import gradio as gr

#api = gr.Interface.load("huggingface/EleutherAI/gpt-j-6B")
api = gr.Interface.load("models/bigscience/bloom")


def complete_with_gpt(text):
    # Use the last 50 characters of the text as context
    # return text[:-50] + api(text[-50:])
    # Use the last 100 characters of the text as context
    return text[:-100] + api(text[-100:])


with gr.Blocks() as demo:
    with gr.Row():
        textbox = gr.Textbox(placeholder="Type here and press enter...", lines=14)
        with gr.Column():
            btn = gr.Button("Generate")

    btn.click(complete_with_gpt, textbox, textbox)

    with gr.Row():
        gr.Markdown("""
        
# Outline of Exciting AI Developments! 🤖💻🔬

Here is an outline of some of the most exciting recent developments in AI:

## Language Models 🗣️

🏆 Bloom sets new record for most performant and efficient AI model in science! 🌸

### Comparison of Large Language Models

| Model Name        | Model Size (in Parameters) |
| ----------------- | -------------------------- |
| BigScience-tr11-176B | 176 billion |
| GPT-3             | 175 billion               |
| OpenAI's DALL-E 2.0 | 500 million               |
| NVIDIA's Megatron | 8.3 billion               |
| Transformer-XL    | 250 million               |
| XLNet             | 210 million               |

## ChatGPT Datasets 📚

- WebText
- Common Crawl
- BooksCorpus
- English Wikipedia
- Toronto Books Corpus
- OpenWebText

## Big Science Model 🚀

- 📜 Papers:
  1. BLOOM: A 176B-Parameter Open-Access Multilingual Language Model [Paper](https://arxiv.org/abs/2211.05100)
  2. Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism [Paper](https://arxiv.org/abs/1909.08053)
  3. 8-bit Optimizers via Block-wise Quantization [Paper](https://arxiv.org/abs/2110.02861)
  4. Train Short, Test Long: Attention with Linear Biases Enables Input Length Extrapolation [Paper](https://arxiv.org/abs/2108.12409)
  5. [Other papers related to Big Science](https://huggingface.co/models?other=doi:10.57967/hf/0003)
  6. [217 other models optimized for use with Bloom](https://huggingface.co/models?other=bloom)

- 📚 Datasets:
  1. [Universal Dependencies](https://paperswithcode.com/dataset/universal-dependencies)
  2. [WMT 2014](https://paperswithcode.com/dataset/wmt-2014)
  3. [The Pile](https://paperswithcode.com/dataset/the-pile)
  4. [HumanEval](https://paperswithcode.com/dataset/humaneval)
  5. [FLORES-101](https://paperswithcode.com/dataset/flores-101)
  6. [CrowS-Pairs](https://paperswithcode.com/dataset/crows-pairs)
  7. [WikiLingua](https://paperswithcode.com/dataset/wikilingua)
  8. [MTEB](https://paperswithcode.com/dataset/mteb)
  9. [xP3](https://paperswithcode.com/dataset/xp3)
  10. [DiaBLa](https://paperswithcode.com/dataset/diabla)

# Deep RL ML Strategy 🧠

The AI strategies are:
- Language Model Preparation using Human Augmented with Supervised Fine Tuning 🤖
- Reward Model Training with Prompts Dataset Multi-Model Generate Data to Rank 🎁
- Fine Tuning with Reinforcement Reward and Distance Distribution Regret Score 🎯
- Proximal Policy Optimization Fine Tuning 🤝
- Variations - Preference Model Pretraining 🤔
- Use Ranking Datasets Sentiment - Thumbs Up/Down, Distribution 📊
- Online Version Getting Feedback 💬
- OpenAI - InstructGPT - Humans generate LM Training Text 🔍
- DeepMind - Advantage Actor Critic Sparrow, GopherCite 🦜
- Reward Model Human Prefence Feedback 🏆

For more information on specific techniques and implementations, check out the following resources:
- OpenAI's paper on [GPT-3](https://arxiv.org/abs/2005.14165) which details their Language Model Preparation approach
- DeepMind's paper on [SAC](https://arxiv.org/abs/1801.01290) which describes the Advantage Actor Critic algorithm
- OpenAI's paper on [Reward Learning](https://arxiv.org/abs/1810.06580) which explains their approach to training Reward Models
- OpenAI's blog post on [GPT-3's fine-tuning process](https://openai.com/blog/fine-tuning-gpt-3/)

        """)

demo.launch()