Instructions to use agentlans/Llama3.1-8B-drill with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use agentlans/Llama3.1-8B-drill with Transformers:

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

pipe = pipeline("text-generation", model="agentlans/Llama3.1-8B-drill")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("agentlans/Llama3.1-8B-drill")
model = AutoModelForCausalLM.from_pretrained("agentlans/Llama3.1-8B-drill")
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

vLLM

How to use agentlans/Llama3.1-8B-drill with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "agentlans/Llama3.1-8B-drill"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "agentlans/Llama3.1-8B-drill",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/agentlans/Llama3.1-8B-drill

SGLang

How to use agentlans/Llama3.1-8B-drill 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 "agentlans/Llama3.1-8B-drill" \
    --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": "agentlans/Llama3.1-8B-drill",
		"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 "agentlans/Llama3.1-8B-drill" \
        --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": "agentlans/Llama3.1-8B-drill",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use agentlans/Llama3.1-8B-drill with Docker Model Runner:
```
docker model run hf.co/agentlans/Llama3.1-8B-drill
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Llama3.1-8B-drill

This is a merge of some highest scoring Llama 3.1 8B models on the IFEval task on the Open LLM Leaderboard.
The purpose is to create a model that can follow instructions well.
It's not meant to be a particularly smart, stylish, or uncensored model.

Update: This turned out to be a mediocre model compared to its parents at even the IFEval task. Consider those models instead. 😕

Merge Details

Merge Method

This model was merged using the Model Stock merge method using Meta-Llama-3.1-8B-Instruct as a base using mergekit.

Models Merged

The following models were included in the merge:

Configuration

The following YAML configuration was used to produce this model:

models:
  - model: Configurable-Llama-3.1-8B-Instruct
  - model: Llama-3.1-Tulu-3-8B
  - model: Llama-3.1-8B-Ultra-Instruct
  - model: Llama-3.1-Storm-8B
merge_method: model_stock
base_model: Meta-Llama-3.1-8B-Instruct
parameters:
  normalize: true
  weight: 1.0
dtype: bfloat16

Open LLM Leaderboard Evaluation Results

Detailed results can be found here! Summarized results can be found here!

Metric	Value (%)
Average	26.62
IFEval (0-Shot)	76.52
BBH (3-Shot)	28.79
MATH Lvl 5 (4-Shot)	16.54
GPQA (0-shot)	2.35
MuSR (0-shot)	4.70
MMLU-PRO (5-shot)	30.84

Downloads last month: 30

Safetensors

Model size

8B params

Tensor type

BF16

Model tree for agentlans/Llama3.1-8B-drill

Quantizations

1 model

Paper for agentlans/Llama3.1-8B-drill

Model Stock: All we need is just a few fine-tuned models

Paper • 2403.19522 • Published Mar 28, 2024 • 15

Evaluation results

averaged accuracy on IFEval (0-Shot)
Open LLM Leaderboard

76.520
normalized accuracy on BBH (3-Shot)
test set Open LLM Leaderboard

28.790
exact match on MATH Lvl 5 (4-Shot)
test set Open LLM Leaderboard

16.540
acc_norm on GPQA (0-shot)
Open LLM Leaderboard

2.350
acc_norm on MuSR (0-shot)
Open LLM Leaderboard

4.700
accuracy on MMLU-PRO (5-shot)
test set Open LLM Leaderboard

30.840