Instructions to use softwareweaver/Twilight-Large-123B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use softwareweaver/Twilight-Large-123B with Transformers:

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

pipe = pipeline("text-generation", model="softwareweaver/Twilight-Large-123B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("softwareweaver/Twilight-Large-123B")
model = AutoModelForCausalLM.from_pretrained("softwareweaver/Twilight-Large-123B")
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 softwareweaver/Twilight-Large-123B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "softwareweaver/Twilight-Large-123B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "softwareweaver/Twilight-Large-123B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/softwareweaver/Twilight-Large-123B

SGLang

How to use softwareweaver/Twilight-Large-123B 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 "softwareweaver/Twilight-Large-123B" \
    --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": "softwareweaver/Twilight-Large-123B",
		"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 "softwareweaver/Twilight-Large-123B" \
        --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": "softwareweaver/Twilight-Large-123B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use softwareweaver/Twilight-Large-123B with Docker Model Runner:
```
docker model run hf.co/softwareweaver/Twilight-Large-123B
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Twilight-Large

This is a merge of pre-trained language models created using mergekit by @softwareweaver. Use the prompt format that Mistral Large uses.

EXL2 Quants

https://huggingface.co/softwareweaver/Twilight-Large-123B-EXL2-5bpw

GGUF Quants

https://huggingface.co/mradermacher/Twilight-Large-123B-GGUF

https://huggingface.co/mradermacher/Twilight-Large-123B-i1-GGUF

Use --chat-template llama2 when using llama.cpp

Control Vectors

You can use Control Vectors for Mistral Large https://huggingface.co/jukofyork/creative-writing-control-vectors-v3.0/tree/main/Mistral-Large-Instruct-2407

Control vectors allow fine-tuned control over LLMs, enabling more precise/targeted text generation. More info https://huggingface.co/jukofyork/creative-writing-control-vectors-v3.0

Sample Generations

Some sample generations https://huggingface.co/softwareweaver/Twilight-Large-123B/discussions

Please add your own generations to the community tab. This allows others to evaluate the model outputs before downloading it.

Merge Details

Merge Method

This model was merged using the della_linear merge method using mistralai/Mistral-Large-Instruct-2407 as a base.

Models Merged

The following models were included in the merge:

Configuration

The following YAML configuration was used to produce this model:

models:
  - model: TheDrummer/Behemoth-123B-v1
    parameters:
      weight: 0.25
      density: 0.9
  - model: schnapper79/lumikabra-123B_v0.4
    parameters:
      weight: 0.3
      density: 0.9      
merge_method: della_linear
base_model: mistralai/Mistral-Large-Instruct-2407
parameters:
  epsilon: 0.05
  lambda: 1
  int8_mask: true
dtype: bfloat16

Downloads last month: 5

Safetensors

Model size

123B params

Tensor type

BF16

Model tree for softwareweaver/Twilight-Large-123B

TheDrummer/Behemoth-123B-v1

mistralai/Mistral-Large-Instruct-2407

schnapper79/lumikabra-123B_v0.4

Merge model

this model

Quantizations

2 models