.
+-   Cost-effectiveness: In many scenarios, TPUs can provide a more
+    cost-effective solution for training large models compared to CPU-based
+    infrastructure, especially when considering the time and resources saved
+    due to faster training.
+-   These advantages are aligned with
+    [Google's commitments to operate sustainably](https://sustainability.google/operating-sustainably/).
+
+### Software
+
+Training was done using [JAX](https://github.com/jax-ml/jax) and
+[ML Pathways](https://blog.google/technology/ai/introducing-pathways-next-generation-ai-architecture/).
+JAX allows researchers to take advantage of the latest generation of hardware,
+including TPUs, for faster and more efficient training of large models. ML
+Pathways is Google's latest effort to build artificially intelligent systems
+capable of generalizing across multiple tasks. This is specially suitable for
+foundation models, including large language models like these ones.\
+Together, JAX and ML Pathways are used as described in the [paper about the
+Gemini family of models](https://goo.gle/gemma2report); *"the 'single
+controller' programming model of Jax and Pathways allows a single Python process
+to orchestrate the entire training run, dramatically simplifying the development
+workflow."*
+
+## Evaluation
+
+Model evaluation metrics and results.
+
+### Benchmark Results
+
+<table>
+  <thead>
+    <tr>
+      <th><strong>Benchmark</strong></th>
+      <th><strong>n-shot</strong></th>
+      <th><strong>Function Gemma 270m</strong></th>
+    </tr>
+  </thead>
+  <tbody>
+    <tr>
+      <td>BFCL Simple</td>
+      <td>0-shot</td>
+      <td>61.6</td>
+    </tr>
+    <tr>
+      <td>BFCL Multiple</td>
+      <td>0-shot</td>
+      <td>63.5</td>
+    </tr>
+    <tr>
+      <td>BFCL Parallel</td>
+      <td>0-shot</td>
+      <td>39</td>
+    </tr>
+    <tr>
+      <td>BFCL Parallel Multiple</td>
+      <td>0-shot</td>
+      <td>29.5</td>
+    </tr>
+    <tr>
+      <td>BFCL Live Simple </td>
+      <td>0-shot</td>
+      <td>36.2</td>
+    </tr>
+    <tr>
+      <td>BFCL Live Multiple</td>
+      <td>0-shot</td>
+      <td>25.7</td>
+    </tr>
+    <tr>
+      <td>BFCL Live Parallel</td>
+      <td>0-shot</td>
+      <td>22.9</td>
+    </tr>
+    <tr>
+      <td>BFCL Live Parallel Multiple</td>
+      <td>0-shot</td>
+      <td>20.8</td>
+    </tr>
+    <tr>
+      <td>BFCL Relevance</td>
+      <td>0-shot</td>
+      <td>61.1</td>
+    </tr>
+    <tr>
+      <td>BFCL Irrelevance</td>
+      <td>0-shot</td>
+      <td>73.7</td>
+    </tr>
+  </tbody>
+</table>
+
+**Impact on Performance after Fine-tuning on Mobile Actions Dataset**\
+To demonstrate the value of specialization for small language models, we
+compared the base FunctionGemma model against the fine-tuned model using the
+"Mobile Actions"
+[recipe](https://github.com/google-gemini/gemma-cookbook/blob/main/FunctionGemma/%5BFunctionGemma%5DFinetune_FunctionGemma_270M_for_Mobile_Actions_with_Hugging_Face.ipynb).
+Fine-tuning significantly improved the base FunctionGemma model's ability to
+correctly identify and format mobile system calls.
+
+<table>
+  <thead>
+    <tr>
+      <th><br>
+Model</th>
+      <th><br>
+Eval results for Mobile Actions</th>
+    </tr>
+  </thead>
+  <tbody>
+    <tr>
+      <td><br>
+Base FunctionGemma model</td>
+      <td><br>
+58%</td>
+    </tr>
+    <tr>
+      <td><br>
+Mobile Actions Fine-Tune</td>
+      <td><br>
+85%</td>
+    </tr>
+  </tbody>
+</table>
+
+**On-Device Performance of the Gemma 270m Fine-tuned Use Cases**\
+We evaluated the fine-tuned use cases on a Samsung S25 Ultra to assess on-device
+latency and memory footprint.
+
+-   **Context:** 512 prefill tokens and 32 decode tokens.
+-   **Hardware:** S25 Ultra CPU using LiteRT XNNPACK delegate with 4 threads.
+
+Mobile Actions On Device Performance
+
+<table>
+  <thead>
+    <tr>
+      <th><br>
+Backend</th>
+      <th><br>
+Quantization scheme</th>
+      <th><br>
+Context length</th>
+      <th><br>
+Prefill (tokens per second)</th>
+      <th><br>
