Instructions to use deqing/llama-3.2-350M-fourier with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use deqing/llama-3.2-350M-fourier with Transformers:

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

pipe = pipeline("text-generation", model="deqing/llama-3.2-350M-fourier")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("deqing/llama-3.2-350M-fourier")
model = AutoModelForCausalLM.from_pretrained("deqing/llama-3.2-350M-fourier")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use deqing/llama-3.2-350M-fourier with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "deqing/llama-3.2-350M-fourier"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "deqing/llama-3.2-350M-fourier",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/deqing/llama-3.2-350M-fourier

SGLang

How to use deqing/llama-3.2-350M-fourier 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 "deqing/llama-3.2-350M-fourier" \
    --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": "deqing/llama-3.2-350M-fourier",
		"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 "deqing/llama-3.2-350M-fourier" \
        --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": "deqing/llama-3.2-350M-fourier",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use deqing/llama-3.2-350M-fourier with Docker Model Runner:
```
docker model run hf.co/deqing/llama-3.2-350M-fourier
```

llama-3.2-350M-fourier

This model is a fine-tuned version of llama_small_config.json on the None dataset. It achieves the following results on the evaluation set:

Loss: 2.5784

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: 0.0005
train_batch_size: 2
eval_batch_size: 2
seed: 42
distributed_type: multi-GPU
num_devices: 2
gradient_accumulation_steps: 16
total_train_batch_size: 64
total_eval_batch_size: 4
optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 1000
num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss
4.6741	0.0552	1000	4.6001
3.6587	0.1105	2000	3.7151
3.2297	0.1657	3000	3.2263
3.0288	0.2210	4000	3.0349
2.9584	0.2762	5000	2.9393
2.9657	0.3315	6000	2.9893
2.8654	0.3867	7000	2.8074
2.6982	0.4420	8000	2.7580
2.7292	0.4972	9000	2.7214
2.7568	0.5525	10000	2.6956
2.6141	0.6077	11000	2.6669
2.631	0.6630	12000	2.6421
2.6837	0.7182	13000	2.6185
2.6257	0.7734	14000	2.6032
2.5669	0.8287	15000	2.5918
2.6383	0.8839	16000	2.5836
2.5749	0.9392	17000	2.5796
2.613	0.9944	18000	2.5784

Framework versions

Transformers 4.48.2
Pytorch 2.3.1+cu118
Datasets 3.2.0
Tokenizers 0.21.0

Downloads last month: 4

Safetensors

Model size

0.3B params

Tensor type

BF16