EMO-phi-128k

EMO-phi-128k

EMO-phi-128k is an emotional intelligence conversational AI model fine-tuned from Microsoft's Phi-3-mini-128k-instruct model. It is designed to engage in open-ended dialogue while exhibiting emotional understanding and emotional intelligence capabilities.

Model Details

• Developer: OEvortex
• Model Type: Transformer-based language model
• Language: English
• License: MIT
• Base Model: microsoft/Phi-3-mini-128k-instruct

Model Description

EMO-phi-128k is a 128k parameter conversational AI model that has been fine-tuned to incorporate emotional intelligence and emotional understanding capabilities. It aims to engage in emotionally aware and contextual dialogue by recognizing and responding appropriately to the emotional tones and sentiments expressed by the user.

While inheriting the strong language understanding and generation capabilities of its base model, EMO-phi-128k has been specifically optimized for emotional intelligence tasks through additional fine-tuning on emotional dialogue datasets.

Intended Uses

• Emotional Support / Conversational Companion
• Customer Service Chatbots (with emotional intelligence)
• Creative Writing Assistance (with emotional awareness)
• Psychological/Therapeutic Applications

Limitations and Risks

As an AI system, EMO-phi-128k may exhibit biases present in its training data. Its true emotional intelligence capabilities are not fully known or verified. The model should be used with caution, especially in sensitive or high-stakes applications involving mental health, therapy, or counseling. Proper human oversight is recommended.

Additionally, like all language models, EMO-phi-128k is susceptible to generating harmful, biased, or explicit content if prompted in an unsafe manner. Safety considerations should be taken into account when deploying or interacting with the model.

How to Use

You can load and use the EMO-phi-128k model with the Transformers library in Python:

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline

torch.random.manual_seed(0)

model = AutoModelForCausalLM.from_pretrained(
    "OEvortex/EMO-phi-128k", 
    device_map="cuda", 
    torch_dtype="auto", 
    trust_remote_code=True, 
)
tokenizer = AutoTokenizer.from_pretrained("microsoft/Phi-3-mini-128k-instruct")

messages = [
    {"role": "system", "content": "You are a helpful Emotional intelligence named as EMO-phi, remember to always answer users question in EMO style."},
    {"role": "user", "content": "My best friend recently lost their parent to cancer after a long battle. They are understandably devastated and struggling with grief."},
]

pipe = pipeline(
    "text-generation",
    model=model,
    tokenizer=tokenizer,
)

generation_args = {
    "max_new_tokens": 500,
    "return_full_text": False,
    "temperature": 0.6,
    "do_sample": True,
}

output = pipe(messages, **generation_args)
print(output[0]['generated_text'])