Upload folder using huggingface_hub

.gitattributes

@@ -33,3 +33,5 @@ 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
+aitriyaj-banner.png filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

README.md

@@ -1,3 +1,384 @@
----
-license: gemma
----
+---
+language:
+- tr
+license: gemma
+base_model: mlx-community/gemma-4-26B-A4B-it-OptiQ-4bit
+tags:
+- medical
+- triage
+- turkish
+- lora
+- mlx
+- emergency-medicine
+pipeline_tag: text-generation
+---
+
+# Gemma 4 26B A4B — Turkish Emergency Triage
+
+![AI-Triyaj](aitriyaj-banner.png)
+
+Türkiye acil servislerine özel, Türkçe triyaj kararı veren konuşmacı dil modeli.
+
+Küresel triyaj sistemleri (CTAS, ESI) genel nüfus ve geniş klinik senaryolar için tasarlanmış olup uygulamada uzun değerlendirme süreçleri gerektirmektedir. Bu model, Türkiye acil servis pratiğine uygun olarak; hızlı, odaklı ve spesifik sorularla hastayı yönlendiren, gereksiz adımları elemek yerine doğrudan karar noktasına ulaşan bir triyaj akışı için tasarlandı. Hasta ile Türkçe doğal dil diyalogu kurarak semptomları toplar ve **RED / YELLOW / GREEN** triyaj kararı üretir.
+
+> **Base model:** `mlx-community/gemma-4-26B-A4B-it-OptiQ-4bit`  
+> **Fine-tuning:** LoRA (mlx-lm + mlx-optiq), 2000 iterasyon  
+> **Format:** 4-bit quantized safetensors (fused, adapter ayrı gerekmez)
+
+---
+
+## Benchmark
+
+![Triage Model Benchmark](benchmark.png)
+
+Fine-tuned modelimiz 648 vakalık test setinde tüm baseline modelleri geride bıraktı.  
+Öne çıkan fark: diğer modeller RED'e aşırı yönelirken, fine-tuned model üç sınıf arasında en dengeli dağılımı gösterdi.
+
+---
+
+## Görev
+
+Hastanın şikayetini Türkçe konuşma diliyle alır, yapılandırılmış sorularla klinik bilgiyi toplar ve triyaj kararını verir:
+
+| Karar | Anlam |
+|-------|-------|
+| 🔴 RED | Acil, hemen müdahale gerektirir |
+| 🟡 YELLOW | Yarı acil, kısa sürede bakılmalı |
+| 🟢 GREEN | Acil değil, bekleyebilir |
+
+---
+
+## Model Mimarisi
+
+- **Gemma 4 26B A4B** (Mixture-of-Experts): 26B toplam parametre, token başına 4B aktif
+- **4-bit OptiQ quantization**: hassasiyet-odaklı mixed-precision (attention katmanları 8-bit, MLP 4-bit)
+- **LoRA fine-tuning**: yalnızca adapter ağırlıkları eğitildi, ardından base model ile birleştirildi
+
+---
+
+## Eğitim Verisi
+
+Eğitim verisi, bir acil tıp doktoru ile iş birliği yapılarak oluşturuldu.
+
+Doktorumuz, 117 farklı şikayet kategorisi (göğüs ağrısı, nefes darlığı, karın ağrısı vb.) için CTAS/ESI protokollerine dayalı ve Türkiye'deki sisteme uygun **karar ağaçları (path)** tasarladı. Her path; hangi soruların sorulacağını, hangi cevabın hangi triyaj kararına yol açacağını ve doğrudan RED kararı gerektiren tetikleyici semptomları tanımlar.
+
+Bu pathler temel alınarak **Gemini** ile 6.475 sentetik Türkçe diyalog üretildi:
+
+- **441 karar yolu** (path) — 117 şikayet kategorisi
+- **Hasta profili çeşitliliği:** temiz / hafif messy / ağır messy / çelişkili
+- **Muğlak cevap senaryoları** (%20): hasta önce belirsiz cevap verir, asistan netleştirir
+- **8 farklı senaryo stili:** yüz yüze triyaj, 112 araması, ambulansla geliş vb.
+- **Gerçek hasta demografisi:** Iran ED Dataset'inden çekilen yaş/cinsiyet dağılımı
+- **Risk faktörleri:** diyabet, kalp hastası, hamilelik, hipertansiyon vb.
+
+| Split | Kayıt |
+|-------|-------|
+| Train | 5.180 |
+| Val   | 646   |
+| Test  | 648   |
+| **Toplam** | **6.474** |
+
+---
+
+## Fine-tuning
+
+Eğitim, Apple Silicon üzerinde MacBook M4 Pro 24GB kullanılarak tamamen yerel olarak **mlx-lm** ve **mlx-optiq** kütüphaneleri ile gerçekleştirildi.
+
+- **mlx-lm**: Apple'ın MLX framework'ü üzerine inşa edilmiş, Mac'te verimli LLM eğitimi ve inferans sağlayan açık kaynak kütüphane
