Add files using upload-large-folder tool

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

README.md

@@ -1,3 +1,233 @@
----
-license: apache-2.0
----
+---
+base_model: crane-ai-labs/ganda-gemma-1b
+language:
+- en
+- lg
+library_name: transformers
+license: gemma
+tags:
+- luganda
+- translation
+- conversational
+- unsloth
+- gemma
+- gemma3
+- gguf
+- quantized
+pipeline_tag: text-generation
+---
+
+# Ganda Gemma 1B - GGUF
+
+Quantized GGUF versions of **Ganda Gemma 1B**, a fine-tuned Gemma 3 1B instruction model specialized for **English-to-Luganda translation and Luganda conversational AI**. The model accepts input in both English and Luganda but outputs responses exclusively in Luganda.
+
+## 🚀 Quick Start
+
+```bash
+# Download the recommended Q4_K_M quantization
+pip install huggingface_hub
+
+# Python download
+from huggingface_hub import snapshot_download
+snapshot_download(
+    repo_id="crane-ai-labs/ganda-gemma-1b-GGUF",
+    local_dir="ganda-gemma-1b-GGUF",
+    allow_patterns=["Q4_K_M/*"]  # Download only Q4_K_M version
+)
+```
+
+## 📊 Available Quantizations
+
+| Quantization | Folder | File Size | Quality | Use Case |
+|-------------|--------|-----------|---------|----------|
+| `F32` | F32/ | ~2.4GB | Highest | Research & benchmarking |
+| `F16` | F16/ | ~1.2GB | Highest | Maximum quality inference |
+| `Q8_0` | Q8_0/ | ~600MB | Very High | Production with ample resources |
+| `Q5_K_M` | Q5_K_M/ | ~500MB | High | Balanced quality/size |
+| `Q4_K_M` | Q4_K_M/ | ~400MB | Good | **Recommended** for most users |
+| `Q4_K_S` | Q4_K_S/ | ~350MB | Good | Resource-constrained environments |
+| `Q3_K_M` | Q3_K_M/ | ~300MB | Fair | Mobile/edge deployment |
+| `Q2_K` | Q2_K/ | ~250MB | Lower | Minimal resource usage |
+
+## 🌍 Model Capabilities
+
+- **Input Languages**: English + Luganda
+- **Output Language**: Luganda only
+- **Specializations**:
+  - English-to-Luganda translation
+  - Luganda conversational AI
+  - Text summarization in Luganda
+  - Question answering in Luganda
+
+## 💻 Usage with llama.cpp
+
+### Download and Setup
+
+```bash
+# Install llama.cpp
+git clone https://github.com/ggerganov/llama.cpp
+cd llama.cpp
+make
+
+# Download specific quantization
+from huggingface_hub import snapshot_download
+snapshot_download(
+    repo_id="crane-ai-labs/ganda-gemma-1b-GGUF",
+    local_dir="ganda-gemma-1b-GGUF",
+    allow_patterns=["Q4_K_M/*"]
+)
+```
+
+### Basic Translation
+
+```bash
+# English to Luganda translation
+./llama-cli \
+  --model ganda-gemma-1b-GGUF/Q4_K_M/ganda-gemma-1b-q4_k_m.gguf \
+  --prompt "Translate to Luganda: Hello, how are you today?" \
+  --temp 0.3 \
+  --top-p 0.95 \
+  --top-k 64 \
+  --repeat-penalty 1.1 \
+  -n 128
+```
+
+### Luganda Conversation
+
+```bash
+# Chat in Luganda
+./llama-cli \
+  --model ganda-gemma-1b-GGUF/Q4_K_M/ganda-gemma-1b-q4_k_m.gguf \
+  --prompt "Mwasuze mutya! Nnyinza ntya okukuyamba leero?" \
+  --temp 0.3 \
+  --interactive
+```
+
+## 🔧 Usage with Ollama
+
+### Create Modelfile
+
+```dockerfile
+# Save as Modelfile
+FROM ganda-gemma-1b-GGUF/Q4_K_M/ganda-gemma-1b-q4_k_m.gguf
+
+TEMPLATE """<start_of_turn>user
+{{ .Prompt }}<end_of_turn>
+<start_of_turn>model
+"""
+
+PARAMETER temperature 0.3
+PARAMETER top_p 0.95
+PARAMETER top_k 64
+PARAMETER repeat_penalty 1.1
+PARAMETER num_ctx 4096
+
+SYSTEM """You are a helpful AI assistant that specializes in English-to-Luganda translation and Luganda conversation. Always respond in Luganda."""
