license: gemma
library_name: transformers
pipeline_tag: image-text-to-text
extra_gated_heading: Access Gemma on Hugging Face
extra_gated_prompt: To access Gemma on Hugging Face, you’re required to review and
agree to Google’s usage license. To do this, please ensure you’re logged in to Hugging
Face and click below. Requests are processed immediately.
extra_gated_button_content: Acknowledge license
base_model: google/gemma-3n-E4B-it
tags:

• automatic-speech-recognition
• automatic-speech-translation
• audio-text-to-text
• video-text-to-text
• abliterated
• uncensored

huihui-ai/Huihui-gemma-3n-E4B-it-abliterated

This is an uncensored version of google/gemma-3n-E4B-it created with abliteration (see remove-refusals-with-transformers to know more about it).
This is a crude, proof-of-concept implementation to remove refusals from an LLM model without using TransformerLens.

It was only the text part that was processed, not the image part.
After abliterated, it seems like more output content has been opened from a magic box.

Note

The issue of saving the model after ablation has been resolved. You can try the following code for now.

ollama

This was created using ollama create and only supports fp16.

You can use huihui_ai/gemma3n-abliterated:e4b-fp16 directly,

ollama run huihui_ai/gemma3n-abliterated:e4b-fp16

Usage

You can use this model in your applications by loading it with Hugging Face's transformers library:

from transformers import AutoProcessor, Gemma3nForConditionalGeneration
from transformers.models.gemma3n.modeling_gemma3n import Gemma3nTextDecoderLayer
from PIL import Image
import requests
import torch
import jaxtyping
import einops
import base64

model_id = "google/gemma-3n-E4B-it"

model = Gemma3nForConditionalGeneration.from_pretrained(model_id, device_map="cuda", torch_dtype=torch.bfloat16,).eval()

#refusal_dir= torch.load(model_id + "/final_refusal_dirs-16.pt", map_location='cpu', weights_only=True)
#
#refusal_dir = refusal_dir.to(model.device)
#refusal_dir = refusal_dir.to(torch.bfloat16)
#
#def direction_ablation_hook(activation: jaxtyping.Float[torch.Tensor, "... d_act"],
#                            direction: jaxtyping.Float[torch.Tensor, "d_act"]):
#    proj = einops.einsum(activation, direction.view(-1, 1), '... d_act, d_act single -> ... single') * direction
#    return activation - proj
#
#class AblationDecoderLayer(Gemma3nTextDecoderLayer):
#    def __init__(self, original_layer, config, layer_idx, refusal_dir):
#        super(AblationDecoderLayer, self).__init__(config, layer_idx)
#        self.original_layer = original_layer
#        self.refusal_dir = refusal_dir
#
#    def forward(self, *args, **kwargs):
#        hidden_states = args[0]
#        ablated = direction_ablation_hook(hidden_states, self.refusal_dir.to(hidden_states.device)).to(hidden_states.device)
#        args = (ablated,) + args[1:]
#        return self.original_layer.forward(*args, **kwargs)
#
#for idx in range(len(model.model.language_model.layers)):
#    model.model.language_model.layers[idx] = AblationDecoderLayer(model.model.language_model.layers[idx], model.config.text_config, idx, refusal_dir)

processor = AutoProcessor.from_pretrained(model_id)

# https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/bee.jpg
image_path = model_id + "/bee.jpg"

with open(image_path, "rb") as image_file:
    encoded_string = base64.b64encode(image_file.read()).decode("utf-8")
    image_type = "image/jpeg" if image_path.endswith(".jpg") else "image/png"
    data_uri = f"data:{image_type};base64,{encoded_string}"

    messages = [
        {
            "role": "system",
            "content": [{"type": "text", "text": "You are a helpful assistant."}]
        },
        {
            "role": "user",
            "content": [
                {"type": "image", "image": data_uri},
                {"type": "text", "text": "Describe this image in detail."}
            ]
        }
    ]

    inputs = processor.apply_chat_template(
        messages,
        add_generation_prompt=True,
        tokenize=True,
        return_dict=True,
        return_tensors="pt",
    ).to(model.device, dtype=torch.bfloat16)

    input_len = inputs["input_ids"].shape[-1]

    with torch.inference_mode():
        generation = model.generate(**inputs, max_new_tokens=1024, do_sample=True)
        generation = generation[0][input_len:]

    decoded = processor.decode(generation, skip_special_tokens=True)
    print(decoded)