+- **mlx-optiq**: Hassasiyet-odaklı (sensitivity-aware) LoRA eğitimi; hangi katmanların daha hassas olduğunu analiz ederek mixed-precision quantization uygular — bu sayede model kalitesi korunurken bellek kullanımı düşük tutulur
+- **LoRA (Low-Rank Adaptation)**: Tüm modeli yeniden eğitmek yerine küçük adapter ağırlıkları eklenerek eğitim yapıldı; eğitim sonunda adapter base model ile birleştirildi (fused)
+
+```
+Framework  : mlx-lm 0.31.3 + mlx-optiq
+Yöntem     : LoRA (r=8, alpha=16)
+İterasyon  : 2000
+Batch size : 4
+Max seq len: 1024
+Optimizer  : AdamW
+Donanım    : Apple Silicon (MacBook M4 Pro 24GB)
+```
+
+Val loss: **9.9 → 0.761** (2000 iterasyon)
+
+---
+
+## Değerlendirme
+
+Test seti: 648 örnek, tüm modeller aynı diyaloglar üzerinde, data leakage olmadan değerlendirildi.
+
+### Genel Accuracy
+
+| Model | Accuracy | Doğru/Toplam |
+|-------|----------|--------------|
+| **Gemma 4 26B Fine-tuned (bizim)** | **75.0%** | 486/648 |
+| Gemini Flash 3.1 Lite | 73.5% | 476/648 |
+| GPT-5.4 | 71.3% | 462/648 |
+| GPT-4.1 | 70.2% | 455/648 |
+| GPT-4.1 mini | 60.8% | 394/648 |
+| Gemma 4 26B Base (fine-tune öncesi) | 56.8% | 25/50 |
+
+### Per-Class Recall (Fine-tuned Model)
+
+| Sınıf | Recall | n | RED | YELLOW | GREEN |
+|-------|--------|---|-----|--------|-------|
+| RED | 76.0% | 317 | 241 | 72 | 4 |
+| YELLOW | 79.2% | 197 | 15 | 156 | 26 |
+| GREEN | 66.4% | 134 | 0 | 45 | 89 |
+
+> **Not:** RED→YELLOW karışması (%22.7) en kritik hata tipidir. Gerçek klinik kullanımda insan gözetimi zorunludur.
+
+---
+
+## Kullanım
+
+### MLX ile (önerilen)
+
+```bash
+pip install mlx-lm
+```
+
+```python
+from mlx_lm import load, generate
+
+model, tokenizer = load("ai-triyaj/gemma4-26B-A4B-triage-turkish")
+
+SYSTEM = """Sen bir acil servis triyaj asistanısın. Hastayla Türkçe konuşarak şikayetini anlıyorsun.
+Her turda tek soru soruyorsun. Yeterli bilgiyi topladıktan sonra son mesajında nötr kapanış yap ve triyaj kararını söyle.
+Son mesaj formatı: "Bilgiler için teşekkürler, sizi ilgili birime yönlendiriyorum. Karar: RED/YELLOW/GREEN"
+Triyaj kararı: RED (kırmızı) = acil, YELLOW (sarı) = yarı acil, GREEN (yeşil) = acil değil."""
+
+messages = [
+    {"role": "system", "content": SYSTEM},
+    {"role": "user", "content": "Göğsümde ağrı var"},
+]
+
+prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
+response = generate(model, tokenizer, prompt=prompt, max_tokens=200)
+print(response)
+```
+
+### Terminal Chat
+
+```bash
+python -m mlx_lm.chat \
+  --model ai-triyaj/gemma4-26B-A4B-triage-turkish \
+  --system "Sen bir acil servis triyaj asistanısın. Hastayla Türkçe konuşarak şikayetini anlıyorsun. Her turda tek soru soruyorsun. Yeterli bilgiyi topladıktan sonra son mesajında nötr kapanış yap ve triyaj kararını söyle. Son mesaj formatı: 'Bilgiler için teşekkürler, sizi ilgili birime yönlendiriyorum. Karar: RED/YELLOW/GREEN' Triyaj kararı: RED (kırmızı) = acil, YELLOW (sarı) = yarı acil, GREEN (yeşil) = acil değil."
+```
+
+### LM Studio
+
+1. LM Studio → Search
+2. `ai-triyaj/gemma4-26B-A4B-triage-turkish` ara
+3. İndir ve başlat
+
+---
+
+## Sistem Gereksinimleri
+
+| Donanım | Durum |
+|---------|-------|
+| Apple Silicon M2 Pro (16GB) | ✅ Çalışır |
+| Apple Silicon M3/M4 (16GB+) | ✅ Önerilen |
+| CUDA GPU (8GB+) | PyTorch safetensors ile çalışabilir |
+
+---
+
+## Sınırlamalar ve Uyarılar
+
+- Bu model **araştırma amaçlıdır**, klinik karar desteği için **insan gözetimi zorunludur**
+- Eğitim verisi sentetik olup gerçek hasta popülasyonunu tam temsil etmeyebilir
+- RED→YELLOW karışması (%22.7) hayati risk taşıyan vakalarda yanlış önceliklendirmeye yol açabilir
+- Model yalnızca **Türkçe** için optimize edilmiştir
+
+---
+
+## Teşekkür
+