+```
+
+### Run with Ollama
+
+```bash
+# Create model
+ollama create ganda-gemma-1b -f Modelfile
+
+# Use for translation
+ollama run ganda-gemma-1b "Translate to Luganda: Good morning, how did you sleep?"
+
+# Use for conversation
+ollama run ganda-gemma-1b "Nkwagala nnyo okuyiga Oluganda. Oyinza okumperereza?"
+```
+
+## 📋 Download Specific Quantizations
+
+```python
+from huggingface_hub import snapshot_download
+
+# Download only F16 (highest quality)
+snapshot_download(
+    repo_id="crane-ai-labs/ganda-gemma-1b-GGUF",
+    local_dir="ganda-gemma-1b-GGUF",
+    allow_patterns=["F16/*"]
+)
+
+# Download only Q2_K (smallest size)
+snapshot_download(
+    repo_id="crane-ai-labs/ganda-gemma-1b-GGUF", 
+    local_dir="ganda-gemma-1b-GGUF",
+    allow_patterns=["Q2_K/*"]
+)
+
+# Download multiple quantizations
+snapshot_download(
+    repo_id="crane-ai-labs/ganda-gemma-1b-GGUF",
+    local_dir="ganda-gemma-1b-GGUF", 
+    allow_patterns=["Q4_K_M/*", "Q8_0/*"]
+)
+```
+
+## 🎯 Performance Metrics
+
+**Translation Quality (BLEU Scores)**
+- **Ganda Gemma 1B**: 6.98 BLEU, 40.63 chrF++
+- **ChatGPT-4o-latest**: 6.46 BLEU, 39.98 chrF++
+- **Llama-4-Maverick**: 4.29 BLEU, 33.52 chrF++
+
+*Evaluated on 1,012 English-to-Luganda translation samples*
+
+## 🎨 Use Cases
+
+- **Language Learning**: Practice English-Luganda translation
+- **Cultural Preservation**: Create and document Luganda content
+- **Educational Tools**: Luganda learning assistants
+- **Content Localization**: Translate materials to Luganda
+- **Conversational Practice**: Improve Luganda dialogue skills
+- **Text Summarization**: Summarize content in Luganda
+
+## 💡 Generation Parameters
+
+Recommended settings for optimal results:
+
+```yaml
+temperature: 0.3      # Focused, coherent responses
+top_p: 0.95          # Nucleus sampling
+top_k: 64            # Top-k sampling
+max_tokens: 128      # Response length limit
+repeat_penalty: 1.1  # Reduces repetition
+```
+
+## 🔗 Related Models
+
+- **Main Model**: [crane-ai-labs/ganda-gemma-1b](https://huggingface.co/crane-ai-labs/ganda-gemma-1b) - Original PyTorch model
+- **Collection**: [Ganda Gemma 1B](https://huggingface.co/collections/crane-ai-labs/ganda-gemma-1b) - All model variants
+
+## ⚠️ Limitations
+
+- **Language Output**: Responds only in Luganda
+- **Factual Knowledge**: General knowledge only, not trained on specific factual datasets
+- **No Coding/Math**: Not designed for programming or mathematical tasks
+- **Context Length**: Limited to 4,096 tokens for optimal performance
+- **Specialized Domains**: May require domain-specific fine-tuning
+
+## 🤝 Training Details
+
+- **Base Model**: Gemma 3 1B Instruction Tuned
+- **Training Method**: GRPO (Generalized Reward Preference Optimization)
+- **Context Length**: 4,096 tokens
+- **Architecture**: Transformer with sliding window attention
+- **Tokenizer**: SentencePiece with 262K vocabulary
+
+## 📄 License
+
+This model is released under the [Gemma Terms of Use](https://ai.google.dev/gemma/terms). Please review the terms before use.
+
+## 🙏 Acknowledgments
+
+- **Google**: For the Gemma 3 base model
+- **Unsloth**: For quantization techniques and methodology
+- **Community**: For Luganda language resources and datasets
+
+---
+
+**Built with ❤️ by Crane AI Labs**
+
+*Ganda Gemma - Your helpful Luganda AI companion*

