Category

Continue

Open-source AI coding assistant for VS Code and JetBrains that bundles autocomplete, chat, inline edit, and an agent mode behind one config, letting each capability use any model provider rather than a single locked-in vendor.

Cognee

Gives AI agents persistent long-term memory: ingests documents in any format and continuously builds a self-hosted knowledge graph fusing vector embeddings, graph reasoning, and ontology grounding, so agents recall and reason over connected facts.

Maxun

Turns any website into structured data or an API without code: record clicks once to capture lists and tables, or describe fields in plain language for AI extraction. Also crawls full sites, scrapes pages to Markdown, and runs filtered searches.

MLX

Provides a NumPy-like array framework for building and training ML on Apple silicon, with Python, C/C++, and Swift APIs plus PyTorch-style higher-level modules. Features lazy evaluation, composable AD/vectorization, and a unified-memory multi-device model so arrays can be used on CPU and GPU without explicit copies.

Mamba

A selective State Space Model architecture and PyTorch implementation for linear-time sequence modeling. Hardware-aware, designed for information-dense tasks (e.g. language modeling), with pretrained weights on Hugging Face; requires CUDA-enabled PyTorch.

VLMEvalKit

Runs one-command evaluation of vision-language models across 80+ multimodal benchmarks, handling data download, inference, and metric scoring in a single pass. Supports 220+ LMMs; adding a new model means writing one generate_inner() function.

LitServe | Deploy any AI model Lightning fast

Builds custom AI inference servers in pure Python on top of FastAPI, keeping full control over request logic while batching, GPU autoscaling, streaming, and OpenAI-spec endpoints come built in. Claims a 2x+ throughput edge over plain FastAPI.

torchtitan

Provides a PyTorch-native platform for experimenting with and scaling generative AI training, including composable parallelism, checkpointing, float8, logging, and Llama recipes.

Flash Linear Attention

Triton kernels and PyTorch layers for linear-attention, state-space, and sparse-attention token mixers (GLA, RWKV, Mamba2, GSA) as drop-in replacements for multihead attention. Runs on NVIDIA, AMD, and Intel GPUs with Hugging Face support.

LLM Transparency Tool

Traces how Transformer LLMs route information from input to output, attributing each block's effect to individual attention heads and feed-forward neurons. Click any edge to see what a head promotes or suppresses in vocabulary space.

Claude-Mem

Claude-Mem is a persistent memory compression system built for Claude Code. It automatically captures tool usage observations during coding sessions, generates semantic summaries using Claude's agent-sdk, and injects relevant context into future sessions to maintain continuity of project knowledge.

CrewAI

Orchestrates teams of role-based autonomous agents that collaborate on multi-step tasks, plus event-driven Flows for deterministic control. Built from scratch with no LangChain dependency; runs 450M+ agentic workflows monthly.