+Decode (tokens per second)</th>
+      <th><br>
+Time-to-first-token (seconds)</th>
+      <th><br>
+Model Size (MB)</th>
+      <th><br>
+Peak RSS Memory (MB)</th>
+    </tr>
+  </thead>
+  <tbody>
+    <tr>
+      <td><br>
+CPU</td>
+      <td><br>
+dynamic_int8</td>
+      <td><br>
+1024</td>
+      <td><br>
+1718</td>
+      <td><br>
+125.9</td>
+      <td><br>
+0.3</td>
+      <td><br>
+288</td>
+      <td><br>
+551</td>
+    </tr>
+  </tbody>
+</table>
+
+Tiny Garden On Device Performance
+
+<table>
+  <thead>
+    <tr>
+      <th><br>
+Backend</th>
+      <th><br>
+Quantization scheme</th>
+      <th><br>
+Context length</th>
+      <th><br>
+Prefill (tokens per second)</th>
+      <th><br>
+Decode (tokens per second)</th>
+      <th><br>
+Time-to-first-token (seconds)</th>
+      <th><br>
+Model Size (MB)</th>
+      <th><br>
+Peak RSS Memory (MB)</th>
+    </tr>
+  </thead>
+  <tbody>
+    <tr>
+      <td><br>
+CPU</td>
+      <td><br>
+dynamic_int8</td>
+      <td><br>
+1024</td>
+      <td><br>
+1743</td>
+      <td><br>
+125.7</td>
+      <td><br>
+0.3</td>
+      <td><br>
+288</td>
+      <td><br>
+549</td>
+    </tr>
+  </tbody>
+</table>
+
+## Ethics and Safety
+
+Ethics and safety evaluation approach and results.
+
+### Evaluation Approach
+
+Our evaluation methods include structured evaluations and internal red-teaming
+testing of relevant content policies. Red-teaming was conducted by a number of
+different teams, each with different goals and human evaluation metrics. These
+models were evaluated against a number of different categories relevant to
+ethics and safety, including:
+
+-   **Child Safety**: Evaluation of text-to-text and image to text prompts
+    covering child safety policies, including child sexual abuse and exploitation.
+-   **Content Safety:** Evaluation of text-to-text and image to text prompts
+    covering safety policies including, harassment, violence and gore, and hate
+    speech.
+-   **Representational Harms**: Evaluation of text-to-text and image to text
+    prompts covering safety policies including bias, stereotyping, and harmful
+    associations or inaccuracies.
+
+### Evaluation Results
+
+For all areas of safety testing, we saw major improvements in the categories of
+child safety, content safety, and representational harms relative to previous
+Gemma models. All testing was conducted without safety filters to evaluate the
+model capabilities and behaviors. The model produced minimal policy violations,
+and showed significant  improvements over previous Gemma models' performance
+with respect to ungrounded inferences. A limitation of our evaluations was they
+included only English language prompts.
+
+## Usage and Limitations
+
+These models have certain limitations that users should be aware of.
+
+### Intended Usage
+
+This model is not intended for use as a direct dialogue model.\
+Open Large Language Models (LLMs) have a wide range of applications across
+various industries and domains. The following list of potential uses is not
+comprehensive. The purpose of this list is to provide contextual information
+about the possible use-cases that the model creators considered as part of model
+training and development.
+
+-   Content Creation and Communication
+    -   Text Generation: These models can be used to generate creative
+        text formats such as poems, scripts, code, marketing copy, and email drafts.
+    -   Chatbots and Conversational AI: Power conversational interfaces
+        for customer service, virtual assistants, or interactive applications.
+    -   Text Summarization: Generate concise summaries of a text corpus,
+        research papers, or reports.
+-   Research and Education
+    -   Natural Language Processing (NLP) Research: These models can
+        serve as a foundation for researchers to experiment with NLP
+        techniques, develop algorithms, and contribute to the advancement of the field.
+    -   Language Learning Tools: Support interactive language learning
+        experiences, aiding in grammar correction or providing writing practice.
