Mesolitica
We develop Multimodality Artificial Intelligence for South East Asia.
Pinned Loading
Repositories
Showing 10 of 45 repositories
- Chunk-loss-LoRA Public
Fused kernel chunk loss to include LoRA to reduce memory, support DeepSpeed ZeRO3.
- initial-paged-flash-attention Public Forked from huggingface/transformers
🤗 Transformers: State-of-the-art Machine Learning for Pytorch, TensorFlow, and JAX.
- transformers-openai-api Public
Lightweight continuous batching OpenAI compatibility using HuggingFace Transformers include T5 and Whisper.
- accelerate-torch-compile-speechlm Public Forked from huggingface/accelerate
🚀 A simple way to launch, train, and use PyTorch models on almost any device and distributed configuration, automatic mixed precision (including fp8), and easy-to-configure FSDP and DeepSpeed support
- dynamic-batch-TTS-pipeline Public
Dynamic batching for Speech Enhancement, Speech Tokenizer and TTS.
- picotron-zero1 Public Forked from huggingface/picotron
Minimalistic 4D-parallelism distributed training framework for education purpose
