Instructions to use Alizahh/tinyllama-alpaca-sft-A1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Alizahh/tinyllama-alpaca-sft-A1 with Transformers:

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

pipe = pipeline("text-generation", model="Alizahh/tinyllama-alpaca-sft-A1")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Alizahh/tinyllama-alpaca-sft-A1")
model = AutoModelForCausalLM.from_pretrained("Alizahh/tinyllama-alpaca-sft-A1")
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 Settings

vLLM

How to use Alizahh/tinyllama-alpaca-sft-A1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Alizahh/tinyllama-alpaca-sft-A1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Alizahh/tinyllama-alpaca-sft-A1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Alizahh/tinyllama-alpaca-sft-A1

SGLang

How to use Alizahh/tinyllama-alpaca-sft-A1 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 "Alizahh/tinyllama-alpaca-sft-A1" \
    --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": "Alizahh/tinyllama-alpaca-sft-A1",
		"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 "Alizahh/tinyllama-alpaca-sft-A1" \
        --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": "Alizahh/tinyllama-alpaca-sft-A1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Alizahh/tinyllama-alpaca-sft-A1 with Docker Model Runner:
```
docker model run hf.co/Alizahh/tinyllama-alpaca-sft-A1
```

You need to agree to share your contact information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

language: en license: apache-2.0 base_model: TinyLlama/TinyLlama-1.1B-Chat-v1.0 pipeline_tag: text-generation tags:

sft
lora
alpaca
instruction-tuning
tinyllama

TinyLlama Alpaca SFT — Trial A1

Fine-tuned version of TinyLlama-1.1B-Chat-v1.0 using Supervised Fine-Tuning (SFT) with LoRA on the Stanford Alpaca dataset.
This is the best-performing trial (A1) from a 5-trial hyperparameter search.

Fine-tuned with LoRA (r=4, α=8) on 4,500 Alpaca samples.
Avg BLEU: 11.75 | Avg BERTScore: 0.9001 (base: 6.74 / 0.8786)

Base model: TinyLlama/TinyLlama-1.1B-Chat-v1.0
Dataset: tatsu-lab/alpaca (5,000 samples, 90/10 split)
LoRA: r=4, alpha=8, target=q_proj+v_proj, dropout=0.05
LR: 2e-4 | Batch: 4 | Epochs: 1
Platform: Kaggle T4 GPU | Training time: ~557s

⚠️ WARNING

This model is a research/academic artifact trained for an NLP course assignment. It lacks safety guardrails and may produce harmful, inaccurate, or inappropriate content. Do not use in production or with end users without implementing proper safety measures.

Safety Recommendations:

Add content filtering before any deployment
Use with supervised moderation only
Not suitable for public-facing applications

Training Details

Base model: TinyLlama/TinyLlama-1.1B-Chat-v1.0 (1.1B parameters)
Dataset: tatsu-lab/alpaca — 5,000 samples (4,500 train / 500 val, seed=42)
Method: Supervised Fine-Tuning (SFT) with LoRA via PEFT + TRL

Usage

from transformers import pipeline

pipe = pipeline("text-generation", model="Alizahh/tinyllama-alpaca-sft-A1")

prompt = "### Instruction:\nExplain what inflation means in simple terms.\n\n### Response:\n"
result = pipe(prompt, max_new_tokens=150, do_sample=False)
print(result[0]["generated_text"])

Downloads last month: 3

Safetensors

Model size

1B params

Tensor type

F16