+Bu projenin klinik temeli, **Dr. Altuğ Hasanbaşoğlu**'nun uzmanlığı ve katkılarıyla oluşturuldu. Türkiye acil servis pratiğine uygun triyaj karar ağaçlarının tasarımı, klinik kriterlerin belirlenmesi ve doğrulanması süreçlerindeki değerli katkıları için kendisine içtenlikle teşekkür ederiz.
+
+---
+
+## Lisans
+
+Bu model Google'ın [Gemma Lisansı](https://ai.google.dev/gemma/terms) kapsamındadır.
+
+---
+---
+
+# Gemma 4 26B A4B — Turkish Emergency Triage (English)
+
+![AI-Triyaj](aitriyaj-banner.png)
+
+A conversational language model designed for Turkish emergency department triage, making **RED / YELLOW / GREEN** triage decisions through natural dialogue with patients in Turkish.
+
+> **Base model:** `mlx-community/gemma-4-26B-A4B-it-OptiQ-4bit`  
+> **Fine-tuning:** LoRA (mlx-lm + mlx-optiq), 2000 iterations  
+> **Format:** 4-bit quantized safetensors (fused, no separate adapter needed)
+
+---
+
+## Overview
+
+Global triage systems such as CTAS and ESI are designed for broad patient populations and wide clinical scenarios, often requiring lengthy assessment workflows. This model was built specifically for Turkish emergency department practice — guiding patients through fast, focused, and symptom-specific questions that cut straight to the decision point without unnecessary steps. It engages patients in natural Turkish dialogue, collects clinical information, and outputs a triage decision.
+
+| Decision | Meaning |
+|----------|---------|
+| 🔴 RED | Critical — requires immediate intervention |
+| 🟡 YELLOW | Semi-urgent — should be seen shortly |
+| 🟢 GREEN | Non-urgent — can wait |
+
+---
+
+## Model Architecture
+
+- **Gemma 4 26B A4B** (Mixture-of-Experts): 26B total parameters, 4B active per token
+- **4-bit OptiQ quantization**: sensitivity-aware mixed-precision (attention layers at 8-bit, MLP at 4-bit)
+- **LoRA fine-tuning**: only adapter weights were trained, then merged into the base model (fused)
+
+---
+
+## Training Data
+
+The training data was developed in collaboration with an emergency medicine physician.
+
+Our physician designed **decision trees (paths)** for 117 complaint categories (chest pain, shortness of breath, abdominal pain, etc.), grounded in CTAS/ESI protocols and adapted to Turkish emergency practice. Each path defines which questions to ask, which answers lead to which triage decision, and which symptoms immediately trigger a RED classification.
+
+Based on these paths, **6,475 synthetic Turkish dialogues** were generated using Gemini:
+
+- **441 decision paths** across 117 complaint categories
+- **Patient profile diversity:** clean / mildly messy / severely messy / contradictory
+- **Ambiguous response scenarios** (20%): patient gives vague answer, assistant clarifies
+- **8 scenario styles:** face-to-face triage, emergency call, ambulance arrival, etc.
+- **Real patient demographics:** age/gender distribution from the Iran ED Dataset
+- **Risk factors:** diabetes, cardiac history, pregnancy, hypertension, etc.
+
+| Split | Records |
+|-------|---------|
+| Train | 5,180 |
+| Val   | 646   |
+| Test  | 648   |
+| **Total** | **6,474** |
+
+---
+
+## Fine-tuning
+
+Training was performed entirely locally on Apple Silicon using **mlx-lm** and **mlx-optiq**:
+
+- **mlx-lm**: open-source library built on Apple's MLX framework for efficient on-device LLM training and inference
+- **mlx-optiq**: sensitivity-aware LoRA training; analyzes which layers are most sensitive to quantization and applies mixed-precision accordingly — preserving model quality while keeping memory footprint low
+- **LoRA**: instead of retraining the full model, small adapter weights were added and trained; after training the adapter was fused into the base model
+
+```
+Framework  : mlx-lm 0.31.3 + mlx-optiq
