|
|
--- |
|
|
license: mit |
|
|
language: |
|
|
- en |
|
|
library_name: transformers |
|
|
datasets: |
|
|
- Cynaptics/persona-chat |
|
|
--- |
|
|
# DialoGPT-Chat-Finetune |
|
|
|
|
|
This is a fine-tuned version of the DialoGPT model. It has been fine-tuned on persona-based data to generate human-like conversational responses. |
|
|
|
|
|
## Model Description |
|
|
The model is based on the DialoGPT architecture and has been fine-tuned for conversational tasks, specifically targeting persona-based interactions. |
|
|
|
|
|
## Model Details |
|
|
- **Architecture:** DialoGPT-medium |
|
|
- **Pretraining Data:** The original model was pretrained on a large corpus of text data. |
|
|
- **Fine-tuning Data:** This model was fine-tuned on persona-based conversational data. |
|
|
|
|
|
## Library |
|
|
- Framework: PyTorch |
|
|
- Model: DialoGPT-medium |
|
|
|
|
|
## Example Usage |
|
|
You can use this model via the Hugging Face `transformers` library. |
|
|
To use the fine-tuned model for text generation based on a persona, follow these steps: |
|
|
|
|
|
```python |
|
|
from transformers import pipeline, AutoModelForCausalLM, AutoTokenizer |
|
|
|
|
|
# Load the fine-tuned model and tokenizer |
|
|
model_name = "hello12w/persona_chatbot" |
|
|
model = AutoModelForCausalLM.from_pretrained(model_name) |
|
|
tokenizer = AutoTokenizer.from_pretrained(model_name) |
|
|
|
|
|
# Define the persona and prompt |
|
|
prompt = prompt = f""" |
|
|
|
|
|
Person B has the following Persona information. |
|
|
|
|
|
Persona of Person B: My name is Sarah and I'm a 28 year old software engineer. |
|
|
|
|
|
Persona of Person B: I love coding and developing new software applications. |
|
|
|
|
|
Persona of Person B: In my free time, I enjoy reading sci-fi novels and playing board games. |
|
|
|
|
|
Instruct: Person A and Person B are now having a conversation. |
|
|
|
|
|
Following the conversation below, write a response that Person B would say based on the above Persona information. |
|
|
|
|
|
Please carefully consider the flow and context of the conversation below, and use the Person B's Persona information appropriately to generate a response that you think is the most appropriate reply for Person B. |
|
|
|
|
|
Persona A: Hi Sarah, I heard you're working on a cool project at work. Can you tell me more about it? |
|
|
|
|
|
Output: |
|
|
|
|
|
""" |
|
|
input_ids = tokenizer(prompt, return_tensors="pt", truncation=True) |
|
|
attention_mask = input_ids.attention_mask |
|
|
input_ids = input_ids.input_ids |
|
|
|
|
|
# Inference |
|
|
with torch.no_grad(): |
|
|
outputs = model.generate( |
|
|
input_ids=input_ids, |
|
|
attention_mask=attention_mask, |
|
|
max_new_tokens=200, |
|
|
do_sample=True, |
|
|
top_p=0.95, |
|
|
temperature=0.9 |
|
|
) |
|
|
|
|
|
# Decode output tokens |
|
|
decoded_outputs = tokenizer.batch_decode(outputs, skip_special_tokens=True) |
|
|
output = decoded_outputs[0][len(prompt):] |
|
|
|
|
|
print(output) |
