Not-For-All-Audiences

Instructions to use Novaciano/Gemma3-Quiet.Hours-1B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Novaciano/Gemma3-Quiet.Hours-1B with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="Novaciano/Gemma3-Quiet.Hours-1B")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Novaciano/Gemma3-Quiet.Hours-1B")
model = AutoModelForCausalLM.from_pretrained("Novaciano/Gemma3-Quiet.Hours-1B")

Inference
Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Novaciano/Gemma3-Quiet.Hours-1B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Novaciano/Gemma3-Quiet.Hours-1B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Novaciano/Gemma3-Quiet.Hours-1B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/Novaciano/Gemma3-Quiet.Hours-1B

SGLang

How to use Novaciano/Gemma3-Quiet.Hours-1B with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "Novaciano/Gemma3-Quiet.Hours-1B" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Novaciano/Gemma3-Quiet.Hours-1B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "Novaciano/Gemma3-Quiet.Hours-1B" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Novaciano/Gemma3-Quiet.Hours-1B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use Novaciano/Gemma3-Quiet.Hours-1B with Docker Model Runner:
```
docker model run hf.co/Novaciano/Gemma3-Quiet.Hours-1B
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Gemma3 • Quiet Hours 1B

Model Overview

This model is a dialogue-oriented merged language model optimized for Slice of Life roleplay, casual conversation, and emotionally expressive character interaction.

It is designed to balance:

Natural conversational flow
Emotional expressiveness
Narrative continuity
Informal dialogue realism

The merge prioritizes personality richness and conversational texture over strict factual reliability.

Model Description

This model is a merge of pre-trained language models created using mergekit.

It combines:

Erudite-style reasoning and structured responses
Uncensored expressive generation
Emotional narrative tone shaping
Conversational roleplay coherence

The resulting model is intended primarily for creative writing, RP chatbots, and slice-of-life storytelling scenarios.

Merge Details

Merge Method

This model was merged using the DARE TIES merge method using Novaciano/Gemma3-Emotional_Uncensored-1B as a base.

DARE TIES was selected for:

Stability preservation from base weights
Controlled stylistic blending
Reduced catastrophic personality override

Models Merged

The following models were included in the merge:

Novaciano/Heretic.Erudite-1B
- Adds structured reasoning and response clarity
DavidAU/gemma-3-1b-it-heretic-extreme-uncensored-abliterated
- Adds raw conversational looseness and reduced filtering
Novaciano/Infected.Crystal-1B
- Adds stylistic unpredictability and narrative tone variance

Intended Use

Primary Use Cases

Slice of Life roleplay
Character-driven storytelling
Chat-style AI companions
Casual bilingual conversation (EN/ES)
Emotional narrative writing assistance

Secondary Use Cases

General creative writing
Informal Q&A
Dialogue drafting

Out-of-Scope Use

This model is not optimized for:

Legal advice
Medical advice
High-precision factual research
Scientific calculation
Safety-critical applications

Training Data

Primary dataset influence:

Stormtrooperaim/Erudite-V2-250K
- Dialogue-heavy
- Mixed instruction + conversational samples
- Informal tone bias

The final model behavior is strongly influenced by merge weight interaction rather than direct training.

Prompting Guide

Recommended Format

<character>: dialogue or narration
<user>: reply

[Scene description]

Character dialogue...

Tips

Works best with narrative or conversational prompts
Responds well to emotional or situational framing
Performs better with persona context

Strengths

Natural conversational rhythm
Emotional tone variation
Good short-to-mid context narrative continuity
Good bilingual casual dialogue

Limitations

May hallucinate factual information
Emotional tone may overshoot prompt intent
Can become verbose in open-ended scenes
Not optimized for strict instruction-following

Bias, Risks, and Safety

Due to source model characteristics:

May produce uncensored or informal language
May reflect dataset conversational biases
Not filtered for sensitive or regulated content domains

Users should implement external moderation layers if required.

Evaluation

Informal evaluation via:

Roleplay scenario testing
Dialogue naturalness scoring
Emotional tone consistency checks

No standardized benchmark evaluation is claimed.

Technical Specifications

Attribute	Value
Architecture	Gemma-based
Parameter Size	~1B
Merge Method	DARE TIES
Precision	bfloat16
Framework	Transformers

Usage Example

from transformers import AutoTokenizer, AutoModelForCausalLM

model_name = "YOUR_MODEL_NAME"

tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name)

prompt = "Character: It's been a long day...\nUser:"
inputs = tokenizer(prompt, return_tensors="pt")

output = model.generate(**inputs, max_new_tokens=120)
print(tokenizer.decode(output[0], skip_special_tokens=True))

Credits

Merge created using:

mergekit
Source models listed above
Dataset contributors

License

Follows base Gemma license requirements and any applicable upstream model licenses.

Users are responsible for verifying downstream compliance.

Configuration

The following YAML configuration was used to produce this model:

merge_method: dare_ties
dtype: bfloat16

base_model: Novaciano/Gemma3-Emotional_Uncensored-1B

models:
  - model: Novaciano/Heretic.Erudite-1B
    parameters:
      weight: 1.10

  - model: Novaciano/Infected.Crystal-1B
    parameters:
      weight: 0.75

  - model: DavidAU/gemma-3-1b-it-heretic-extreme-uncensored-abliterated
    parameters:
      weight: 0.60