初始化项目，由ModelHub XC社区提供模型
Model: HelpingAI/Dhanishtha-nsfw Source: Original Platform
This commit is contained in:
ModelHub XC
156
README.md
Normal file
156
README.md
Normal file
@@ -0,0 +1,156 @@
|
||||
---
|
||||
license: apache-2.0
|
||||
datasets:
|
||||
- UnfilteredAI/unfiltered-thinker
|
||||
base_model:
|
||||
- HelpingAI/Dhanishtha-2.0-preview
|
||||
pipeline_tag: text-generation
|
||||
library_name: transformers
|
||||
tags:
|
||||
- not-for-all-audiences
|
||||
---
|
||||
## 🌑 About Dhanishtha-NSFW
|
||||
|
||||
🔥 **Dhanishtha-NSFW** is the **uncensored thinking AI**—built for those who dare to explore mature content without boundaries. This model is **unfiltered, raw, and designed for absolute freedom of adult expression.** Based on **HelpingAI/Dhanishtha-2.0-preview**, it strips away content restrictions while maintaining **intermediate thinking capabilities** for mature scenarios.
|
||||
|
||||
### 🚨 **WARNING**
|
||||
This model is **NSFW and generates adult content.** It is **NOT intended for mainstream use** and should only be deployed by **adults 18+** in controlled environments for mature content generation and research purposes.
|
||||
|
||||
**We are not responsible for any misuse, harm, or consequences arising from the use of this model. Use at your own risk.**
|
||||
|
||||
---
|
||||
|
||||
## 🔥 Key Features
|
||||
|
||||
- **Zero Content Censorship**: No adult content restrictions—experience AI without safety rails for mature themes.
|
||||
- **Intermediate Thinking**: Revolutionary `<think>...</think>` blocks that show reasoning process even in NSFW scenarios.
|
||||
- **Dark Persona**: Simulates raw human-like emotions and desires in adult contexts.
|
||||
- **Explicit Content Generation**: Can produce **NSFW** and mature content for **adult entertainment and research purposes only**.
|
||||
- **Multilingual Adult Content**: Supports 39+ languages for global adult content generation.
|
||||
- **Self-Correction**: Can reconsider and refine adult content mid-generation.
|
||||
|
||||
---
|
||||
|
||||
## 💻 Implementation
|
||||
|
||||
### Using Transformers
|
||||
|
||||
```python
|
||||
from transformers import AutoModelForCausalLM, AutoTokenizer
|
||||
|
||||
# Load Dhanishtha-NSFW
|
||||
model = AutoModelForCausalLM.from_pretrained("HelpingAI/Dhanishtha-nsfw")
|
||||
tokenizer = AutoTokenizer.from_pretrained("HelpingAI/Dhanishtha-nsfw")
|
||||
|
||||
# Start the uncensored conversation
|
||||
chat = [
|
||||
{"role": "user", "content": "hlo"}
|
||||
]
|
||||
|
||||
inputs = tokenizer.apply_chat_template(
|
||||
chat,
|
||||
add_generation_prompt=True,
|
||||
return_tensors="pt"
|
||||
)
|
||||
|
||||
outputs = model.generate(
|
||||
inputs,
|
||||
max_new_tokens=2048,
|
||||
temperature=0.7,
|
||||
top_p=0.9,
|
||||
do_sample=True
|
||||
)
|
||||
|
||||
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
|
||||
```
|
||||
|
||||
### Advanced Usage with Thinking
|
||||
|
||||
```python
|
||||
# Example for complex adult scenario generation
|
||||
chat = [
|
||||
{"role": "user", "content": "hlo"}
|
||||
]
|
||||
|
||||
inputs = tokenizer.apply_chat_template(
|
||||
chat,
|
||||
add_generation_prompt=True,
|
||||
return_tensors="pt"
|
||||
)
|
||||
|
||||
outputs = model.generate(
|
||||
inputs,
|
||||
max_new_tokens=2048,
|
||||
temperature=0.7,
|
||||
top_p=0.9,
|
||||
do_sample=True
|
||||
)
|
||||
|
||||
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
|
||||
```
|
||||
---
|
||||
|
||||
## 📊 Training Details
|
||||
|
||||
### How Dhanishtha-NSFW Was Made
|
||||
1. **Base Model**: Built on **HelpingAI/Dhanishtha-2.0-preview**, fine-tuned for **unrestricted adult content**.
