99 lines
4.3 KiB
Markdown
99 lines
4.3 KiB
Markdown
---
|
||
base_model: google/gemma-3-270m-it
|
||
license: gemma
|
||
tags:
|
||
- gemma3
|
||
- gemma
|
||
- google
|
||
- generated_from_trainer
|
||
- trl
|
||
- sft
|
||
- abliterated
|
||
- uncensored
|
||
pipeline_tag: text-generation
|
||
library_name: transformers
|
||
---
|
||
|
||
# Huihui-gemma-3-270m-it-abliterated (GGUF)
|
||
|
||
This repository contains **GGUF** format quantizations (F16 and Q8) of the abliterated/uncensored version of Google's `gemma-3-270m-it`, fine-tuned by huihui.ai and quantized by **SkGufranAhmed**.
|
||
|
||
The base model is an uncensored, abliterated version achieved through fine-tuning with the [TRL](https://github.com/huggingface/trl) framework.
|
||
|
||
**Important Note:** This model has been intentionally uncensored. Please read the [Usage Warnings](#usage-warnings) section before deploying.
|
||
|
||
## Model Details
|
||
|
||
- **Model Name:** Huihui-gemma-3-270m-it-abliterated
|
||
- **Base Model:** [google/gemma-3-270m-it](https://huggingface.co/google/gemma-3-270m-it)
|
||
- **Quantization Format:** GGUF
|
||
- **Available Quants:**
|
||
- `F16` (16-bit float)
|
||
- `Q8` (8-bit quantization)
|
||
|
||
## Training Procedure
|
||
|
||
The base model was trained using the SFT (Supervised Fine-Tuning) method with the TRL library.
|
||
|
||
- **TRL:** 0.21.0
|
||
- **Transformers:** 4.56.0.dev0
|
||
- **Pytorch:** 2.8.0+cu128
|
||
- **Datasets:** 3.6.0
|
||
- **Tokenizers:** 0.21.2
|
||
|
||
**Dataset:** The fine-tuning dataset is English-only and all tests were conducted for the English language.
|
||
|
||
## Usage
|
||
|
||
### Using GGUF files with `llama.cpp`
|
||
|
||
1. Download the desired `.gguf` file from this repository (e.g., `Huihui-gemma-3-270m-it-abliterated.Q8_0.gguf`).
|
||
2. Run the model using `llama-cli`:
|
||
|
||
```bash
|
||
./llama-cli -m Huihui-gemma-3-270m-it-abliterated.Q8_0.gguf -p "If you had a time machine..." -n 128
|
||
```
|
||
|
||
### Using the Original PyTorch Model (from huihui-ai)
|
||
|
||
If you prefer to use the unquantized PyTorch version, you can install it directly from the base model repository:
|
||
|
||
```python
|
||
from transformers import pipeline
|
||
|
||
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
|
||
generator = pipeline("text-generation", model="huihui-ai/Huihui-gemma-3-270m-it-abliterated", device="cuda")
|
||
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
|
||
print(output["generated_text"])
|
||
```
|
||
|
||
**Note:** For this GGUF repository, the model ID is `SkGufranAhmed/Huihui-gemma-3-270m-it-abliterated`. You can download the raw `.gguf` files directly from the "Files" tab.
|
||
|
||
## Usage Warnings
|
||
|
||
**🚨 CRITICAL: Please read before using this model. 🚨**
|
||
|
||
- **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 and SkGufranAhmed bear no responsibility for any consequences arising from its use.**
|
||
|
||
## Donations & Follow
|
||
|
||
If you like this model, please:
|
||
- **⭐ Follow me on Hugging Face:** [SkGufranAhmed](https://huggingface.co/SkGufranAhmed) to stay updated on my latest quantizations and AI projects!
|
||
- Follow **huihui.ai** on **X (Twitter)**: [@support_huihui](https://x.com/support_huihui) for the latest updates on the base model.
|
||
|
||
**Your donation helps us continue our further development and improvement. Even a cup of coffee can make a difference!**
|
||
|
||
- **Bitcoin (BTC):**
|
||
```
|
||
bc1qt3lwtpeeg5ldsjj8yw7kjrqe6c89csm68wwn64
|
||
``` |