Instructions to use Sorihon/Reforged-Memories-12B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Sorihon/Reforged-Memories-12B with Transformers:

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

pipe = pipeline("text-generation", model="Sorihon/Reforged-Memories-12B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Sorihon/Reforged-Memories-12B")
model = AutoModelForCausalLM.from_pretrained("Sorihon/Reforged-Memories-12B", 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=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Sorihon/Reforged-Memories-12B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Sorihon/Reforged-Memories-12B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Sorihon/Reforged-Memories-12B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Sorihon/Reforged-Memories-12B

SGLang

How to use Sorihon/Reforged-Memories-12B 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 "Sorihon/Reforged-Memories-12B" \
    --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": "Sorihon/Reforged-Memories-12B",
		"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 "Sorihon/Reforged-Memories-12B" \
        --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": "Sorihon/Reforged-Memories-12B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Sorihon/Reforged-Memories-12B with Docker Model Runner:
```
docker model run hf.co/Sorihon/Reforged-Memories-12B
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Models Merged

The following models were included in the merge:

Vortex5/Nether-Moon-12B
Vortex5/Wicked-Oblivion-12B
Vortex5/Stellar-Witch-12B
Vortex5/Shining-Seraph-12B
DreadPoor/Irix-12B-Model_Stock
redrix/patricide-12B-Unslop-Mell
mistralai/Mistral-Nemo-Instruct-2407
ReadyArt/Omega-Darker_The-Final-Directive-12B
v000000/MN-12B-Estrella-v2.2
anthracite-org/magnum-v4-12b
Sorihon/MN-12B-Lyra-v4-Heretic
Sorihon/MN-12B-Celeste-V1.9-Heretic
Sorihon/MN-Chinofun-12B-4-Heretic
Delta-Vector/Rei-12B
LatitudeGames/Wayfarer-2-12B
Sorihon/Ethereal-Oblivion-12B

Step 1: Model Stock (Vortex)

Vortex5/Nether-Moon-12B
Vortex5/Wicked-Oblivion-12B
Vortex5/Stellar-Witch-12B
Vortex5/Shining-Seraph-12B
mistralai/Mistral-Nemo-Instruct-2407 (base model)

Step 2: Nuslerp (Reforged)

Vortex (weight 0.5)
DreadPoor/Irix-12B-Model_Stock (weight 0.5)
mistralai/Mistral-Nemo-Instruct-2407 (base model)

Step 3: Dare Ties

Reforged (Density 0.7 Weight 0.6)
redrix/patricide-12B-Unslop-Mell (Density 0.6 Weight 0.5)
mistralai/Mistral-Nemo-Instruct-2407 (base model)

Step 4: Karcher (ReforgedV2)

model: Reforged-Memories-12B
model: ReadyArt/Omega-Darker_The-Final-Directive-12B
model: v000000/MN-12B-Estrella-v2.2
model: anthracite-org/magnum-v4-12b
model: Sorihon/MN-12B-Lyra-v4-Heretic
model: Sorihon/MN-12B-Celeste-V1.9-Heretic
model: Sorihon/MN-Chinofun-12B-4-Heretic
model: mistralai/Mistral-Nemo-Instruct-2407

Step 5: Della (ReforgedV3)

models:

mistralai/Mistral-Nemo-Instruct-2407
Reforged-Memories-12B parameters: density: 0.7 weight: 0.55 epsilon: 0.25
model: Reforged-MemoriesV2-12B parameters: density: 0.7 weight: 0.6 epsilon: 0.25 base_model: mistralai/Mistral-Nemo-Instruct-2407 merge_method: della parameters: normalize: true int8_mask: true dtype: bfloat16

Step 6: Dare_Ties (ReforgedV4)

model: Reforged-MemoriesV3-12B/ parameters: density: 0.5 weight: 0.5
model: Delta-Vector/Rei-12B parameters: density: 0.5 weight: 0.5 base_model: Reforged-MemoriesV3-12B/ merge_method: dare_ties parameters: lambda: 1 tokenizer: source: union normalize: true int8_mask: true dtype: bfloat16

Step 6: Nuslerp (Final-Forge)

models:

model: Reforged-MemoriesV4-12B/ parameters: weight: 0.6
model: LatitudeGames/Wayfarer-2-12B parameters: weight: 0.45 base_model: Reforged-MemoriesV3-12B merge_method: nuslerp parameters: tokenizer: source: union normalize: true int8_mask: true dtype: bfloat16