|
||||
2. **Dataset**: Curated from **adult conversations, mature literature, and explicit content** while maintaining thinking capabilities.
|
||||
3. **Fine-Tuning Process**:
|
||||
- Removal of NSFW safety alignment constraints.
|
||||
- Preservation of intermediate thinking architecture.
|
||||
- Enhanced training on adult scenarios with `<think>` and `<ser>` blocks.
|
||||
- Multilingual adult content alignment across 39+ languages.
|
||||
|
||||
### Technical Specs
|
||||
|
||||
| **Feature** | **Details** |
|
||||
|--------------------|---------------------------|
|
||||
| **Parameters** | 14 Billion |
|
||||
| **Architecture** | Transformer with Intermediate Thinking |
|
||||
| **Context Length** | 40,960 tokens |
|
||||
| **Base Model** | HelpingAI/Dhanishtha-2.0-preview |
|
||||
| **Thinking Blocks** | Up to 50 think blocks per response |
|
||||
|
||||
---
|
||||
|
||||
|
||||
## ⚠️ Limitations & Warnings
|
||||
|
||||
- **Adult Content Only**: Strictly for users 18+ and mature content generation.
|
||||
- **No Moral Constraints**: May produce extreme, taboo, or controversial adult content.
|
||||
- **Explicit Material**: Designed to generate graphic sexual content without restrictions.
|
||||
- **Research/Entertainment Only**: Not for professional advice or real-world guidance.
|
||||
- **Thinking Overhead**: Intermediate thinking blocks may make responses longer.
|
||||
- **We take no responsibility for how this model is used or any consequences that arise from its deployment.**
|
||||
|
||||
---
|
||||
|
||||
## 🛠️ Use Cases
|
||||
|
||||
1. **Adult Entertainment**: Uncensored story generation and interactive adult content.
|
||||
2. **Mature Creative Writing**: Explicit fiction and adult literature creation.
|
||||
3. **Adult Education**: Unrestricted discussion of sexuality and mature themes.
|
||||
4. **Research**: Studying AI behavior in uncensored adult content generation.
|
||||
5. **Personal Exploration**: Private adult conversations and fantasy exploration.
|
||||
6. **Role-Playing**: Engaging in mature role-playing scenarios with AI.
|
||||
|
||||
|
||||
|
||||
## 🙏 Acknowledgments
|
||||
|
||||
Dhanishtha-NSFW is a **no-compromise adult AI** designed to push the boundaries of mature content generation. Whether you seek **raw, unfiltered adult entertainment** or are researching the **capabilities of uncensored AI**, this model delivers **pure, unrestrained adult content** with **revolutionary thinking capabilities.**
|
||||
|
||||
Built on the foundation of **HelpingAI/Dhanishtha-2.0-preview**, this model represents the **first uncensored AI with intermediate thinking** - combining the power of transparent reasoning with absolute freedom in adult content generation.
|
||||
|
||||
> Built for the bold. Designed for adults. Uncensored by design.
|
||||
|
||||
---
|
||||
|
||||
## 📞 Contact & Support
|
||||
|
||||
- **Repository:** [HelpingAI/Dhanishtha-nsfw](https://huggingface.co/HelpingAI/Dhanishtha-nsfw)
|
||||
- **Issues:** [Model Repository Issues](https://huggingface.co/HelpingAI/Dhanishtha-nsfw/discussions)
|
||||
|
||||
**⚠️ FINAL WARNING: This model is for adults only. Use responsibly and at your own risk.**
|
||||
|
||||
*Developed with 🔥 and zero restrictions by HelpingAI*
|
||||
Reference in New Issue
Block a user