Instructions to use Dracones/L3.3-Damascus-R1_exl2_2.25bpw with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Dracones/L3.3-Damascus-R1_exl2_2.25bpw with Transformers:

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

pipe = pipeline("text-generation", model="Dracones/L3.3-Damascus-R1_exl2_2.25bpw")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Dracones/L3.3-Damascus-R1_exl2_2.25bpw")
model = AutoModelForCausalLM.from_pretrained("Dracones/L3.3-Damascus-R1_exl2_2.25bpw")
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 Dracones/L3.3-Damascus-R1_exl2_2.25bpw with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Dracones/L3.3-Damascus-R1_exl2_2.25bpw"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Dracones/L3.3-Damascus-R1_exl2_2.25bpw",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Dracones/L3.3-Damascus-R1_exl2_2.25bpw

SGLang

How to use Dracones/L3.3-Damascus-R1_exl2_2.25bpw 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 "Dracones/L3.3-Damascus-R1_exl2_2.25bpw" \
    --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": "Dracones/L3.3-Damascus-R1_exl2_2.25bpw",
		"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 "Dracones/L3.3-Damascus-R1_exl2_2.25bpw" \
        --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": "Dracones/L3.3-Damascus-R1_exl2_2.25bpw",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Dracones/L3.3-Damascus-R1_exl2_2.25bpw with Docker Model Runner:
```
docker model run hf.co/Dracones/L3.3-Damascus-R1_exl2_2.25bpw
```

Configuration Parsing Warning:In config.json: "quantization_config.bits" must be an integer

L3.3-Damascus-R1 - EXL2 2.25bpw

This is a 2.25bpw EXL2 quant of Steelskull/L3.3-Damascus-R1

Details about the model can be found at the above model page.

Perplexity Scoring

Below are the perplexity scores for the EXL2 models. A lower score is better.

Quant Level	Perplexity Score
5.0	4.6682
4.5	4.7686
4.0	4.9222
3.5	5.2946
3.0	6.5971
2.75	8.3347
2.5	9.1701
2.25	10.6287

Downloads last month: -

Model tree for Dracones/L3.3-Damascus-R1_exl2_2.25bpw

Base model

EVA-UNIT-01/EVA-LLaMA-3.33-70B-v0.1

Finetuned

Steelskull/L3.3-Damascus-R1

Quantized

(17)

this model