Instructions to use SuperPeaceBusters/Jintan-LLaMA3.2-1B-v1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use SuperPeaceBusters/Jintan-LLaMA3.2-1B-v1 with Transformers:

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

pipe = pipeline("text-generation", model="SuperPeaceBusters/Jintan-LLaMA3.2-1B-v1")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("SuperPeaceBusters/Jintan-LLaMA3.2-1B-v1")
model = AutoModelForCausalLM.from_pretrained("SuperPeaceBusters/Jintan-LLaMA3.2-1B-v1", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use SuperPeaceBusters/Jintan-LLaMA3.2-1B-v1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "SuperPeaceBusters/Jintan-LLaMA3.2-1B-v1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "SuperPeaceBusters/Jintan-LLaMA3.2-1B-v1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/SuperPeaceBusters/Jintan-LLaMA3.2-1B-v1

SGLang

How to use SuperPeaceBusters/Jintan-LLaMA3.2-1B-v1 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 "SuperPeaceBusters/Jintan-LLaMA3.2-1B-v1" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "SuperPeaceBusters/Jintan-LLaMA3.2-1B-v1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

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 "SuperPeaceBusters/Jintan-LLaMA3.2-1B-v1" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "SuperPeaceBusters/Jintan-LLaMA3.2-1B-v1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use SuperPeaceBusters/Jintan-LLaMA3.2-1B-v1 with Docker Model Runner:
```
docker model run hf.co/SuperPeaceBusters/Jintan-LLaMA3.2-1B-v1
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Jintan-LLaMA3.2-1B-v1

This model is a fine-tuned version of Dolphin3.0-Llama3.2-1B trained on a cleaned version of the Anohana-Subtitles dataset.

Model description

Jintan-LLaMA3.2-1B-v1 is an experimental fine-tune that imitate the personality, tone, and speaking style of Jintan from Anohana: The Flower We Saw That Day.

The model is intended for lightweight casual conversations. The goal was not to create a powerful or reliable assistant, but rather a small character-focused model that feels somewhat like talking to Jintan.

Limitations

This model is highly experimental and the results are limited

The character imitation is inconsistent.
Responses can become repetitive or incoherent.
The model was trained on a relatively small subtitle dataset, which significantly limits its general capabilities.
Overall performance is modest due to the small model size (1B parameters).