facebookresearch / memory
Memory layers use a trainable key-value lookup mechanism to add extra parameters to a model without increasing FLOPs. Conceptually, sparsely activated memory layers complement compute-heavy dense feed-forward layers, providing dedicated capacity to store and retrieve information cheaply.
☆311Updated 3 months ago
Alternatives and similar repositories for memory:
Users that are interested in memory are comparing it to the libraries listed below
- 🌾 OAT: A research-friendly framework for LLM online alignment, including preference learning, reinforcement learning, etc.☆300Updated last week
- PyTorch building blocks for the OLMo ecosystem☆177Updated this week
- PyTorch implementation of models from the Zamba2 series.☆178Updated 2 months ago
- Repo for "LoLCATs: On Low-Rank Linearizing of Large Language Models"☆226Updated 2 months ago
- ☆163Updated last month
- ☆401Updated this week
- [ICML 2024] CLLMs: Consistency Large Language Models☆388Updated 4 months ago
- ☆173Updated 3 months ago
- A framework to study AI models in Reasoning, Alignment, and use of Memory (RAM).☆210Updated last week
- Towards Economical Inference: Enabling DeepSeek's Multi-Head Latent Attention in Any Transformer-based LLMs☆149Updated last week
- Normalized Transformer (nGPT)☆164Updated 4 months ago
- An extension of the nanoGPT repository for training small MOE models.☆109Updated 3 weeks ago
- LongRoPE is a novel method that can extends the context window of pre-trained LLMs to an impressive 2048k tokens.☆212Updated 7 months ago
- PyTorch implementation of Infini-Transformer from "Leave No Context Behind: Efficient Infinite Context Transformers with Infini-attention…☆287Updated 10 months ago
- Q-GaLore: Quantized GaLore with INT4 Projection and Layer-Adaptive Low-Rank Gradients.☆195Updated 8 months ago
- Muon optimizer: +>30% sample efficiency with <3% wallclock overhead☆539Updated last week
- Efficient LLM Inference over Long Sequences☆365Updated last month
- 🚀 Efficiently (pre)training foundation models with native PyTorch features, including FSDP for training and SDPA implementation of Flash…☆234Updated this week
- [NeurIPS 2024] Official Repository of The Mamba in the Llama: Distilling and Accelerating Hybrid Models☆209Updated last month
- Build your own visual reasoning model☆320Updated last week
- Tree Attention: Topology-aware Decoding for Long-Context Attention on GPU clusters☆125Updated 4 months ago
- A project to improve skills of large language models☆260Updated this week
- OpenCoconut implements a latent reasoning paradigm where we generate thoughts before decoding.☆169Updated 2 months ago
- Implementation of 🥥 Coconut, Chain of Continuous Thought, in Pytorch☆161Updated 3 months ago
- Code for "LayerSkip: Enabling Early Exit Inference and Self-Speculative Decoding", ACL 2024☆279Updated last month
- Benchmark and research code for the paper SWEET-RL Training Multi-Turn LLM Agents onCollaborative Reasoning Tasks☆140Updated this week
- Pretraining code for a large-scale depth-recurrent language model☆709Updated 2 weeks ago
- Chain of Experts (CoE) enables communication between experts within Mixture-of-Experts (MoE) models☆151Updated 2 weeks ago
- Implementation of 💍 Ring Attention, from Liu et al. at Berkeley AI, in Pytorch☆509Updated 5 months ago
- ☆195Updated 3 months ago