Instructions to use Nitral-AI/Hathor_Stable-v0.2-L3-8B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Nitral-AI/Hathor_Stable-v0.2-L3-8B with Transformers:

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

pipe = pipeline("text-generation", model="Nitral-AI/Hathor_Stable-v0.2-L3-8B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Nitral-AI/Hathor_Stable-v0.2-L3-8B")
model = AutoModelForCausalLM.from_pretrained("Nitral-AI/Hathor_Stable-v0.2-L3-8B")
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]:]))

Inference
Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use Nitral-AI/Hathor_Stable-v0.2-L3-8B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Nitral-AI/Hathor_Stable-v0.2-L3-8B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Nitral-AI/Hathor_Stable-v0.2-L3-8B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Nitral-AI/Hathor_Stable-v0.2-L3-8B

SGLang

How to use Nitral-AI/Hathor_Stable-v0.2-L3-8B 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 "Nitral-AI/Hathor_Stable-v0.2-L3-8B" \
    --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": "Nitral-AI/Hathor_Stable-v0.2-L3-8B",
		"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 "Nitral-AI/Hathor_Stable-v0.2-L3-8B" \
        --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": "Nitral-AI/Hathor_Stable-v0.2-L3-8B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Nitral-AI/Hathor_Stable-v0.2-L3-8B with Docker Model Runner:
```
docker model run hf.co/Nitral-AI/Hathor_Stable-v0.2-L3-8B
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Hathor-L3-8B-v.02-exl2

Notes: Hathor is trained on 3 epochs of private data, synthetic opus instructions, a mix of light/classical novel data, roleplaying chat pairs over llama 3 8B instruct. (expanded)

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	25.70
IFEval (0-Shot)	71.75
BBH (3-Shot)	32.83
MATH Lvl 5 (4-Shot)	9.21
GPQA (0-shot)	4.92
MuSR (0-shot)	5.56
MMLU-PRO (5-shot)	29.96

Downloads last month: 1,459

Model tree for Nitral-AI/Hathor_Stable-v0.2-L3-8B

Merges

15 models

Quantizations

12 models

Spaces using Nitral-AI/Hathor_Stable-v0.2-L3-8B 10

Evaluation results

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

71.750
normalized accuracy on BBH (3-Shot)
Open LLM Leaderboard

32.830
exact match on MATH Lvl 5 (4-Shot)
Open LLM Leaderboard

9.210
acc_norm on GPQA (0-shot)
Open LLM Leaderboard

4.920
acc_norm on MuSR (0-shot)
Open LLM Leaderboard

5.560
accuracy on MMLU-PRO (5-shot)
test set Open LLM Leaderboard

29.960