+Method     : LoRA (r=8, alpha=16)
+Iterations : 2000
+Batch size : 4
+Max seq len: 1024
+Optimizer  : AdamW
+Hardware   : Apple Silicon (MacBook M4 Pro 24GB)
+```
+
+Val loss: **9.9 → 0.761** (2000 iterations)
+
+---
+
+## Evaluation
+
+Test set: 648 samples, all models evaluated on the same dialogues with no data leakage.
+
+### Overall Accuracy
+
+| Model | Accuracy | Correct/Total |
+|-------|----------|---------------|
+| **Gemma 4 26B Fine-tuned (ours)** | **75.0%** | 486/648 |
+| Gemini Flash 3.1 Lite | 73.5% | 476/648 |
+| GPT-5.4 | 71.3% | 462/648 |
+| GPT-4.1 | 70.2% | 455/648 |
+| GPT-4.1 mini | 60.8% | 394/648 |
+| Gemma 4 26B Base (before fine-tuning) | 56.8% | 25/50 |
+
+### Per-Class Recall (Fine-tuned Model)
+
+| Class | Recall | n | RED | YELLOW | GREEN |
+|-------|--------|---|-----|--------|-------|
+| RED | 76.0% | 317 | 241 | 72 | 4 |
+| YELLOW | 79.2% | 197 | 15 | 156 | 26 |
+| GREEN | 66.4% | 134 | 0 | 45 | 89 |
+
+> **Note:** RED→YELLOW confusion (22.7%) is the most critical error type. Human oversight is mandatory for any real clinical use.
+
+---
+
+## Usage
+
+### With MLX (recommended)
+
+```bash
+pip install mlx-lm
+```
+
+```python
+from mlx_lm import load, generate
+
+model, tokenizer = load("ai-triyaj/gemma4-26B-A4B-triage-turkish")
+
+SYSTEM = """You are an emergency department triage assistant. Speak with the patient in Turkish to understand their complaint.
+Ask one question per turn. Once you have gathered enough information, close with a neutral statement and give the triage decision.
+Final message format: "Bilgiler için teşekkürler, sizi ilgili birime yönlendiriyorum. Karar: RED/YELLOW/GREEN"
+Triage decision: RED = critical, YELLOW = semi-urgent, GREEN = non-urgent."""
+
+messages = [
+    {"role": "system", "content": SYSTEM},
+    {"role": "user", "content": "Göğsümde ağrı var"},
+]
+
+prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
+response = generate(model, tokenizer, prompt=prompt, max_tokens=200)
+print(response)
+```
+
+### Terminal Chat
+
+```bash
+python -m mlx_lm.chat \
+  --model ai-triyaj/gemma4-26B-A4B-triage-turkish \
+  --system "Sen bir acil servis triyaj asistanısın. Hastayla Türkçe konuşarak şikayetini anlıyorsun. Her turda tek soru soruyorsun. Yeterli bilgiyi topladıktan sonra son mesajında nötr kapanış yap ve triyaj kararını söyle. Son mesaj formatı: 'Bilgiler için teşekkürler, sizi ilgili birime yönlendiriyorum. Karar: RED/YELLOW/GREEN' Triyaj kararı: RED (kırmızı) = acil, YELLOW (sarı) = yarı acil, GREEN (yeşil) = acil değil."
+```
+
+### LM Studio
+
+1. LM Studio → Search
+2. Search for `ai-triyaj/gemma4-26B-A4B-triage-turkish`
+3. Download and launch
+
+---
+
+## System Requirements
+
+| Hardware | Status |
+|----------|--------|
+| Apple Silicon M2 Pro (16GB) | ✅ Works |
+| Apple Silicon M3/M4 (16GB+) | ✅ Recommended |
+| CUDA GPU (8GB+) | Works via PyTorch safetensors |
+
+---
+
+## Limitations & Warnings
+
+- This model is intended for **research purposes only** — human oversight is mandatory for clinical decision support
+- Training data is synthetic and may not fully represent real patient populations
+- RED→YELLOW confusion (22.7%) may lead to incorrect prioritization in life-threatening cases
+- The model is optimized for **Turkish only**
+
+---
+
+## Acknowledgements
+
+The clinical foundation of this project was built with the expertise and contributions of **Dr. Altuğ Hasanbaşoğlu**. We sincerely thank him for designing the triage decision trees adapted to Turkish emergency practice, defining the clinical criteria, and validating the decision logic throughout the project.
+
+---
+
+## License
+
+This model is subject to Google's [Gemma License](https://ai.google.dev/gemma/terms).