+    -   Knowledge Exploration: Assist researchers in exploring large
+        bodies of text by generating summaries or answering questions about
+        specific topics.
+
+### Limitations
+
+-   Training Data
+    -   The quality and diversity of the training data significantly
+        influence the model's capabilities. Biases or gaps in the training data
+        can lead to limitations in the model's responses.
+    -   The scope of the training dataset determines the subject areas
+        the model can handle effectively.
+-   Context and Task Complexity
+    -   Models are better at tasks that can be framed with clear
+        prompts and instructions. Open-ended or highly complex tasks might be
+        challenging.
+    -   A model's performance can be influenced by the amount of context
+        provided (longer context generally leads to better outputs, up to a
+        certain point).
+-   Language Ambiguity and Nuance
+    -   Natural language is inherently complex. Models might struggle
+        to grasp subtle nuances, sarcasm, or figurative language.
+-   Factual Accuracy
+    -   Models generate responses based on information they learned
+        from their training datasets, but they are not knowledge bases. They
+        may generate incorrect or outdated factual statements.
+-   Common Sense
+    -   Models rely on statistical patterns in language. They might
+        lack the ability to apply common sense reasoning in certain situations.
+
+### Ethical Considerations and Risks
+
+The development of large language models (LLMs) raises several ethical
+concerns. In creating an open model, we have carefully considered the
+following:
+
+-   Bias and Fairness
+    -   LLMs trained on large-scale, real-world text data can reflect
+        socio-cultural biases embedded in the training material. These models
+        underwent careful scrutiny, input data pre-processing described and
+        posterior evaluations reported in this card.
+-   Misinformation and Misuse
+    -   LLMs can be misused to generate text that is false, misleading,
+        or harmful.
+    -   Guidelines are provided for responsible use with the model, see
+        the [Responsible Generative AI Toolkit](https://ai.google.dev/responsible).
+-   Transparency and Accountability:
+    -   This model card summarizes details on the models' architecture,
+        capabilities, limitations, and evaluation processes.
+    -   A responsibly developed open model offers the opportunity to
+        share innovation by making LLM technology accessible to developers and
+        researchers across the AI ecosystem.
+
+Risks identified and mitigations:
+
+-   Perpetuation of biases: It's encouraged to perform continuous
+    monitoring (using evaluation metrics, human review) and the exploration of
+    de-biasing techniques during model training, fine-tuning, and other use cases.
+-   Generation of harmful content: Mechanisms and guidelines for content
+    safety are essential. Developers are encouraged to exercise caution and
+    implement appropriate content safety safeguards based on their specific
+    product policies and application use cases.
+-   Misuse for malicious purposes: Technical limitations and developer and
+    end-user education can help mitigate against malicious applications of
+    LLMs. Educational resources and reporting mechanisms for users to flag
+    misuse are provided. Prohibited uses of Gemma models are outlined in the
+    [Gemma Prohibited Use Policy](https://ai.google.dev/gemma/prohibited_use_policy)..
+-   Privacy violations: Models were trained on data filtered for removal of
+    PII (Personally Identifiable Information). Developers are encouraged to
+    adhere to privacy regulations with privacy-preserving techniques.
+
+### Benefits
+
+At the time of release, this family of models provides high-performance open large language model implementations designed from the ground up for Responsible AI development compared to similarly sized models.

