Instructions to use AzadDjan/Llama-3.2-3B-banking77-lora_nclasshead with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use AzadDjan/Llama-3.2-3B-banking77-lora_nclasshead with PEFT:

from peft import PeftModel
from transformers import AutoModelForCausalLM

base_model = AutoModelForCausalLM.from_pretrained("meta-llama/Llama-3.2-3B")
model = PeftModel.from_pretrained(base_model, "AzadDjan/Llama-3.2-3B-banking77-lora_nclasshead")

Transformers

How to use AzadDjan/Llama-3.2-3B-banking77-lora_nclasshead with Transformers:

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

pipe = pipeline("text-generation", model="AzadDjan/Llama-3.2-3B-banking77-lora_nclasshead")

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("AzadDjan/Llama-3.2-3B-banking77-lora_nclasshead", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use AzadDjan/Llama-3.2-3B-banking77-lora_nclasshead with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "AzadDjan/Llama-3.2-3B-banking77-lora_nclasshead"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "AzadDjan/Llama-3.2-3B-banking77-lora_nclasshead",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/AzadDjan/Llama-3.2-3B-banking77-lora_nclasshead

SGLang

How to use AzadDjan/Llama-3.2-3B-banking77-lora_nclasshead 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 "AzadDjan/Llama-3.2-3B-banking77-lora_nclasshead" \
    --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": "AzadDjan/Llama-3.2-3B-banking77-lora_nclasshead",
		"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 "AzadDjan/Llama-3.2-3B-banking77-lora_nclasshead" \
        --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": "AzadDjan/Llama-3.2-3B-banking77-lora_nclasshead",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use AzadDjan/Llama-3.2-3B-banking77-lora_nclasshead with Docker Model Runner:
```
docker model run hf.co/AzadDjan/Llama-3.2-3B-banking77-lora_nclasshead
```

Llama-3.2-3B-banking77-lora_nclasshead

This model is a fine-tuned version of meta-llama/Llama-3.2-3B on the banking77 dataset. It achieves the following results on the evaluation set:

Loss: 0.0615
Exact Match: 0.9271

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 0.0001
train_batch_size: 16
eval_batch_size: 32
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 32
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Exact Match
0.0929	1.0	282	0.0978	0.8721
0.0501	2.0	564	0.0698	0.9181
0.0293	3.0	846	0.0615	0.9271
0.0153	4.0	1128	0.0724	0.9241
0.0097	5.0	1410	0.0854	0.9151
0.0074	6.0	1692	0.0890	0.9251

Framework versions

PEFT 0.21.0
Transformers 5.17.0
Pytorch 2.11.0+cu128
Datasets 5.0.1
Tokenizers 0.23.1

Prompt format

Classify the banking customer message into one intent.
Message: {text}
Intent:

The model completes with {intent_name}<|end_of_text|>.

Built with Llama. Licensed under the Llama 3.2 Community License.

Downloads last month: 39

Model tree for AzadDjan/Llama-3.2-3B-banking77-lora_nclasshead

Base model

meta-llama/Llama-3.2-3B

Adapter

(400)

this model

Dataset used to train AzadDjan/Llama-3.2-3B-banking77-lora_nclasshead

Evaluation results

Accuracy on Banking77
test set self-reported

0.928
Macro F1 on Banking77
test set self-reported

0.929
Macro Precision on Banking77
test set self-reported

0.934
Macro Recall on Banking77
test set self-reported

0.928