Hebrew Gpt Neo Tiny

🚀 Hebrew GPT-Neo Tiny

A Hebrew text generation model based on EleutherAI's gpt-neo, offering efficient and effective Hebrew text generation capabilities.

✨ Features

• Trained on a TPUv3 - 8 provided via the TPU Research Cloud Program.
• Utilizes a variety of Hebrew corpuses for training.

📦 Installation

The installation steps are included in the usage sample code. You can install the necessary libraries using the following command:

!pip install tokenizers==0.10.2 transformers==4.6.0

💻 Usage Examples

Basic Usage

The following is a simple usage example of the Hebrew GPT - Neo Tiny model:

!pip install tokenizers==0.10.2 transformers==4.6.0

from transformers import AutoTokenizer, AutoModelForCausalLM
  
tokenizer = AutoTokenizer.from_pretrained("Norod78/hebrew-gpt_neo-tiny")
model = AutoModelForCausalLM.from_pretrained("Norod78/hebrew-gpt_neo-tiny", pad_token_id=tokenizer.eos_token_id)

prompt_text = "אני אוהב שוקולד ועוגות"
max_len = 512
sample_output_num = 3
seed = 1000

import numpy as np
import torch

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
n_gpu = 0 if torch.cuda.is_available()==False else torch.cuda.device_count()

print(f"device: {device}, n_gpu: {n_gpu}")

np.random.seed(seed)
torch.manual_seed(seed)
if n_gpu > 0:
    torch.cuda.manual_seed_all(seed)

model.to(device)

encoded_prompt = tokenizer.encode(
    prompt_text, add_special_tokens=False, return_tensors="pt")

encoded_prompt = encoded_prompt.to(device)

if encoded_prompt.size()[-1] == 0:
        input_ids = None
else:
        input_ids = encoded_prompt

print("input_ids = " + str(input_ids))

if input_ids != None:
  max_len += len(encoded_prompt[0])
  if max_len > 1024:
    max_len = 1024

print("Updated max_len = " + str(max_len))

stop_token = "<|endoftext|>"
new_lines = "\n\n\n"

sample_outputs = model.generate(
    input_ids,
    do_sample=True, 
    max_length=max_len, 
    top_k=50, 
    top_p=0.95, 
    num_return_sequences=sample_output_num
)

print(100 * '-' + "\n\t\tOutput\n" + 100 * '-')
for i, sample_output in enumerate(sample_outputs):

  text = tokenizer.decode(sample_output, skip_special_tokens=True)
  
  # Remove all text after the stop token
  text = text[: text.find(stop_token) if stop_token else None]

  # Remove all text after 3 newlines
  text = text[: text.find(new_lines) if new_lines else None]

  print("\n{}: {}".format(i, text))
  print("\n" + 100 * '-')

Advanced Usage

You can access the Google Colab Notebook for more advanced usage scenarios here.

📚 Documentation

Datasets

1. An assortment of various Hebrew corpuses - Available here.
2. oscar / unshuffled_deduplicated_he - Homepage | Dataset Permalink.
   • The Open Super - large Crawled ALMAnaCH coRpus is a huge multilingual corpus obtained by language classification and filtering of the Common Crawl corpus using the goclassy architecture.

Training Config

The training configuration is available here.

📄 License

This project is licensed under the MIT license.