Instructions to use w-ahmad/tiny-llama-3-relu with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use w-ahmad/tiny-llama-3-relu with Transformers:

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

pipe = pipeline("text-generation", model="w-ahmad/tiny-llama-3-relu")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("w-ahmad/tiny-llama-3-relu", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use w-ahmad/tiny-llama-3-relu with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "w-ahmad/tiny-llama-3-relu"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "w-ahmad/tiny-llama-3-relu",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/w-ahmad/tiny-llama-3-relu

SGLang

How to use w-ahmad/tiny-llama-3-relu 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 "w-ahmad/tiny-llama-3-relu" \
    --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": "w-ahmad/tiny-llama-3-relu",
		"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 "w-ahmad/tiny-llama-3-relu" \
        --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": "w-ahmad/tiny-llama-3-relu",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use w-ahmad/tiny-llama-3-relu with Docker Model Runner:
```
docker model run hf.co/w-ahmad/tiny-llama-3-relu
```

tiny-llama-3-relu

This model is a fine-tuned version of on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 1.3448

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: 0.0003
train_batch_size: 32
eval_batch_size: 32
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 64
optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: constant
training_steps: 10000

Training results

Training Loss	Epoch	Step	Validation Loss
4.7466	0.0319	500	2.3445
3.8322	0.0637	1000	1.9124
3.4850	0.0956	1500	1.7444
3.3109	0.1274	2000	1.6543
3.1761	0.1593	2500	1.5934
3.0842	0.1911	3000	1.5510
3.0527	0.2230	3500	1.5208
2.9735	0.2548	4000	1.4916
2.9263	0.2867	4500	1.4705
2.8899	0.3185	5000	1.4522
2.8697	0.3504	5500	1.4347
2.8389	0.3823	6000	1.4212
2.8109	0.4141	6500	1.4078
2.7935	0.4460	7000	1.3972
2.7701	0.4778	7500	1.3853
2.7457	0.5097	8000	1.3772
2.7358	0.5415	8500	1.3679
2.7298	0.5734	9000	1.3606
2.6985	0.6052	9500	1.3523
2.6766	0.6371	10000	1.3448

Framework versions

Transformers 5.15.0.dev0
Pytorch 2.6.0+cu124
Datasets 5.0.1
Tokenizers 0.22.2

Downloads last month: 10

Safetensors

Model size

52.1M params

Tensor type

F32