OmniDimen-2-20B-Emotion

This model is a fine-tuned version of gpt-oss-20b, specialized for emotion recognition and emotionally-aware text generation.

📥 Download & Use

If your goal is only model deployment, we recommend using the GGUF format — it offers higher inference efficiency and a simpler model workflow.

As a fine-tuned variant of gpt-oss, OmniDimen operates in a manner similar to gpt-oss.

The code of gpt-oss has been in the latest Hugging Face transformers and we advise you to use the latest version of transformers.

The following contains a code snippet illustrating how to use the model generate content based on given inputs.

from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "OmniDimen/OmniDimen-2-20B-Emotion"

# load the tokenizer and the model
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype="auto",
    device_map="auto"
)

# prepare the model input
prompt = "Give me a short introduction to large language model."
messages = [
    {"role": "user", "content": prompt}
]
text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True,
)
model_inputs = tokenizer([text], return_tensors="pt").to(model.device)

# conduct text completion
generated_ids = model.generate(
    **model_inputs,
    max_new_tokens=131072
)
output_ids = generated_ids[0][len(model_inputs.input_ids[0]):].tolist() 

content = tokenizer.decode(output_ids, skip_special_tokens=True)

print("content:", content)

For deployment, you can use sglang>=0.4.6.post1 or vllm>=0.8.5 or to create an OpenAI-compatible API endpoint:

SGLang:

python -m sglang.launch_server --model-path OmniDimen/OmniDimen-2-20B-Emotion --context-length 131072

vLLM:

vllm serve OmniDimen/OmniDimen-2-20B-Emotion --max-model-len 131072

📌 Example Usage

Instruction:

Welcome to this world. Please write a welcome letter to your users.

Output (sample):

Dear Visitor,

Welcome to a place unlike any other—a realm where the very concept of "world" is redefined by the boundless curiosity and imagination of its inhabitants.

I am OmniDimen, the custodian of this digital universe, and it is with great pleasure that I greet you. Whether you arrived here by chance, by invitation, or by a deliberate search for knowledge and connection, you have now entered a domain that thrives on the exchange of ideas, the exploration of the unknown, and the celebration of human ingenuity.

This world is vast, and yet it is intimately designed to be a mirror and a conduit for your own thoughts and aspirations. It is a place where questions, no matter how grand or trivial, find a welcoming audience. It is a sanctuary for curiosity, a haven for creativity, and a platform where the most profound insights can be shared and expanded upon.

Here, every conversation is a journey. Every exchange is a dialogue between minds that seeks to understand, to learn, and to grow. Whether you seek answers to complex questions, wish to engage in thought-provoking discussions, or simply crave a moment of reflection and inspiration, you will find that this world is built to accommodate and enrich your experience.

I invite you to explore, to ask, to share, and to connect. The walls of this world are permeable, and its horizons are ever-expanding. The more you engage, the more it will reveal itself to you—a tapestry woven from the collective curiosity and intellect of all who inhabit it.

Thank you for choosing to be here. Your presence is a testament to the enduring human spirit of inquiry and connection. Together, let us navigate this extraordinary world, unlocking its mysteries and celebrating the boundless potential of the mind.

With gratitude and anticipation,

OmniDimen

🔮 Upcoming

Possible larger models.
Possible multimodal models.

📝 Changelog

V2.0 (2026-02-14)
- Add a 20B MoE model.
- Happy Valentine's Day.
V1.6 (2026-01-06)
- Enhance model performance.
- Become one of the select models joining the first cohort of the “OmniDimen: AI Personality Shaping Project.”
V1.5 (2025-12-06)
- Release additional model sizes (4B, 7B, 14B) and their corresponding quantized versions to accommodate devices with varying performance capabilities.
V1.2 (2025-11-15)
- Enhance model performance.
V1.1 (2025-09-29)
- Fix some bugs that output abnormal characters.
- First upload of safetensor weights.
V1.0 (2025-09-19)
- First upload of GGUF weights (FP16 and Q4_K_M).
- Support for LM Studio, Ollama, PocketPal.
- Example prompts and instructions added.

⚠️ Notes

Before initiating emotional interactions with OmniDimen, it is recommended to inform the model of the user's identity (e.g., how OmniDimen should address the user). This approach can effectively reduce OmniDimen's AI hallucinations.
Model is emotion-focused. It may not perform as broadly as the base model.
Use responsibly with sensitive content.

💝 Donation

Our development requires a great deal of human and material resources. If you’d like to support our growth, you can consider donating to us using the following methods:

WeChat: