Instructions to use WillisBack/gemma-Summarizer-2b-it-LORA-bnb-4bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use WillisBack/gemma-Summarizer-2b-it-LORA-bnb-4bit with Transformers:

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

pipe = pipeline("text-generation", model="WillisBack/gemma-Summarizer-2b-it-LORA-bnb-4bit")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("WillisBack/gemma-Summarizer-2b-it-LORA-bnb-4bit")
model = AutoModelForCausalLM.from_pretrained("WillisBack/gemma-Summarizer-2b-it-LORA-bnb-4bit")
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=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use WillisBack/gemma-Summarizer-2b-it-LORA-bnb-4bit with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "WillisBack/gemma-Summarizer-2b-it-LORA-bnb-4bit"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "WillisBack/gemma-Summarizer-2b-it-LORA-bnb-4bit",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/WillisBack/gemma-Summarizer-2b-it-LORA-bnb-4bit

SGLang

How to use WillisBack/gemma-Summarizer-2b-it-LORA-bnb-4bit 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 "WillisBack/gemma-Summarizer-2b-it-LORA-bnb-4bit" \
    --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": "WillisBack/gemma-Summarizer-2b-it-LORA-bnb-4bit",
		"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 "WillisBack/gemma-Summarizer-2b-it-LORA-bnb-4bit" \
        --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": "WillisBack/gemma-Summarizer-2b-it-LORA-bnb-4bit",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Unsloth Studio new

How to use WillisBack/gemma-Summarizer-2b-it-LORA-bnb-4bit with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for WillisBack/gemma-Summarizer-2b-it-LORA-bnb-4bit to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for WillisBack/gemma-Summarizer-2b-it-LORA-bnb-4bit to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for WillisBack/gemma-Summarizer-2b-it-LORA-bnb-4bit to start chatting

Load model with FastModel

pip install unsloth
from unsloth import FastModel
model, tokenizer = FastModel.from_pretrained(
    model_name="WillisBack/gemma-Summarizer-2b-it-LORA-bnb-4bit",
    max_seq_length=2048,
)

Docker Model Runner
How to use WillisBack/gemma-Summarizer-2b-it-LORA-bnb-4bit with Docker Model Runner:
```
docker model run hf.co/WillisBack/gemma-Summarizer-2b-it-LORA-bnb-4bit
```

Uploaded as lora model

Developed by: Labagaite
License: apache-2.0
Finetuned from model : unsloth/gemma-2b-it-bnb-4bit

Training Logs

Traning metrics

Evaluation score

Évaluation des rapports générés par les deux modèles d'IA

Modèle de base (unsloth/gemma-2b-it-bnb-4bit)

Performance de la structuration du rapport: 6/10
Qualité du langage: 7/10
Cohérence: 6/10

Modèle fine-tuned (gemma-Summarizer-2b-it-bnb-4bit)

Performance de la structuration du rapport: 8/10
Qualité du langage: 8/10
Cohérence: 8/10

Score global

Modèle de base: 6.3/10
Modèle fine-tuned: 8/10

Conclusion

Le modèle fine-tuned a clairement surpassé le modèle de base en termes de structuration du rapport, qualité du langage et cohérence. Le rapport généré par le modèle fine-tuned est plus clair, plus fluide et mieux organisé. Il offre une analyse plus approfondie et une meilleure compréhension des sujets abordés. En revanche, le modèle de base présente quelques lacunes en termes de cohérence et de structuration. Il pourrait bénéficier d'une amélioration pour offrir des rapports plus percutants et informatifs. Evaluation report and scoring

Wandb logs

You can view the training logs .

Training details

training data

Dataset : fr-summarizer-dataset
Data-size : 7.65 MB
train : 1.97k rows
validation : 440 rows
roles : user , assistant
Format chatml "role": "role", "content": "content", "user": "user", "assistant": "assistant"
*French audio podcast transcription*

Project details

Fine-tuned on French audio podcast transcription data for summarization task. As a result, the model is able to summarize French audio podcast transcription data. The model will be used for an AI application: Report Maker wich is a powerful tool designed to automate the process of transcribing and summarizing meetings. It leverages state-of-the-art machine learning models to provide detailed and accurate reports.

This gemma model was trained 2x faster with Unsloth and Huggingface's TRL library. This gemma was trained with LLM summarizer trainer LLM summarizer trainer