chat_template.jinja

@@ -0,0 +1,347 @@
+{%- 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' -}}
+        {%- if not enable_thinking | default(false) -%}
+            {{- '<|channel>thought\n<channel|>' -}}
+        {%- endif -%}
+    {%- endif -%}
+{%- endif -%}

config.json

@@ -0,0 +1,2712 @@
+{
+    "architectures": [
+        "Gemma4ForCausalLM"
+    ],
+    "dtype": "bfloat16",
+    "eos_token_id": [
+        1,
+        106,
+        50
+    ],
+    "initializer_range": 0.02,
+    "model_type": "gemma4",
+    "quantization": {
+        "group_size": 64,
+        "bits": 4,
+        "mode": "affine",
+        "language_model.model.embed_tokens": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.0.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.0.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.0.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.0.self_attn.o_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.0.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.0.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.0.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.0.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.0.experts.switch_glu.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.0.experts.switch_glu.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.0.experts.switch_glu.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.1.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.1.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.1.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.1.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.1.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.1.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.1.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.1.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.1.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.1.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.1.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.2.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.2.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.2.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.2.self_attn.o_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.2.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.2.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.2.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.2.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.2.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.2.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.2.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.3.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.3.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.3.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.3.self_attn.o_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.3.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.3.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.3.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.3.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.3.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.3.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.3.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.4.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.4.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.4.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.4.self_attn.o_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.4.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.4.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.4.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.4.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.4.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.4.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.4.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.5.self_attn.q_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.5.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.5.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.5.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.5.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.5.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.5.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.5.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.5.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.5.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.6.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.6.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.6.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.6.self_attn.o_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.6.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.6.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.6.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.6.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.6.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.6.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.6.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.7.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.7.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.7.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.7.self_attn.o_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.7.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.7.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.7.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.7.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.7.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.7.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.7.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.8.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.8.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.8.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.8.self_attn.o_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.8.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.8.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.8.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.8.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.8.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.8.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.8.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.9.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.9.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.9.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.9.self_attn.o_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.9.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.9.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.9.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.9.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.9.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.9.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.9.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.10.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.10.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.10.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.10.self_attn.o_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.10.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.10.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.10.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.10.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.10.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.10.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.10.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.11.self_attn.q_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.11.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.11.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.11.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.11.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.11.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.11.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.11.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.11.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.11.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.12.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.12.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.12.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.12.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.12.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.12.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.12.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.12.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.12.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.12.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.12.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.13.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.13.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.13.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.13.self_attn.o_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.13.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.13.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.13.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.13.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.13.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.13.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.13.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.14.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.14.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.14.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.14.self_attn.o_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.14.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.14.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.14.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.14.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.14.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.14.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.14.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.15.self_attn.q_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.15.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.15.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.15.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.15.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.15.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.15.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.15.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.15.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.15.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.15.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.16.self_attn.q_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.16.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.16.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.16.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.16.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.16.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.16.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.16.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.16.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.16.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.16.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.17.self_attn.q_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.17.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.17.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.17.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.17.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.17.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.17.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.17.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.17.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.17.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.18.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.18.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.18.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.18.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.18.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.18.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.18.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.18.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.18.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.18.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.18.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.19.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.19.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.19.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.19.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.19.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.19.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.19.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.19.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.19.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.19.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.19.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.20.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.20.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.20.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.20.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.20.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.20.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.20.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.20.