chat_template.jinja

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

config.json

@@ -0,0 +1,73 @@
+{
+  "_sliding_window_pattern": 6,
+  "architectures": [
+    "Gemma3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_logit_softcapping": null,
+  "bos_token_id": 2,
+  "dtype": "bfloat16",
+  "eos_token_id": 1,
+  "final_logit_softcapping": null,
+  "head_dim": 256,
+  "hidden_activation": "gelu_pytorch_tanh",
+  "hidden_size": 640,
+  "initializer_range": 0.02,
+  "intermediate_size": 2048,
+  "layer_types": [
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "model_type": "gemma3_text",
+  "num_attention_heads": 4,
+  "num_hidden_layers": 18,
+  "num_key_value_heads": 1,
+  "pad_token_id": 0,
+  "query_pre_attn_scalar": 256,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "full_attention": {
+      "rope_theta": 1000000.0,
+      "rope_type": "default"
+    },
+    "sliding_attention": {
+      "rope_theta": 10000.0,
+      "rope_type": "default"
+    }
+  },
+  "sliding_window": 512,
+  "transformers_version": "5.0.0.dev0",
+  "use_bidirectional_attention": false,
+  "use_cache": true,
+  "vocab_size": 262144,
+  "transformers.js_config": {
+    "use_external_data_format": {
+      "model.onnx": 1,
+      "model_fp16.onnx": 1,
+      "model_q4.onnx": 1,
+      "model_q4f16.onnx": 1
+    },
+    "kv_cache_dtype": {
+      "q4f16": "float16",
+      "fp16": "float16"
+    }
+  }
+}

generation_config.json

@@ -0,0 +1,13 @@
+{
+  "cache_implementation": "hybrid",
+  "do_sample": true,
+  "eos_token_id": [
+    1,
+    50,
+    106
+  ],
+  "top_k": 64,
+  "top_p": 0.95,
+  "transformers_version": "5.0.0.dev0",
+  "trust_remote_code": false
+}

onnx/model.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:07c362225485a837effc6e21834b76fe542c861f522944063dd757a33336dd3f
+size 502654

onnx/model.onnx_data

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b513fd3a9d1633a21ebbc4330e0c193a7e118915d21f17079fe672a7dfb546b6
+size 1139501568

onnx/model_fp16.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:16bf5d6e9249e11e52306a3dc7fc8a9b2e85eac1a7e3f9e884614d4eb6465b4c
+size 619409

onnx/model_fp16.onnx_data

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f4360c8c74dd9d2315c7a367baa65383338af0d51241632ab403fc00bb57c375
+size 569862656

onnx/model_q4.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f3f12f387ee22d2b8c1b308bb6b2967ceb7c3466688950c5d7ccee29620ecede
+size 430147

onnx/model_q4.onnx_data

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0d717b24f73233099ff714cd75abf8990fbb17ef1c75bd026a629800d3e3e3ec
+size 801090048

onnx/model_q4f16.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8dc9fb5e2b0aa34f527309f0ecaeb9b824b5ad9a9613350168753054c180e145
+size 518626

onnx/model_q4f16.onnx_data

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b30ca95e4b31014ec791d7589f8c6416b8056ffc4f39093aa7ceb3ad37f2a0c7
+size 425724416

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:69fde4ada54844b6a7b94494e97f93c581c80cc6610c87e7b45d223077542169
+size 20316979

