facebookresearch / memoryLinks
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.
☆366Updated last year
Alternatives and similar repositories for memory
Users that are interested in memory are comparing it to the libraries listed below
Sorting:
- 🌾 OAT: A research-friendly framework for LLM online alignment, including reinforcement learning, preference learning, etc.☆609Updated this week
- Tina: Tiny Reasoning Models via LoRA☆312Updated 3 months ago
- [ICML 2024] CLLMs: Consistency Large Language Models☆411Updated last year
- Parallel Scaling Law for Language Model — Beyond Parameter and Inference Time Scaling☆467Updated 7 months ago
- Repo for "LoLCATs: On Low-Rank Linearizing of Large Language Models"☆249Updated 11 months ago
- Code for "LayerSkip: Enabling Early Exit Inference and Self-Speculative Decoding", ACL 2024☆351Updated 8 months ago
- Chain of Experts (CoE) enables communication between experts within Mixture-of-Experts (MoE) models☆228Updated 2 months ago
- PyTorch implementation of Infini-Transformer from "Leave No Context Behind: Efficient Infinite Context Transformers with Infini-attention…☆294Updated last year
- A project to improve skills of large language models☆756Updated this week
- A framework to study AI models in Reasoning, Alignment, and use of Memory (RAM).☆340Updated 3 weeks ago
- OpenCoconut implements a latent reasoning paradigm where we generate thoughts before decoding.☆174Updated 11 months ago
- PyTorch building blocks for the OLMo ecosystem☆681Updated this week
- LongRoPE is a novel method that can extends the context window of pre-trained LLMs to an impressive 2048k tokens.☆276Updated 2 months ago
- ☆224Updated last month
- Implementation of 🥥 Coconut, Chain of Continuous Thought, in Pytorch☆181Updated 6 months ago
- Public repository for "The Surprising Effectiveness of Test-Time Training for Abstract Reasoning"☆341Updated 2 months ago
- ☆204Updated last year
- Official PyTorch implementation for Hogwild! Inference: Parallel LLM Generation with a Concurrent Attention Cache☆137Updated 4 months ago
- Pretraining and inference code for a large-scale depth-recurrent language model☆857Updated last week
- Code to train and evaluate Neural Attention Memory Models to obtain universally-applicable memory systems for transformers.☆346Updated last year
- An extension of the nanoGPT repository for training small MOE models.☆224Updated 10 months ago
- Reproducible, flexible LLM evaluations☆316Updated last month
- Exploring Applications of GRPO☆251Updated 4 months ago
- Code for NeurIPS'24 paper 'Grokked Transformers are Implicit Reasoners: A Mechanistic Journey to the Edge of Generalization'☆234Updated 5 months ago
- Q-GaLore: Quantized GaLore with INT4 Projection and Layer-Adaptive Low-Rank Gradients.☆201Updated last year
- Towards Economical Inference: Enabling DeepSeek's Multi-Head Latent Attention in Any Transformer-based LLMs☆198Updated last month
- [NeurIPS 2024] Official Repository of The Mamba in the Llama: Distilling and Accelerating Hybrid Models☆233Updated 2 months ago
- Single File, Single GPU, From Scratch, Efficient, Full Parameter Tuning library for "RL for LLMs"☆575Updated 3 months ago
- Code for the paper: "Learning to Reason without External Rewards"☆385Updated 6 months ago
- Minimal hackable GRPO implementation☆309Updated 11 months ago