File size: 3,631 Bytes
e6695af
 
 
 
 
 
 
 
5d7af2a
3a3e2bb
e4b3146
3a3e2bb
e6695af
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6aa47a4
 
e6695af
0e0be5a
 
 
e6695af
 
 
 
0e0be5a
 
e6695af
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d4a3cd5
e6695af
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
---
license: mit
datasets:
- neural-bridge/rag-dataset-12000
- neural-bridge/rag-dataset-1200
language:
- en
---
# VERY IMPORTANT
- This model is in alpha phase and is NOT yet recommended for use.
- This model is obsolete today, recommended model [here.](https://huggingface.co/BueormLLC/RAGPT-2_Turbo)
# RAGPT-2 (unfunctional): Fine-tuned GPT-2 for Context-Based Question Answering

## Model Description

RAGPT-2 is a fine-tuned version of [GPT-2 small](https://huggingface.co/BueormLLC/CleanGPT), specifically adapted for context-based question answering tasks. This model has been trained to generate relevant answers based on a given context and question, similar to a Retrieval-Augmented Generation (RAG) system.

### Key Features

- Based on the GPT-2 small architecture (124M parameters)
- Fine-tuned on the "neural-bridge/rag-dataset-12000" and others dataset from Hugging Face
- Capable of generating answers based on provided context and questions
- Suitable for various question-answering applications

## Training Data

The model was fine-tuned using the "neural-bridge/rag-dataset-12000" and "neural-bridge/rag-dataset-1200" dataset, which contains:
- Context passages
- Questions related to the context
- Corresponding answers

## Fine-tuning Process

The fine-tuning process involved:
1. Loading the pre-trained GPT-2 small model
2. Preprocessing the dataset to combine context, question, and answer into a single text
3. Training the model to predict the next token given the context and question

### Hyperparameters

- Base model: GPT-2 small
- Number of training epochs: 8
- Batch size: 4
- Learning rate: Default AdamW optimizer settings
- Max sequence length: 512 tokens

## Usage

To use the model:

```python
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("BueormLLC/RAGPT-2_unfunctional")
model = AutoModelForCausalLM.from_pretrained("BueormLLC/RAGPT-2_unfunctional")

context = "Mount Everest is the highest mountain in the world, with a height of 8,848 meters."
question = "What is the height of Mount Everest?"
input_text = f"Context: {context}\nquestion: {question}\nanswer:"

input_ids = tokenizer.encode(input_text, return_tensors="pt")
output = model.generate(input_ids, max_length=150, num_return_sequences=1)
answer = tokenizer.decode(output[0], skip_special_tokens=True)

print(f"Respuesta generada: {answer}")
```

## Limitations

- The model's knowledge is limited to its training data and the base GPT-2 model.
- It may sometimes generate irrelevant or incorrect answers, especially for topics outside its training domain.
- The model does not have access to external information or real-time data.

## Ethical Considerations

Users should be aware that this model, like all language models, may reflect biases present in its training data. It should not be used as a sole source of information for critical decisions.

## Future Improvements

- Fine-tuning on a larger and more diverse dataset
- Experimenting with larger base models (e.g., GPT-2 medium or large)
- Implementing techniques to improve factual accuracy and reduce hallucinations

## Support us

- [Paypal](https://paypal.me/bueorm)
- [Patreon](https://patreon.com/bueorm)
### We appreciate your support, without you we could not do what we do.

## Citation

If you use this model in your research, please cite:

```
@misc{RAGPT,
  author = {Bueorm},
  title = {RAGPT-2: Fine-tuned GPT-2 for Context-Based Question Answering},
  year = {2024},
  publisher = {GitHub},
  journal = {None},
  howpublished = {\url{https://huggingface.co/BueormLLC/RAGPT-2_unfunctional}}
}
```