added_tokens.json

@@ -0,0 +1,3 @@
+{
+  "<image_soft_token>": 262144
+}

chat_template.jinja

@@ -0,0 +1,47 @@
+{{ bos_token }}
+{%- if messages[0]['role'] == 'system' -%}
+    {%- if messages[0]['content'] is string -%}
+        {%- set first_user_prefix = messages[0]['content'] + '
+
+' -%}
+    {%- else -%}
+        {%- set first_user_prefix = messages[0]['content'][0]['text'] + '
+
+' -%}
+    {%- endif -%}
+    {%- set loop_messages = messages[1:] -%}
+{%- else -%}
+    {%- set first_user_prefix = "" -%}
+    {%- set loop_messages = messages -%}
+{%- endif -%}
+{%- for message in loop_messages -%}
+    {%- if (message['role'] == 'user') != (loop.index0 % 2 == 0) -%}
+        {{ raise_exception("Conversation roles must alternate user/assistant/user/assistant/...") }}
+    {%- endif -%}
+    {%- if (message['role'] == 'assistant') -%}
+        {%- set role = "model" -%}
+    {%- else -%}
+        {%- set role = message['role'] -%}
+    {%- endif -%}
+    {{ '<start_of_turn>' + role + '
+' + (first_user_prefix if loop.first else "") }}
+    {%- if message['content'] is string -%}
+        {{ message['content'] | trim }}
+    {%- elif message['content'] is iterable -%}
+        {%- for item in message['content'] -%}
+            {%- if item['type'] == 'image' -%}
+                {{ '<start_of_image>' }}
+            {%- elif item['type'] == 'text' -%}
+                {{ item['text'] | trim }}
+            {%- endif -%}
+        {%- endfor -%}
+    {%- else -%}
+        {{ raise_exception("Invalid content type") }}
+    {%- endif -%}
+    {{ '<end_of_turn>
+' }}
+{%- endfor -%}
+{%- if add_generation_prompt -%}
+    {{ '<start_of_turn>model
+' }}
+{%- endif -%}

config.json

@@ -0,0 +1,64 @@
+{
+  "_sliding_window_pattern": 6,
+  "architectures": [
+    "Gemma3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_logit_softcapping": null,
+  "bos_token_id": 2,
+  "cache_implementation": "hybrid",
+  "eos_token_id": 106,
+  "final_logit_softcapping": null,
+  "head_dim": 256,
+  "hidden_activation": "gelu_pytorch_tanh",
+  "hidden_size": 1152,
+  "initializer_range": 0.02,
+  "intermediate_size": 6912,
+  "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"
+  ],
+  "max_position_embeddings": 32768,
+  "model_type": "gemma3_text",
+  "num_attention_heads": 4,
+  "num_hidden_layers": 26,
+  "num_key_value_heads": 1,
+  "pad_token_id": 0,
+  "query_pre_attn_scalar": 256,
+  "rms_norm_eps": 1e-06,
+  "rope_local_base_freq": 10000,
+  "rope_scaling": null,
+  "rope_theta": 1000000,
+  "sliding_window": 512,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.54.1",
+  "unsloth_fixed": true,
+  "unsloth_version": "2025.7.11",
+  "use_cache": true,
+  "vocab_size": 262144
+}

generation_config.json

@@ -0,0 +1,14 @@
+{
+  "bos_token_id": 2,
+  "cache_implementation": "hybrid",
+  "do_sample": true,
+  "eos_token_id": [
+    1,
+    106
+  ],
+  "max_length": 32768,
+  "pad_token_id": 0,
+  "top_k": 64,
+  "top_p": 0.95,
+  "transformers_version": "4.54.1"
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c23f49045a133162ef10793f5ea7f6eadc0d2802c66ac53f9b191f117481d3a9
+size 1999811208

special_tokens_map.json

@@ -0,0 +1,33 @@
+{
+  "boi_token": "<start_of_image>",
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eoi_token": "<end_of_image>",
+  "eos_token": {
+    "content": "<end_of_turn>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "image_token": "<image_soft_token>",
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d786405177734910d7a3db625c2826640964a0b4e5cdbbd70620ae3313a01bef
+size 33384722

tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1299c11d7cf632ef3b4e11937501358ada021bbdf7c47638d13c0ee982f2e79c
+size 4689074