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.20.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.20.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.20.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.21.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.21.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.21.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.21.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.21.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.21.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.21.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.21.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.21.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.21.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.21.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.22.self_attn.q_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.22.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.22.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.22.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.22.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.22.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.22.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.22.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.22.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.22.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.22.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.23.self_attn.q_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.23.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.23.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.23.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.23.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.23.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.23.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.23.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.23.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.23.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.24.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.24.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.24.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.24.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.24.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.24.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.24.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.24.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.24.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.24.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.24.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.25.self_attn.q_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.25.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.25.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.25.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.25.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.25.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.25.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.25.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.25.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.25.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.25.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.26.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.26.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.26.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.26.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.26.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.26.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.26.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.26.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.26.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.26.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.26.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.27.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.27.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.27.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.27.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.27.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.27.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.27.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.27.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.27.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.27.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.27.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.28.self_attn.q_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.28.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.28.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.28.self_attn.o_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.28.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.28.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.28.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.28.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.28.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.28.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.28.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.29.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.29.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.29.self_attn.o_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.29.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.29.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.29.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.29.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.29.experts.switch_glu.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.29.experts.switch_glu.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.29.experts.switch_glu.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        }
+    },
+    "quantization_config": {
+        "group_size": 64,
+        "bits": 4,
+        "mode": "affine",
+        "language_model.model.embed_tokens": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.0.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.0.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.0.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.0.self_attn.o_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.0.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.0.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.0.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.0.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.0.experts.switch_glu.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.0.experts.switch_glu.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.0.experts.switch_glu.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.1.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.1.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.1.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.1.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.1.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.1.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.1.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.1.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.1.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.1.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.1.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.2.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.2.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.2.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.2.self_attn.o_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.2.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.2.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.2.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.2.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.2.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.2.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.2.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.3.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.3.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.3.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.3.self_attn.o_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.3.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.3.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.3.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.3.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.3.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.3.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.3.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.4.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.4.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.4.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.4.self_attn.o_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.4.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.4.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.4.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.4.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.4.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.4.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.4.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.5.self_attn.q_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.5.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.5.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.5.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.5.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.5.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.5.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.5.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.5.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.5.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.6.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.6.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.6.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.6.self_attn.o_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.6.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.6.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.6.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.6.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.6.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.6.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.6.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.7.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.7.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.7.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.7.self_attn.o_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.7.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.7.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.7.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.7.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.7.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.7.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.7.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.8.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.8.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.8.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.8.self_attn.o_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.8.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.8.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.8.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.8.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.8.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.8.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.8.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.9.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.9.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.9.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.9.self_attn.o_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.9.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.9.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.9.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.9.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.9.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.9.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.9.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.10.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.10.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.10.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.10.self_attn.o_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.10.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.10.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.10.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.10.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.10.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.10.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.10.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.11.self_attn.q_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.11.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.11.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.11.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.11.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.11.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.11.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.11.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.11.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.11.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.12.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.12.