Instructions to use JoshBrew/TagaLLM-Gemma-4-E2B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use JoshBrew/TagaLLM-Gemma-4-E2B with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf JoshBrew/TagaLLM-Gemma-4-E2B:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf JoshBrew/TagaLLM-Gemma-4-E2B:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf JoshBrew/TagaLLM-Gemma-4-E2B:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf JoshBrew/TagaLLM-Gemma-4-E2B:Q4_K_M

Build from source code

git clone https://github.com/ggml-org/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-app
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama serve -hf JoshBrew/TagaLLM-Gemma-4-E2B:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama cli -hf JoshBrew/TagaLLM-Gemma-4-E2B:Q4_K_M

Use Docker

docker model run hf.co/JoshBrew/TagaLLM-Gemma-4-E2B:Q4_K_M

Llama new
LM Studio
Jan
Ollama
How to use JoshBrew/TagaLLM-Gemma-4-E2B with Ollama:
```
ollama run hf.co/JoshBrew/TagaLLM-Gemma-4-E2B:Q4_K_M
```
Unsloth Desktop

How to use JoshBrew/TagaLLM-Gemma-4-E2B with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf JoshBrew/TagaLLM-Gemma-4-E2B:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "JoshBrew/TagaLLM-Gemma-4-E2B:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use JoshBrew/TagaLLM-Gemma-4-E2B with Docker Model Runner:
```
docker model run hf.co/JoshBrew/TagaLLM-Gemma-4-E2B:Q4_K_M
```

Lemonade

How to use JoshBrew/TagaLLM-Gemma-4-E2B with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull JoshBrew/TagaLLM-Gemma-4-E2B:Q4_K_M

Run and chat with the model

lemonade run user.TagaLLM-Gemma-4-E2B-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use JoshBrew/TagaLLM-Gemma-4-E2B with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf JoshBrew/TagaLLM-Gemma-4-E2B:Q4_K_M

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default JoshBrew/TagaLLM-Gemma-4-E2B:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use JoshBrew/TagaLLM-Gemma-4-E2B with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf JoshBrew/TagaLLM-Gemma-4-E2B:Q4_K_M

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "JoshBrew/TagaLLM-Gemma-4-E2B:Q4_K_M" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

TagaLLM Gemma 4 E2B

The evaluated chatbot uses the Google Gemma 4 E2B instruction checkpoint at revision 4adf15748de9e1a0097b6274cee9c15836c41c37. The checkpoint was converted to BF16 GGUF and quantized to Q4_K_M with llama.cpp. Its SHA-256 is 1b71213c391f309cd3775c2fb489591d5806d2f2e59eccf6400d2b96f42fed47. The file is 3,427,874,880 bytes.

The evaluated configuration uses the quantized base weights, a Tagalog system prompt, curated cultural references, and separate language and scope routing. The pilot adapters are separate experiments. The GGUF embeds the tokenizer. A compatible llama.cpp build serves the model with a 4,096-token context in the app.

In the 20-conversation unblinded development comparison, the full app received full marks on naturalness for 110 of 120 replies and on state tracking for 37 of 72 checks. Three conversations passed all six exchanges. These are app-level results. The model can miss updates to people, places and times. The two inference processes together used about 3.8 GiB of resident RAM in a local Windows sample. The two checkpoints require about 6.9 GB of storage.

During TagaLLM development, a native Tagalog speaker helped review selected model replies for grammar and natural phrasing, with AI assistance.

See the upstream model card and local setup guide for dependencies and inference commands. The Apache 2.0 license is included in this package. The upstream configuration and chat template are provided for reference. The GGUF conversion and quantization are TagaLLM project modifications.