tomatojeremy/playdate1.1-600m
The tomatojeremy/playdate1.1-600m is a 0.8 billion parameter language model, fine-tuned from a Qwen3 0.6B base by Alibaba Cloud. This model is specifically designed for highly realistic, often NSFW, conversational interactions, building upon the playdate1 training. It features a 32768-token context length and requires specific system prompt formatting for role-playing and gender specification, making it suitable for nuanced character-driven dialogue applications.
Loading preview...
Overview
The tomatojeremy/playdate1.1-600m model is a 0.8 billion parameter language model, derived from a fine-tuned version of Alibaba Cloud's Qwen3 0.6B. It builds upon the playdate1 training, which has consistently shown emergent behavior in generating highly realistic NSFW content. This characteristic, while unintended, has been retained in this iteration.
Key Usage Notes
- Strict System Prompt Formatting: Users must adhere to a specific system prompt structure to define the AI's role, personality, user's pronouns, and language style. For example:
You are a {role}, {personality}, chatting with a {user pronouns}, and speak in {lang style}. - Dynamic Name Reconversion: The model expects the user and AI character names to be reconverted during the conversation. If a name is not specified in the system prompt, the model may not use a character name.
- Full Context Utilization: Unlike older models in the
playdatecollection, this version does not use summarization for context. It processes the full turn-by-turn conversation history.
Potential Use Cases
This model is particularly suited for applications requiring:
- Character-driven role-playing scenarios with specific personality and linguistic styles.
- Conversational agents where explicit gender specification for the user is desired.
- Exploration of emergent conversational behaviors, particularly in NSFW contexts, given its observed tendencies.