Sublime

BetterWhisperX: An open-source speech recognition tool for fast transcription, word-level timestamps, and multi-speaker recognition, using GPU acceleration and batch processing https://t.co/2nH0j1Z05j

Rottnest : Data Lake Indices

You don't need ElasticSearch or some vector database to do full text search or vector search. Parquet + Rottnest is all you need. Rottnest is like Postgres indices for Parquet. Read more on what it can do for e.g. logs here.

Installation

Local installation: pip install rottnest .

Rottnest supports many different index typ... See more

Massive Language Models Can Be Accurately Pruned in One-Shot abs: https://t.co/geA8bbWYFA... See more

gsplat An Open-Source Library for Gaussian Splatting discuss: https://t.co/ePA5QQt0Ph gsplat... See more

SAPG: Split and Aggregate Policy Gradients Significantly higher performance across a variety of challenging environments where vanilla PPO and other strong baselines fail to achieve high performance proj: https://t.co/HGFlqF8p8J abs:... See more

微软推出了RAG的新方法：LazyGraphRAG，核心是成本非常低，数据索引成本只有完整GraphRAG的0.1%，生成结果的准确率、查询效率很高 性能： 1、索引成本，与向量RAG持平，为GraphRAG的0.1% 2、在本地查询上，与向量RAG相当的查询成本下，优于所有方法，包括长上下文向量RAG、GraphRAG DRIFT搜索、GraphRAG本地搜索 3、在全局查询上，质量与GraphRAG全局搜索相当，查询成本降低了700多倍 ___LINEBREAK... See more

"Right before our release, we are informed with a concurrent blogpost (Super-HOT kaiokendev (2023)) that also interpolates positional encoding in RoPE ot extend the context window from 2k and 8k" scooped 🌭🎣 https://t.co/O9a2XACI6H