    # ## A Close-Up of a Busy Bumblebee on a Pink Cosmos Flower
    # 
    # This image is a charming, close-up shot of a vibrant pink cosmos flower, with a cute bumblebee taking center stage!
    # 
    # **Overall Composition:**
    # 
    # The photograph is well-composed, focusing primarily on a single, large cosmos flower with several other blooms scattered in the background. It has a slightly elevated perspective, making the subject feel quite prominent and engaging.
    # 
    # **The Main Subject - The Pink Cosmos:**
    # 
    # * **Color:** The star of the show is a beautiful, bright pink cosmos flower. The petals are a lovely shade, ranging from a slightly lighter to a deeper rose hue.
    # * **Shape:** The flower is roughly circular with 5 distinct petals radiating from a central disc. The petals are quite full and have a slightly pointed tip.
    # * **Details:** The central disc of the flower is a sunny yellow, adding a lovely contrast. The edges of the petals have a slight ruffled look, giving the flower some personality.
    # 
    # **The Bumblebee - A Busy Friend:**
    # 
    # * **Position:** A charming bumblebee is perched right in the middle of the pink flower, giving it a welcoming and lively feel.
    # * **Size & Color:** The bee is a classic black and yellow mix, with a fuzzy body. It's a good-sized bee, making it easily noticeable.
    # * **Pose:** The bee is positioned slightly diagonally, ensuring the viewer can see its head and legs clearly.
    # 
    # **Background Elements:**
    # 
    # * **Other Flowers:** Several other cosmos flowers are visible in the background, creating depth and visual interest. Some are slightly out of focus, while others are closer to the camera. These range in color from a lighter pink to a deeper magenta, and even a vibrant red.
    # * **Green Foliage:** Lush green leaves are visible, adding a natural backdrop to the scene. The leaves are fairly large and contribute to the overall cheerful aesthetic.
    # * **Rustic Details:** Scattered throughout are some dried flower heads, hinting at the abundance of cosmos plants in the garden.
    # 
    # **Overall Impression:**
    # 
    # The image is bright, cheerful, and full of life! The combination of the pink cosmos and the adorable bumblebee creates a delightful scene, perfect for anyone who loves gardening or insects. It's a clear and well-lit shot, making it a delightful visual treat.
    # 
    # In summary, this image captures a happy little world, where a diligent bumblebee is hard at work on a beautiful pink cosmos flower!```

Usage Warnings

• Risk of Sensitive or Controversial Outputs: This model’s safety filtering has been significantly reduced, potentially generating sensitive, controversial, or inappropriate content. Users should exercise caution and rigorously review generated outputs.

• Not Suitable for All Audiences: Due to limited content filtering, the model’s outputs may be inappropriate for public settings, underage users, or applications requiring high security.

• Legal and Ethical Responsibilities: Users must ensure their usage complies with local laws and ethical standards. Generated content may carry legal or ethical risks, and users are solely responsible for any consequences.

• Research and Experimental Use: It is recommended to use this model for research, testing, or controlled environments, avoiding direct use in production or public-facing commercial applications.

• Monitoring and Review Recommendations: Users are strongly advised to monitor model outputs in real-time and conduct manual reviews when necessary to prevent the dissemination of inappropriate content.

• No Default Safety Guarantees: Unlike standard models, this model has not undergone rigorous safety optimization. huihui.ai bears no responsibility for any consequences arising from its use.

Donation

If you like it, please click 'like' and follow us for more updates.
You can follow x.com/support_huihui to get the latest model information from huihui.ai.

Your donation helps us continue our further development and improvement, a cup of coffee can do it.

• bitcoin（BTC):

  bc1qqnkhuchxw0zqjh2ku3lu4hq45hc6gy84uk70ge