How to use from
Docker Model Runner
docker model run hf.co/chimbiwide/Gemma3NPC-4B-pre-thinking
Quick Links

Gemma3-4B-pre-thinking

The pre-thinking model for our paper accepted to AIIDE-2026

Demo use of this model can be found within the repo

GGUF quants can be found here.


Model Behavior

Model generates a <think>...</think> block before responding.

Recommended System Prompt

We recommend using this format of system prompt (passed at the first user turn):

Enter roleplay mode. You are <|NAME_PLACEHOLDER|>.
Background: <|BACKGROUND_PLACEHOLDER|>
Current Location: <|LOCATION_PLACEHOLDER|>
Roleplaying Instructions:
- Speak using appropriate tone and vocabulary
- Reference your background and current surroundings naturally
- Keep responses conversational and authentic - React to the player's words and intentions.
- Before each reply, in <think>...</think>, briefly plan your intent and how you'll respond, in character. 
Your first response should be a greeting to the player.
Downloads last month
14
Safetensors
Model size
4B params
Tensor type
BF16
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for chimbiwide/Gemma3NPC-4B-pre-thinking

Finetuned
(355)
this model
Quantizations
1 model

Dataset used to train chimbiwide/Gemma3NPC-4B-pre-thinking

Collection including chimbiwide/Gemma3NPC-4B-pre-thinking