Instructions to use OccultAI/Doppelganger-Twist-24B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use OccultAI/Doppelganger-Twist-24B with Transformers:

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

pipe = pipeline("text-generation", model="OccultAI/Doppelganger-Twist-24B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("OccultAI/Doppelganger-Twist-24B")
model = AutoModelForCausalLM.from_pretrained("OccultAI/Doppelganger-Twist-24B", 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]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use OccultAI/Doppelganger-Twist-24B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "OccultAI/Doppelganger-Twist-24B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "OccultAI/Doppelganger-Twist-24B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/OccultAI/Doppelganger-Twist-24B

SGLang

How to use OccultAI/Doppelganger-Twist-24B 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 "OccultAI/Doppelganger-Twist-24B" \
    --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": "OccultAI/Doppelganger-Twist-24B",
		"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 "OccultAI/Doppelganger-Twist-24B" \
        --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": "OccultAI/Doppelganger-Twist-24B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use OccultAI/Doppelganger-Twist-24B with Docker Model Runner:
```
docker model run hf.co/OccultAI/Doppelganger-Twist-24B
```

More focused

by redaihf - opened about 11 hours ago

Discussion

redaihf

about 11 hours ago

•

edited about 11 hours ago

This model continues to be opinionated like Goetia 1.4 while being more coherent and compliant. It tends to weave its complaints about the subject matter into its responses with at least some finesse. It is somewhat better at instruction following than its progenitor and reveals more of its knowledge in its storytelling.

Naphula

Occult AI org about 10 hours ago

I'm testing a della merge of all 3 Goetia 1.4 variants right now. Not sure but perhaps the slightly lower norm_div of twist from base helps with instruction.

[DELLA Audit] Layer: model.layers.20.mlp.down_proj.weight | Lambda=1.00
  [BASE] mistralai--Magistral-Small-2509/textonly
  Doppelganger-Twist-24B                            : ████████████████                                    33.1% (W:0.50 D:0.90 N:8.35 E:0.09)
  Goetia-24B-v1.4                                   : ████████████████                                    33.4% (W:0.50 D:0.90 N:8.43 E:0.09)
  Ouroboros-24B-v1.4                                : ████████████████                                    33.4% (W:0.50 D:0.90 N:8.43 E:0.09)
Executing graph:  34%|████████████████████▊                                         | 855/2546 [06:28<13:40,  2.06it/s]

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