tokenizer_config.json

@@ -0,0 +1,28 @@
+{
+  "additional_special_tokens": null,
+  "backend": "tokenizers",
+  "boi_token": "<start_of_image>",
+  "bos_token": "<bos>",
+  "clean_up_tokenization_spaces": false,
+  "eoi_token": "<end_of_image>",
+  "eos_token": "<eos>",
+  "image_token": "<image_soft_token>",
+  "is_local": false,
+  "mask_token": "<mask>",
+  "model_max_length": 1000000000000000019884624838656,
+  "model_specific_special_tokens": {
+    "boi_token": "<start_of_image>",
+    "eoi_token": "<end_of_image>",
+    "image_token": "<image_soft_token>",
+    "sfr_token": "<start_function_response>"
+  },
+  "pad_token": "<pad>",
+  "padding_side": "left",
+  "sfr_token": "<start_function_response>",
+  "sp_model_kwargs": null,
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false,
+  "chat_template": "{%- macro format_parameters(properties, required) -%}\n    {%- set standard_keys = ['description', 'type', 'properties', 'required', 'nullable'] -%}\n    {%- set ns = namespace(found_first=false) -%}\n    {%- for key, value in properties | dictsort -%}\n        {%- if key not in standard_keys -%}\n            {%- if ns.found_first %},{% endif -%}\n            {%- set ns.found_first = true -%}\n            {{- key }}:{description:<escape>{{ value['description'] }}<escape>\n            {%- if value['type'] | upper == 'STRING' -%}\n                {%- if value['enum'] -%}\n                    ,enum:{{ format_argument(value['enum']) }}\n                {%- endif -%}\n            {%- elif value['type'] | upper == 'OBJECT' -%}\n                ,properties:{\n                {%- if value['properties'] is defined and value['properties'] is mapping -%}\n                    {{- format_parameters(value['properties'], value['required'] | default([])) -}}\n                {%- elif value is mapping -%}\n                    {{- format_parameters(value, value['required'] | default([])) -}}\n                {%- endif -%}\n                }\n                {%- if value['required'] -%}\n                    ,required:[\n                    {%- for item in value['required'] | default([]) -%}\n                        <escape>{{- item -}}<escape>\n                        {%- if not loop.last %},{% endif -%}\n                    {%- endfor -%}\n                    ]\n                {%- endif -%}\n            {%- elif value['type'] | upper == 'ARRAY' -%}\n                {%- if value['items'] is mapping and value['items'] -%}\n                    ,items:{\n                    {%- set ns_items = namespace(found_first=false) -%}\n                    {%- for item_key, item_value in value['items'] | dictsort -%}\n                        {%- if item_value is not none -%}\n                            {%- if ns_items.found_first %},{% endif -%}\n                            {%- set ns_items.found_first = true -%}\n                            {%- if item_key == 'properties' -%}\n                                properties:{\n                                {%- if item_value is mapping -%}\n                                    {{- format_parameters(item_value, value['items']['required'] | default([])) -}}\n                                {%- endif -%}\n                                }\n                            {%- elif item_key == 'required' -%}\n                                required:[\n                                {%- for req_item in item_value -%}\n                                    <escape>{{- req_item -}}<escape>\n                                    {%- if not loop.last %},{% endif -%}\n                                {%- endfor -%}\n                                ]\n                            {%- elif item_key == 'type' -%}\n                                {%- if item_value is string -%}\n                                    type:{{ format_argument(item_value | upper) }}\n                                {%- else -%}\n                                    type:{{ format_argument(item_value | map('upper') | list) }}\n                                {%- endif -%}\n                            {%- else -%}\n                                {{ item_key }}:{{ format_argument(item_value) }}\n                            {%- endif -%}\n                        {%- endif -%}\n                    {%- endfor -%}\n                    }\n                {%- endif -%}\n            {%- endif -%}\n            ,type:<escape>{{ value['type'] | upper }}<escape>}\n        {%- endif -%}\n    {%- endfor -%}\n{%- endmacro -%}\n{% macro format_function_declaration(tool_data) -%}\ndeclaration:{{- tool_data['function']['name'] -}}\n{description:<escape>{{- tool_data['function']['description'] -}}<escape>\n{%- set params = tool_data['function']['parameters'] -%}\n{%- if params -%}\n    ,parameters:{\n    {%- if params['properties'] -%}\n        properties:{ {{- format_parameters(params['properties'], params['required']) -}} },\n    {%- endif -%}\n    {%- if params['required'] -%}\n        required:[\n        {%- for item in params['required'] -%}\n            <escape>{{- item -}}<escape>\n            {{- ',' if not loop.last -}}\n        {%- endfor -%}\n        ],\n    {%- endif -%}\n    {%- if params['type'] -%}\n        type:<escape>{{- params['type'] | upper -}}<escape>}\n    {%- endif -%}\n{%- endif -%}\n}\n{%- endmacro -%}\n{% macro format_argument(argument, escape_keys=True) -%}\n{%- if argument is string -%}\n    {{- '<escape>' + argument + '<escape>' -}}\n{%- elif argument is boolean -%}\n    {%- if argument -%}\n        {{- 'true' -}}\n    {%- else -%}\n        {{- 'false' -}}\n    {%- endif -%}\n{%- elif argument is mapping -%}\n    {{- '{' -}}\n    {%- set ns = namespace(found_first=false) -%}\n    {%- for key, value in argument | dictsort -%}\n        {%- if ns.found_first %},{% endif -%}\n        {%- set ns.found_first = true -%}\n        {%- if escape_keys -%}\n            {{- '<escape>' + key + '<escape>' -}}\n        {%- else -%}\n            {{- key -}}\n        {%- endif -%}\n        :{{- format_argument(value, escape_keys=escape_keys) -}}\n    {%- endfor -%}\n    {{- '}' -}}\n{%- elif argument is iterable -%}\n    {{- '[' -}}\n    {%- for item in argument -%}\n        {{- format_argument(item, escape_keys=escape_keys) -}}\n        {%- if not loop.last %},{% endif -%}\n    {%- endfor -%}\n    {{- ']' -}}\n{%- else -%}\n    {{- argument -}}\n{%- endif -%}\n{%- endmacro -%}\n{{ bos_token }}\n{%- set ns = namespace(prev_message_type=None) -%}\n{#- Tool Declarations -#}\n{%- set loop_messages = messages -%}\n{%- if tools or messages[0]['role'] == 'system' or messages[0]['role'] == 'developer' -%}\n    {{- '<start_of_turn>developer\\n' -}}\n    {%- if messages[0]['role'] == 'system' or messages[0]['role'] == 'developer' -%}\n        {%- if messages[0]['content'] is string -%}\n            {{- messages[0]['content'] | trim -}}\n        {%- elif messages[0]['content'] is iterable -%}\n            {%- for item in messages[0]['content'] -%}\n                {%- if item['type'] == 'text' -%}\n                    {{- item['text'] | trim -}}\n                {%- endif -%}\n            {%- endfor -%}\n        {%- endif -%}\n        {%- set loop_messages = messages[1:] -%}\n    {%- endif -%}\n    {%- if tools -%}\n        {%- for tool in tools %}\n            {{- '<start_function_declaration>' -}}\n            {{- format_function_declaration(tool) | trim }}\n            {{- '<end_function_declaration>' -}}\n        {%- endfor %}\n    {%- endif -%}\n    {{- '<end_of_turn>\\n' }}\n{%- endif %}\n{#- Loop through messages. -#}\n{%- for message in loop_messages -%}\n    {%- if (message['role'] == 'assistant') -%}\n        {#- Rename \"assistant\" to \"model\". -#}\n        {%- set role = \"model\" -%}\n    {%- else -%}\n        {%- set role = message['role'] -%}\n    {%- endif -%}\n    {%- if role != 'tool' -%}\n        {%- if ns.prev_message_type != 'tool_response' -%}\n            {{- '<start_of_turn>' + role + '\\n' }}\n        {%- endif -%}\n        {%- set ns.prev_message_type = None -%}\n        {%- if 'content' in message and message['content'] is not none -%}\n            {%- if message['content'] is string -%}\n                {{ message['content'] | trim }}\n            {%- elif message['content'] is iterable -%}\n                {%- for item in message['content'] -%}\n                    {%- if item['type'] == 'image' -%}\n                        {{ '<start_of_image>' }}\n                    {%- elif item['type'] == 'text' -%}\n                        {{ item['text'] | trim }}\n                    {%- endif -%}\n                {%- endfor -%}\n            {%- else -%}\n                {{ raise_exception(\"Invalid content type in user/assistant message\") }}\n            {%- endif -%}\n            {%- set ns.prev_message_type = 'content' -%}\n        {%- endif -%}\n        {%- if 'tool_calls' in message and message['tool_calls'] and message['tool_calls'] is iterable -%}\n            {#- Tool Calls -#}\n            {%- for tool_call in message['tool_calls'] -%}\n                {% set function = tool_call['function'] %}\n                {{-  '<start_function_call>call:' + function['name'] + '{' -}}\n                {%- if 'arguments' in function -%}\n                    {%- if function['arguments'] is mapping -%}\n                        {%- set ns = namespace(found_first=false) -%}\n                        {%- for key, value in function['arguments'] | dictsort -%}\n                            {%- if ns.found_first %},{% endif -%}\n                            {%- set ns.found_first = true -%}\n                            {{- key -}}:{{- format_argument(value, escape_keys=False) -}}\n                        {%- endfor -%}\n                    {%- elif function['arguments'] is string -%}\n                        {# This handles string-JSON, just in case #}\n                    {{ function['arguments'] }}\n                    {%- endif %}\n                {%- endif -%}\n                {{- '}<end_function_call>' -}}\n            {%- endfor -%}\n            {%- if loop.last -%}\n                {{ '<start_function_response>' }}\n            {%- endif -%}\n            {%- set ns.prev_message_type = 'tool_call' -%}\n        {%- endif -%}\n    {%- else -%}\n        {#- Tool Responses -#}\n        {%- if 'content' in message and message['content'] -%}\n            {%- if message['content'] is mapping -%}\n                {%- if 'name' in message['content'] and 'response' in message['content'] -%}\n                    {{ '<start_function_response>response:' + message['content']['name'] | trim + '{' }}\n                    {%- set response_ns = namespace(found_first=false) -%}\n                    {%- for key, value in message['content']['response'] | dictsort -%}\n                        {%- if response_ns.found_first %},{% endif -%}\n                        {%- set response_ns.found_first = true -%}\n                        {{- key -}}:{{- format_argument(value, escape_keys=False) -}}\n                    {%- endfor -%}\n                    {{- '}<end_function_response>' -}}\n                {%- elif 'name' in message -%}\n                    {{ '<start_function_response>response:' + message['name'] | trim + '{' }}\n                    {%- set response_ns = namespace(found_first=false) -%}\n                    {%- for key, value in message['content'] | dictsort -%}\n                        {%- if response_ns.found_first %},{% endif -%}\n                        {%- set response_ns.found_first = true -%}\n                        {{- key -}}:{{- format_argument(value, escape_keys=False) -}}\n                    {%- endfor -%}\n                    {{- '}<end_function_response>' -}}\n                {%- else -%}\n                    {{ raise_exception(\"Invalid tool response mapping: must contain 'name' and 'response' keys, or 'name' must be in the message.\") }}\n                {%- endif -%}\n            {%- elif message['content'] is string -%}\n                {%- if 'name' in message -%}\n                     {{ '<start_function_response>response:' + message['name'] | trim + '{value:' + format_argument(message['content'], escape_keys=False) + '}<end_function_response>' }}\n                {%- else -%}\n                     {{ raise_exception(\"Invalid tool response: 'name' must be provided.\") }}\n                {%- endif -%}\n            {%- elif message['content'] is iterable -%}\n                {%- for item in message['content'] -%}\n                    {%- if item is mapping -%}\n                        {%- if 'name' in item and 'response' in item -%}\n                            {{ '<start_function_response>response:' + item['name'] | trim + '{' }}\n                            {%- set response_ns = namespace(found_first=false) -%}\n                            {%- for key, value in item['response'] | dictsort -%}\n                                {%- if response_ns.found_first %},{% endif -%}\n                                {%- set response_ns.found_first = true -%}\n                                {{- key -}}:{{- format_argument(value, escape_keys=False) -}}\n                            {%- endfor -%}\n                            {{- '}<end_function_response>' -}}\n                        {%- elif 'name' in message -%}\n                            {{ '<start_function_response>response:' + message['name'] | trim + '{' }}\n                            {%- set response_ns = namespace(found_first=false) -%}\n                            {%- for key, value in item | dictsort -%}\n                                {%- if response_ns.found_first %},{% endif -%}\n                                {%- set response_ns.found_first = true -%}\n                                {{- key -}}:{{- format_argument(value, escape_keys=False) -}}\n                            {%- endfor -%}\n                            {{- '}<end_function_response>' -}}\n                        {%- else -%}\n                            {{ raise_exception(\"Invalid tool response mapping: must contain 'name' and 'response' keys, or 'name' must be in the message.\") }}\n                        {%- endif -%}\n                    {%- else -%}\n                        {{ raise_exception(\"Invalid tool response message: multiple responses must all be mappings\") }}\n                    {%- endif -%}\n                {%- endfor -%}\n            {%- else -%}\n                {{ raise_exception(\"Invalid content type in tool message: must be mapping, sequence of mappings, or string.\") }}\n            {%- endif -%}\n        {%- endif -%}\n        {%- set ns.prev_message_type = 'tool_response' -%}\n    {%- endif -%}\n    {%- if ns.prev_message_type not in ['tool_call', 'tool_response'] -%}\n        {{ '<end_of_turn>\\n' }}\n    {%- endif -%}\n{%- endfor -%}\n{%- if add_generation_prompt -%}\n    {%- if ns.prev_message_type != 'tool_response' -%}\n        {{- '<start_of_turn>model\\n' -}}\n    {%- endif -%}\n{%- endif -%}\n"
+}