OmniDimen/OmniDimen-v2.5-4B-Emotion

VISIONConcurrency Cost:1Model Size:4.5BQuant:BF16Ctx Length:32kTool Calling:SupportedPublished:Mar 23, 2026License:mitArchitecture:Transformer0.0K Open Weights Cold

OmniDimen/OmniDimen-v2.5-4B-Emotion is a 4.5 billion parameter language model fine-tuned from Qwen/Qwen3.5-4B, specifically designed for emotion recognition and emotionally-aware text generation. This model leverages a 262,144 token context length and incorporates multi-modal capabilities, making it suitable for applications requiring nuanced understanding and generation of emotional content. It is optimized for deployment with high-efficiency inference frameworks like SGLang and vLLM.

Loading preview...

OmniDimen-v2.5-4B-Emotion: Emotion-Focused LLM

OmniDimen-v2.5-4B-Emotion is a 4.5 billion parameter model, fine-tuned from the Qwen3.5-4B base, with a primary specialization in emotion recognition and emotionally-aware text generation. This model is designed to understand and generate text with nuanced emotional content, making it distinct from general-purpose LLMs.

Key Capabilities & Features

  • Emotion-Focused Generation: Excels at producing text that reflects or responds to emotional cues.
  • Emotion Recognition: Capable of identifying and interpreting emotions within input text.
  • Large Context Window: Features a default context length of 262,144 tokens, enabling complex and extended interactions, though 128K tokens are recommended to preserve 'thinking capabilities' and avoid OOM errors.
  • Multi-modal Adoption: The V2.5 update introduced multi-modal capabilities, enhancing its processing of diverse input types.
  • Optimized for Deployment: Provided in safetensor (BF16) and GGUF (FP16 & Q4_K_M) formats, with recommendations for high-efficiency inference frameworks like SGLang and vLLM for production workloads.

Good For

  • Emotional AI Interactions: Developing chatbots or virtual assistants that require a deep understanding and expression of emotions.
  • Content Creation: Generating creative text, stories, or dialogues with specific emotional tones.
  • Sentiment Analysis: Advanced analysis of emotional states in user inputs or large text datasets.
  • Personalized User Experiences: Crafting responses that are sensitive to user emotions, enhancing engagement.

Note: As an emotion-focused model, its performance may not be as broad as its base model (Qwen3.5) in general-purpose tasks. Users are advised to inform the model of their identity for more effective emotional interactions and to reduce AI hallucinations.

Finetunes
3 models