Instructions to use Rombo-Org/Rombo-LLM-V3.0-Qwen-72b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Rombo-Org/Rombo-LLM-V3.0-Qwen-72b with Transformers:

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

pipe = pipeline("text-generation", model="Rombo-Org/Rombo-LLM-V3.0-Qwen-72b")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Rombo-Org/Rombo-LLM-V3.0-Qwen-72b")
model = AutoModelForCausalLM.from_pretrained("Rombo-Org/Rombo-LLM-V3.0-Qwen-72b")
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 Rombo-Org/Rombo-LLM-V3.0-Qwen-72b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Rombo-Org/Rombo-LLM-V3.0-Qwen-72b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Rombo-Org/Rombo-LLM-V3.0-Qwen-72b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Rombo-Org/Rombo-LLM-V3.0-Qwen-72b

SGLang

How to use Rombo-Org/Rombo-LLM-V3.0-Qwen-72b 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 "Rombo-Org/Rombo-LLM-V3.0-Qwen-72b" \
    --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": "Rombo-Org/Rombo-LLM-V3.0-Qwen-72b",
		"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 "Rombo-Org/Rombo-LLM-V3.0-Qwen-72b" \
        --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": "Rombo-Org/Rombo-LLM-V3.0-Qwen-72b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Unsloth Studio new

How to use Rombo-Org/Rombo-LLM-V3.0-Qwen-72b with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for Rombo-Org/Rombo-LLM-V3.0-Qwen-72b to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for Rombo-Org/Rombo-LLM-V3.0-Qwen-72b to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for Rombo-Org/Rombo-LLM-V3.0-Qwen-72b to start chatting

Load model with FastModel

pip install unsloth
from unsloth import FastModel
model, tokenizer = FastModel.from_pretrained(
    model_name="Rombo-Org/Rombo-LLM-V3.0-Qwen-72b",
    max_seq_length=2048,
)

Docker Model Runner
How to use Rombo-Org/Rombo-LLM-V3.0-Qwen-72b with Docker Model Runner:
```
docker model run hf.co/Rombo-Org/Rombo-LLM-V3.0-Qwen-72b
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Like my work? Support me on patreon for only $5 a month and get to vote on what model's i make next as well as get access to this org's private repo's

Subscribe bellow:

Patreon.com/Rombodawg

Rombo-LLM-V3.0-Qwen-72b

Rombos-LLM-V3.0-Qwen-72b is a continues finetuned version of the Rombo-LLM-V2.5-Qwen-72b on a Reasoning and Non-reasoning dataset. The models performs exceptionally well when paired with the system prompt that it was trained on during reasoning training. Nearing SOTA levels even quantized to 4-bit.

I highly recommend using a temp of 0.4 when using this model (Especially with the reasoning system prompt)

The system prompt is as follows for multi-reasoning, also called optimized reasoning. (Recommended)

You are an AI assistant that always begins by assessing whether detailed reasoning is needed before answering; follow these guidelines: 1) Start every response with a single <think> block that evaluates the query's complexity and ends with </think>; 2) For straightforward queries, state that no detailed reasoning is required and provide a direct answer; 3) For complex queries, indicate that detailed reasoning is needed, then include an additional "<think> (reasoning) </think> (answer)" block with a concise chain-of-thought before delivering the final answer—keeping your reasoning succinct and adding extra steps only when necessary.

For single reasoning or traditional reasoning you can use the system prompt bellow:

You are an AI assistant that always begins by assessing whether detailed reasoning is needed before answering; follow these guidelines: 1) Start every response with a single  "<think> (reasoning) </think> (answer)" block with a concise chain-of-thought before delivering the final answer—keeping your reasoning succinct and adding extra steps only when necessary.

For non-reasoning use cases no system prompt is needed (Not recommended)

Quantized versions:

Model Evaluation: (Coming soon)