---
base_model:
- microsoft/Phi-3-mini-128k-instruct
- marketeam/Phi-Marketing
- OEvortex/EMO-phi-128k
library_name: transformers
tags:
- mergekit
- merge
license: apache-2.0
---
# Phi-3-mini-EmoMarketing-DELLA

This is a model based on [microsoft/Phi-3-mini-128k-instruct](https://huggingface.co/microsoft/Phi-3-mini-128k-instruct) created by merging two fine-tuned versions together, one checkpoint for a domain-specific marketing fine tune, and one for emotional intelligence conversational setting. 

## 🤏 Models Merged

This is a merge of pre-trained language models created using [mergekit](https://github.com/cg123/mergekit). This model was merged using the [DELLA](https://arxiv.org/abs/2406.11617) merge method using [marketeam/Phi-Marketing](https://huggingface.co/marketeam/Phi-Marketing) as a base.

The following models were included in the merge:
* [marketeam/Phi-Marketing](https://huggingface.co/marketeam/Phi-Marketing) <- Base
* [OEvortex/EMO-phi-128k](https://huggingface.co/OEvortex/EMO-phi-128k)

## 🧩 Configuration

The following YAML configuration was used to produce this model:

```yaml
models:
  - model: marketeam/Phi-Marketing
    parameters:
      weight: 1.0
  - model: OEvortex/EMO-phi-128k
    parameters:
      weight: 1.0
merge_method: della
base_model: marketeam/Phi-Marketing
parameters:
  density: 0.7
  lambda: 1.1
  epsilon: 0.2

```

## 💻 Usage

```python
from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

# Load the tokenizer and model
tokenizer = AutoTokenizer.from_pretrained("AdamLucek/Phi-3-mini-EmoMarketing-DELLA", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    "AdamLucek/Phi-3-mini-EmoMarketing-DELLA",
    device_map="cuda",
    torch_dtype=torch.bfloat16,
    trust_remote_code=True
)

# Prepare the input text
input_text = "What are specific actionable ways to market products to technical software engineers with an emotional angle?"
input_ids = tokenizer(input_text, return_tensors="pt").to("cuda")

# Generate the output
outputs = model.generate(
    **input_ids,
    max_new_tokens=256,
    pad_token_id=tokenizer.eos_token_id
)

# Decode and print the generated text
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
```

**output**
>Hello there! 😊 I'd be happy to help you with that. When it comes to marketing products to technical software engineers with an emotional angle, there are several specific actionable ways to approach this. Here are a few ideas:
>1. Highlight the impact of the product on the user's personal and professional life. Emphasize how the product can solve a specific problem or improve the user's overall experience, and how it can positively impact their emotions and well-being.
>2. Use storytelling to create an emotional connection with the audience. Share real-life stories or testimonials from users who have experienced positive emotional outcomes as a result of using the product.
>3. Focus on the user's passions and interests. Understand what motivates and inspires technical software engineers, and tailor the marketing message to resonate with their emotional drivers.
>4. Use visual and sensory elements to evoke emotions. Incorporate imagery, colors, and sounds that align with the emotional tone you want to convey, and create a visually appealing and emotionally