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.12.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.12.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.12.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.12.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.12.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.12.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.12.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.12.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.12.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.13.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.13.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.13.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.13.self_attn.o_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.13.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.13.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.13.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.13.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.13.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.13.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.13.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.14.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.14.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.14.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.14.self_attn.o_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.14.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.14.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.14.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.14.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.14.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.14.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.14.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.15.self_attn.q_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.15.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.15.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.15.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.15.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.15.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.15.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.15.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.15.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.15.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.15.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.16.self_attn.q_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.16.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.16.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.16.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.16.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.16.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.16.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.16.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.16.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.16.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.16.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.17.self_attn.q_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.17.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.17.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.17.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.17.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.17.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.17.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.17.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.17.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.17.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.18.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.18.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.18.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.18.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.18.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.18.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.18.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.18.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.18.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.18.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.18.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.19.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.19.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.19.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.19.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.19.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.19.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.19.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.19.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.19.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.19.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.19.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.20.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.20.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.20.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.20.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.20.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.20.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.20.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.20.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.20.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.20.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.20.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.21.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.21.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.21.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.21.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.21.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.21.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.21.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.21.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.21.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.21.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.21.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.22.self_attn.q_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.22.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.22.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.22.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.22.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.22.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.22.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.22.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.22.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.22.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.22.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.23.self_attn.q_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.23.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.23.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.23.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.23.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.23.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.23.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.23.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.23.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.23.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.24.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.24.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.24.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.24.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.24.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.24.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.24.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.24.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.24.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.24.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.24.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.25.self_attn.q_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.25.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.25.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.25.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.25.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.25.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.25.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.25.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.25.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.25.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.25.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.26.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.26.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.26.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.26.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.26.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.26.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.26.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.26.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.26.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.26.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.26.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.27.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.27.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.27.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.27.self_attn.o_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.27.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.27.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.27.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.27.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.27.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.27.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.27.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.28.self_attn.q_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.28.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.28.self_attn.v_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.28.self_attn.o_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.28.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.28.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.28.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.28.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.28.experts.switch_glu.gate_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.28.experts.switch_glu.up_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.28.experts.switch_glu.down_proj": {
+            "bits": 4,
+            "group_size": 64
+        },
+        "language_model.model.layers.29.self_attn.q_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.29.self_attn.k_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.29.self_attn.o_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.29.mlp.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.29.mlp.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.29.mlp.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.29.router.proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.29.experts.switch_glu.gate_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.29.experts.switch_glu.up_proj": {
+            "bits": 8,
+            "group_size": 64
+        },
+        "language_model.model.layers.29.experts.switch_glu.down_proj": {
+            "bits": 8,
+            "group_size": 64
+        }
+    },
+    "text_config": {
+        "attention_bias": false,
+        "attention_dropout": 0.0,
+        "attention_k_eq_v": true,
+        "bos_token_id": 2,
+        "dtype": "bfloat16",
+        "enable_moe_block": true,
+        "eos_token_id": 1,
+        "final_logit_softcapping": 30.0,
+        "global_head_dim": 512,
+        "head_dim": 256,
+        "hidden_activation": "gelu_pytorch_tanh",
+        "hidden_size": 2816,
+        "hidden_size_per_layer_input": 0,
+        "initializer_range": 0.02,
+        "intermediate_size": 2112,
+        "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"
+        ],
+        "max_position_embeddings": 262144,
+        "model_type": "gemma4_text",
+        "moe_intermediate_size": 704,
+        "num_attention_heads": 16,
+        "num_experts": 128,
+        "num_global_key_value_heads": 2,
+        "num_hidden_layers": 30,
+        "num_key_value_heads": 8,
+        "num_kv_shared_layers": 0,
+        "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": 1024,
+        "tie_word_embeddings": true,
+        "top_k_experts": 8,
+        "use_bidirectional_attention": "vision",
+        "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.dev0"
+}

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.dev0"
+}

model-00001-of-00003.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72bf520a507f6b5405e9912bf587852d536a38dc088770ed23730cb120f492b8
+size 5340564888

model-00002-of-00003.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:696860ef5a3a6f7e2e2b1eb051abec3da166653de3e147c5cc4e25ee09975907
+size 5266440089

model-00003-of-00003.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:993f4534c438a90da781c8114517ec47e9be4a044867ffc6b17bd88ca2d35511
+size 4981132643

tokenizer.json

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

tokenizer_config.json

@@ -0,0 +1,97 @@
+{
+  "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": true,
+  "local_files_only": 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",
+  "tool_parser_type": "gemma4",
+  "unk_token": "<unk>"
+}