Pygmalion-1.3B
| Property | Value |
|---|---|
| Parameter Count | 1.52B parameters |
| Model Type | Dialogue Model |
| License | AGPL-3.0 |
| Training Data Size | 56MB |
| Tensor Type | FP16, U8 |
What is pygmalion-1.3b?
Pygmalion-1.3B is a specialized dialogue model developed by PygmalionAI, built upon EleutherAI's pythia-1.3b-deduped architecture. This proof-of-concept model represents a significant advancement in conversational AI, featuring 1.52 billion parameters and specialized training for dialogue generation.
Implementation Details
The model was fine-tuned using ColossalAI framework, processing approximately 11.4 million tokens over 5440 steps. The training was completed on a single 24GB GPU in under 21 hours, utilizing both real and partially machine-generated conversations for training data.
- Built on pythia-1.3b-deduped architecture
- Fine-tuned using ColossalAI framework
- Implements FP16 and U8 tensor types
- Trained on 56MB of diverse dialogue data
Core Capabilities
- Advanced dialogue generation and response
- Character-based conversation handling
- Context-aware responses with dialogue history support
- Flexible prompt formatting for character interactions
Frequently Asked Questions
Q: What makes this model unique?
The model's distinctive feature is its specialized dialogue capabilities and character-based interaction system, making it particularly suitable for conversational applications with defined personality traits.
Q: What are the recommended use cases?
The model is designed for dialogue generation with character personas, making it suitable for chatbots, interactive storytelling, and conversational agents. However, it's important to note that the model is not suitable for use by minors due to potential X-rated content generation.
