Instructions to use OddTheGreat/Rotor_24B_V.1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use OddTheGreat/Rotor_24B_V.1 with Transformers:

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

pipe = pipeline("text-generation", model="OddTheGreat/Rotor_24B_V.1")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("OddTheGreat/Rotor_24B_V.1")
model = AutoModelForCausalLM.from_pretrained("OddTheGreat/Rotor_24B_V.1")
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 OddTheGreat/Rotor_24B_V.1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "OddTheGreat/Rotor_24B_V.1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "OddTheGreat/Rotor_24B_V.1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/OddTheGreat/Rotor_24B_V.1

SGLang

How to use OddTheGreat/Rotor_24B_V.1 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 "OddTheGreat/Rotor_24B_V.1" \
    --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": "OddTheGreat/Rotor_24B_V.1",
		"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 "OddTheGreat/Rotor_24B_V.1" \
        --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": "OddTheGreat/Rotor_24B_V.1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use OddTheGreat/Rotor_24B_V.1 with Docker Model Runner:
```
docker model run hf.co/OddTheGreat/Rotor_24B_V.1
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Rotor_24B_V.1

This is a merge of pre-trained language models.

Goal of this merge was to create RP model to replace Circuitry as my main model. Highest priority this time was style of prose, I wanted it to be least "mechanical" as possible.

After a ton of experiments, I made this thing.

Model is merge of Mistral-Small 24b finetunes, it's smart, excellent at following instructions, creative and handles 12k context perfectly. (I wish I could test more context on reasonable speed.)

Model is stable, but prone to longer replies. Model handles long dialogues well, and good at keeping balance between sfw and nsfw.

I prefer to play text adventure, so it was most important aspect for me. And model solidly works in that scenario too. It can try to talk for user, but that is rare (~1/20 replies). Characters can say no to you, if it fits their description.

Ru language is tested, was good as assistant, RP was tested too, with positive impression (though I need more cards for tests).

Also, i tested french, in assistant mode, trying to renew my rusty french for fun, and it worked well, so it worth to try to test french rp.

Tested on mistral-tekkenV7 preset, Hamonv1\Shingane sysprompt, T 0.81 topP 0.95 minP 0.05 XTC 0.1 0.1