Daniel Tiarks's starred repositories
RWKV-LM
RWKV is an RNN with transformer-level LLM performance. It can be directly trained like a GPT (parallelizable). So it's combining the best of RNN and transformer - great performance, fast inference, saves VRAM, fast training, "infinite" ctx_len, and free sentence embedding.
Awesome-Multimodal-Large-Language-Models
:sparkles::sparkles:Latest Advances on Multimodal Large Language Models
pyscaffold
🛠 Python project template generator with batteries included
P-tuning-v2
An optimized deep prompt tuning strategy comparable to fine-tuning across scales and tasks
xdocreport
XDocReport means XML Document reporting. It's Java API to merge XML document created with MS Office (docx) or OpenOffice (odt), LibreOffice (odt) with a Java model to generate report and convert it if you need to another format (PDF, XHTML...).
recurrent-memory-transformer
[NeurIPS 22] [AAAI 24] Recurrent Transformer-based long-context architecture.
string2string
String-to-String Algorithms for Natural Language Processing
Layout2Graph
An official implementation of paper "Paragraph2Graph: A Language-independent GNN-based framework for layout analysis"