Not-For-All-Audiences

Instructions to use Tarek07/Dungeonmaster-V2.2-Expanded-LLaMa-70B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Tarek07/Dungeonmaster-V2.2-Expanded-LLaMa-70B with Transformers:

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

pipe = pipeline("text-generation", model="Tarek07/Dungeonmaster-V2.2-Expanded-LLaMa-70B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Tarek07/Dungeonmaster-V2.2-Expanded-LLaMa-70B")
model = AutoModelForCausalLM.from_pretrained("Tarek07/Dungeonmaster-V2.2-Expanded-LLaMa-70B", 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]:]))

Inference
Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Tarek07/Dungeonmaster-V2.2-Expanded-LLaMa-70B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Tarek07/Dungeonmaster-V2.2-Expanded-LLaMa-70B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Tarek07/Dungeonmaster-V2.2-Expanded-LLaMa-70B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Tarek07/Dungeonmaster-V2.2-Expanded-LLaMa-70B

SGLang

How to use Tarek07/Dungeonmaster-V2.2-Expanded-LLaMa-70B 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 "Tarek07/Dungeonmaster-V2.2-Expanded-LLaMa-70B" \
    --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": "Tarek07/Dungeonmaster-V2.2-Expanded-LLaMa-70B",
		"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 "Tarek07/Dungeonmaster-V2.2-Expanded-LLaMa-70B" \
        --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": "Tarek07/Dungeonmaster-V2.2-Expanded-LLaMa-70B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Tarek07/Dungeonmaster-V2.2-Expanded-LLaMa-70B with Docker Model Runner:
```
docker model run hf.co/Tarek07/Dungeonmaster-V2.2-Expanded-LLaMa-70B
```

Wow.

by Alrai2550 - opened Apr 2, 2025

Discussion

Alrai2550

Apr 2, 2025

This was exactly what I was looking for.
Sure smut is fun, but this model really shines in its storytelling. It doesn't try to leave a scene early, but also doesn't get stuck in it. It'll actually progress the story when it's natural to do so without nudging from me. It's also willing to put the user character in danger or difficult situations without being led by the nose to do so.

I'm using the recommended settings although the supplied context template was missing the character description block so I edited that back in.

Tarek07

Owner Apr 3, 2025

This was exactly what I was looking for.
Sure smut is fun, but this model really shines in its storytelling. It doesn't try to leave a scene early, but also doesn't get stuck in it. It'll actually progress the story when it's natural to do so without nudging from me. It's also willing to put the user character in danger or difficult situations without being led by the nose to do so.

I'm using the recommended settings although the supplied context template was missing the character description block so I edited that back in.

Thank you for the feedback! I am glad you like the model! Thanks for the edit too.

yano2mch

Jul 15, 2025

•

edited Jul 15, 2025

I agree. I try to do a totally clean SillyTavern scenario to give a model a chance to stand on it's own, and i'm liking the output so far!

Nice and clean output, doesn't break up things, various complex concepts retained, i'll have it do a story analysis later as a test, but this feels on par with the Anubis and GeneticLemonade models, though how it does for general story telling we'll see.

So far, added to my list of 'potential' models :)

As such i'll get the v2.4 and give that a go too.

edit: This type of model expanded to a 150B or 235B and can overcome some of the limits i've seen in my RPing and testing, would be awesome!

yano2mch

Jul 18, 2025

Been trying v2.4, but i don't see a discernible difference. I don't see a need to make a new thread so i'm just appending my opinion here.

I'm not using controlled seeds with controlled inputs of the same to see if there's an output difference.

Regardless both seem like fine models. Personally it feels like there's a bit more repetitions of phrases from v2.4 than v2.2 (with a character who is eager to please).

Tarek07

Owner Jul 18, 2025

Been trying v2.4, but i don't see a discernible difference. I don't see a need to make a new thread so i'm just appending my opinion here.

I'm not using controlled seeds with controlled inputs of the same to see if there's an output difference.

Regardless both seem like fine models. Personally it feels like there's a bit more repetitions of phrases from v2.4 than v2.2 (with a character who is eager to please).

I appreciate the feedback!

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment