Instructions to use andrijdavid/tinyllama-dare with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use andrijdavid/tinyllama-dare with Transformers:

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

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

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("andrijdavid/tinyllama-dare")
model = AutoModelForCausalLM.from_pretrained("andrijdavid/tinyllama-dare", device_map="auto")
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 andrijdavid/tinyllama-dare with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/andrijdavid/tinyllama-dare

SGLang

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

Docker Model Runner
How to use andrijdavid/tinyllama-dare with Docker Model Runner:
```
docker model run hf.co/andrijdavid/tinyllama-dare
```

TinyLlama Merge

Models Merged

The following models were included in the merge:

Disclaimer

The users of this model (hereinafter referred to as "the Model") should be aware of the following points prior to using the Model:

Factual Inaccuracy: The Model has been trained on a diverse set of data sources. However, it may occasionally generate outputs that contain factual inaccuracies. Users are advised to verify the factual correctness of the information provided by the Model, especially when using it for critical applications.
Potential Biases: Despite efforts to ensure fairness and neutrality, the Model may still exhibit biases inherent in the training data. These biases do not reflect the views or intentions of the developers. Users should be cautious of these potential biases, especially when using the Model in contexts where fairness and impartiality are crucial.
Hallucination: The Model might generate content that is not grounded in reality, commonly referred to as "hallucinations." This is a known limitation of language models and should be considered when interpreting the Model's outputs.
Inappropriate Content: While measures have been taken to minimize the risk, the Model might still generate content that could be deemed offensive, inappropriate, or not suitable for all audiences. We advise users to exercise discretion and consider the context of use.

Use at Your Own Risk: The Model is provided "as is," and the developers make no representations or warranties of any kind concerning the Model's performance or suitability for any particular purpose. The user assumes full responsibility and risk of loss resulting from using the Model.

By using the Model, users acknowledge and agree to the terms stated in this disclaimer. This disclaimer is subject to change without notice, and the latest version can be found on the Model's Hugging Face page.

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	38.64
AI2 Reasoning Challenge (25-Shot)	37.29
HellaSwag (10-Shot)	62.78
MMLU (5-Shot)	25.20
TruthfulQA (0-shot)	39.01
Winogrande (5-shot)	65.90
GSM8k (5-shot)	1.67

Downloads last month: 60

Safetensors

Model size

1B params

Tensor type

BF16

Model tree for andrijdavid/tinyllama-dare

BEE-spoke-data/TinyLlama-3T-1.1bee

TinyLlama/TinyLlama-1.1B-Chat-v1.0

kevin009/lamatama

sreeramajay/TinyLlama-1.1B-orca-v1.0

venkycs/Zyte-1B

vihangd/DopeyTinyLlama-1.1B-v1

Merge model

this model

Evaluation results

normalized accuracy on AI2 Reasoning Challenge (25-Shot)
test set Open LLM Leaderboard

37.290
normalized accuracy on HellaSwag (10-Shot)
validation set Open LLM Leaderboard

62.780
accuracy on MMLU (5-Shot)
test set Open LLM Leaderboard

25.200
mc2 on TruthfulQA (0-shot)
validation set Open LLM Leaderboard

39.010
accuracy on Winogrande (5-shot)
validation set Open LLM Leaderboard

65.900
accuracy on GSM8k (5-shot)
test set Open LLM Leaderboard

1.670