Spaces:

CaioXapelaum

/

GGUF-Playground

like 7

Runtime error

Title: GGUF Playground: Chat with LLMs using llama.cpp python bindings (llama-cpp-python)

Space: GGUF Playground

GGUF Playground is a space designed for efficient inference of large language models (LLMs) using the GGUF format, optimized for performance across various hardware setups. While the space is still growing, with a small but active user base, its purpose is to facilitate the deployment and experimentation with quantized models such as Llama-3.1, Mistral, and others.

Currently, the space has reached the 100GB storage limit, which poses significant constraints on model availability, caching, and overall functionality. Even though the space has only a few likes so far, it is designed with scalability in mind and is useful for users and single/small developers working with tiny models in constrained environments due to accessibi.

Persistent Storage Request:
I am requesting an increase in persistent storage capacity beyond the current 100GB limit to ensure the space remains usable and effective, despite its relatively modest size. The reasons for this request include:

1. Model Storage and Caching: Each LLM model, especially those in GGUF format, takes up substantial disk space. Persistent storage ensures these models remain loaded and accessible, preventing frequent re-downloading and reducing initialization times. Without additional storage, the space will struggle to support even a small number of models simultaneously.

2. Growing Model Sizes: As newer and more complex models are released, such as Llama-2 and Mistral, their larger sizes will quickly exceed the current storage capacity, limiting the ability to provide multiple models for user experimentation.

3. User Experience and Space Stability: Even though the current user base is small, it is essential to offer a seamless experience by maintaining cached models, outputs, and session data. Without sufficient storage, users will face delays and interruptions, which may hinder growth and adoption.

4. Future Growth Potential: Although GGUF Playground currently has only 5 likes, there is strong potential for the space to grow as the demand for efficient LLM deployment increases. Scaling storage now will allow the space to better serve future users as it gains traction.

Proposed Storage:
I am requesting any tier of persistent storage above 100GB to handle current demands and future growth. This will provide the necessary capacity to store larger models and ensure smooth operation for users, no matter the space’s current popularity.

Thank you for considering this request. Expanding the storage capacity will significantly enhance the space’s ability to serve its community and prepare for future growth.

Best regards,
Caio.