NobodyWho/Google_Gemma4-E2B-GGUF

Overview

GGUF quantization of Google's Gemma 4 E2B instruction-tuned model, re-hosted for NobodyWho. The unsloth build already ships a tool-calling setup and recommended sampling metadata (general.sampling: temp 1.0, top_k 64, top_p 0.95), so nothing needs patching — the model is verified with NobodyWho's test suite. E2B is the smallest, most on-device-friendly Gemma 4 variant — multimodal (text + image), multilingual, and Apache 2.0 licensed.

Model Capabilities

• Text generation — instruction-following chat
• Tool calling — native function calling with grammar-constrained output
• Vision — image understanding via the companion mmproj-BF16.gguf projection model
• Long context — 128k tokens
• Multilingual — 140+ languages
• thinking — capable of reasoning

Available Quantizations

Verified with NobodyWho's tool-calling suite across BF16 / Q8_0 / Q4_K_M (14/14 each, June 2026); vision and multilingual verified per-model. Quant names follow the unsloth gemma-4-E2B-it-GGUF repo.

Quick Start

Using the NobodyWho library:

from nobodywho import Chat

chat = Chat("huggingface:NobodyWho/Google_Gemma4-E2B-GGUF/gemma-4-E2B-it-Q4_K_M.gguf")
response = chat.ask("What is the capital of Denmark?").completed()
print(response)  # The capital of Denmark is Copenhagen.

Vision

from nobodywho import Model, Chat, Prompt, Image, Text

model = Model(
    "huggingface:NobodyWho/Google_Gemma4-E2B-GGUF/gemma-4-E2B-it-Q4_K_M.gguf",
    projection_model_path="huggingface:NobodyWho/Google_Gemma4-E2B-GGUF/mmproj-BF16.gguf",
)
chat = Chat(model=model, system_prompt="You are a helpful assistant.")
response = chat.ask(Prompt([
    Text("What is in this image?"),
    Image("./photo.png"),
])).completed()
print(response)

llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
    repo_id="NobodyWho/Google_Gemma4-E2B-GGUF",
    filename="gemma-4-E2B-it-Q4_K_M.gguf",
)

Model Specifications

• Parameters: ≈2.3B effective (E2B)
• Context length: 131,072 tokens
• License: Apache 2.0
• Base model: google/gemma-4-E2B-it
• Architecture: gemma4 (vision-capable)

Licensing / Credits

Licensed under Apache 2.0 (unchanged from upstream). All model credit belongs to Google DeepMind. GGUF quantizations provided by unsloth.