https://github.com/kyegomez/reka-torch

Implementation of the model: "Reka Core, Flash, and Edge: A Series of Powerful Multimodal Language Models" in PyTorch
https://github.com/kyegomez/reka-torch

Last synced: 6 months ago
JSON representation

Implementation of the model: "Reka Core, Flash, and Edge: A Series of Powerful Multimodal Language Models" in PyTorch

• Host: GitHub
• URL: https://github.com/kyegomez/reka-torch
• Owner: kyegomez
• License: mit
• Created: 2024-04-15T20:29:34.000Z (over 1 year ago)
• Default Branch: main
• Last Pushed: 2025-03-17T21:20:24.000Z (7 months ago)
• Last Synced: 2025-03-23T11:18:00.934Z (7 months ago)
• Language: Python
• Size: 2.16 MB
• Stars: 30
• Watchers: 1
• Forks: 3
• Open Issues: 4
• Metadata Files:
  • Readme: README.md
  • Funding: .github/FUNDING.yml
  • License: LICENSE

Awesome Lists containing this project

README

          
[![Multi-Modality](agorabanner.png)](https://discord.gg/qUtxnK2NMf)

# Reka Torch

Implementation of the model: "Reka Core, Flash, and Edge: A Series of Powerful Multimodal Language Models" in PyTorch. [PAPER LINK](https://publications.reka.ai/reka-core-tech-report.pdf)

## Install

`pip3 install -U reka-torch`

## Usage

```python

import torch  # Importing the torch library

from reka_torch.model import Reka  # Importing the Reka model from the reka_torch package

text = torch.randint(0, 10000, (2, 512))  # Generating a random tensor of shape (2, 512) with values between 0 and 10000

img = torch.randn(2, 3, 224, 224)  # Generating a random tensor of shape (2, 3, 224, 224) with values

audio = torch.randn(2, 1000)  # Generating a random tensor of shape (2, 1000) with values

video = torch.randn(2, 3, 16, 224, 224)  # Generating a random tensor of shape (2, 3, 16, 224, 224) with values

model = Reka(512)  # Creating an instance of the Reka model with input size 512

out = model(text, img, audio, video)  # Forward pass through the model with the input tensors

print(out.shape)  # Printing the shape of the output tensor

```

# License

MIT