This repository contains a version of the code for EM-LLM, published in ICLR 2025: [openreview link].
While typical LLMs struggle with processing extensive contexts, the human brain excels at organising and retrieving experiences spanning a lifetime. In this work, we introduce EM-LLM, an architecture that integrates key aspects of human episodic memory and event cognition into LLMs with no fine-tuning, enabling them to handle practically infinite context lengths while maintaining computational efficiency. EM-LLM organises sequences of tokens into coherent episodic events using a combination of Bayesian surprise and graph-theoretic boundary refinement in an online fashion. When needed, these events are retrieved through a two-stage memory process, combining similarity-based and temporally contiguous retrieval for efficient and human-like access to relevant information. Experiments on the LongBench and
Figure 1: Architecture of memory formation and retrieval in each LLM layer. Formation: Input sequence is initially segmented via surprise (purple dashed lines in ①), then segmentation is refined based on group theoretic metrics (green dashed lines in ②). Initial tokens and local context are preserved. Retrieval: via both k-NN search ③ and selecting contiguous events from episodic memory ④.
Click here more complete result tables.
Figure 2: (Left) EM-LLM$_S$ vs. RAG (NV-Embed-v2 retriever) vs. full-context, with LLaMA-3.1-8B as the base LLM, evaluated on LongBench. (Right) Comparison of various long-sequence methods (sorted based on their context window length) on an extended version of
Install requirements:
python3 -m pip install --upgrade pip
pip install -r "${base_dir}/requirements.txt"
pip install -e "${base_dir}/."
The YAML files used for configuration can be found in the config/ directory.
Here is a breakdown of each parameter included in these files:
verbose: false # print the question/prediction/answer after an example has been processed
compute_ppl: true # print and log perplexity for each example/chunk
return_block_size: true # print and log block size for each example/chunk
logging: true # save logs to output directory and label individual worker logs during multiprocessing
em_splitter: surprisal # method by which to split chunks into memory blocks (surprisal, random, sentence)
max_len: 2147483647 # maximum sequence length before truncation is used
chunk_size: 512 # size of chunked input during decoding
conv_type: mistral-inst # conversation template type
extended_passkey: 1024 # length to extend infinite-bench's passkey task to in terms of thousands of tokens (k)
model:
type: em-llm # Which model to use for inference (only em-llm is made available in this version)
path: mistralai/Mistral-7B-Instruct-v0.2 # HuggingFace model path
min_block_size: 8 # the smallest possible block size - blocks smaller than this will be expanded to this size
max_block_size: 128 # the biggest possible block size - blocks bigger than this will be split to this size
n_init: 128 # number of initial tokens to include in context window
n_local: 4096 # number of local tokens to include in context window
n_mem: 2048 # number of retrieved tokens to include in context window (includes both the similarity and contiguity buffers)
repr_topk: 4 # number of top-scoring tokens per memory unit considered as representative elements
max_cached_block: 512 # number of memory blocks to keep in GPU memory - must be greater than n_mem/min_block_size
exc_block_size: 512 # number of tokens queried at a time as an execution block - each execution block performs retrieval of n_mem tokens once
base: 1000000 # RoPE base
distance_scale: 1.0 # RoPE distance scale
surprisal_threshold_gamma: 1.0 # the standard-deviation scaling factor in the surprisal calculation (see paper)
min_free_cpu_memory: 100 # minimum amount CPU RAM (GB) to keep free when allocating memory blocks
disk_offload_threshold: 300000 # number of tokens in a sequence past which disk offloading should be used
vector_offload_threshold: 50000 # number of tokens in a sequence past which representative tokens should be offloaded to CPU memory
similarity_refinement_kwargs: # parameters relating directly to the boundary refinement step of our paper
similarity_refinement: false # whether to use boundary refinement or not
refine_with_buffer: true # if True, the adjacency matrix will include part of the neighbouring chunks in its calculation of the adjacency matrix - designed to make segmentations more compatible with neighbouring chunks, but also increases computation time
refine_from_layer: 20 # which layers to use when calculating the adjacency
similarity_metric: modularity # the metric to use as the objective during refinement: modularity or conductance (or intra_inter_sim but this doesn't work well so far)
contiguity_buffer_kwargs: # parameters relating directly to the contiguity buffer
use_contiguity_buffer: true # whether to use a contiguity buffer
contiguity_buffer_size: 0.3 # proportion of n_mem tokens to dedicate to the contiguity buffer
uniform_blocks: false # ignore em_splitter (above) and segment chunks into fixed-sized blocks of size max_block_size (above)
random_topk_blocks: false # retrieve random blocks rather than the topk most similar blocksData Preparation
We adopt
Response Generation You can evaluate EM-LLM by running the following command. You can also optionally pass in the following arguments to accomodate your hardware resources
bash scripts/run.sh
-m|--model # DEFAULT: mistral; OPTIONS: mistral,llama3,llama31,phi3_mini,phi35_mini - Which base LLM to use during evaluation.
-b|--benchmark # DEFAULT: long-bench; OPTIONS: long-bench,infinite-bench,passkey - Which benchmark to evaluate. Passkey evaluates an extended version of InfiniteBench's passkey retrieval task (see yaml for context length parameter).
-w|--world-size # DEFAULT: number of visible GPUs - Total number of GPUs to be used during evaluation.
-n|--num_gpus_per_job # DEFAULT: 1 - How many GPUs to attribute to each job. If >1, model layers will be evenly spread over multiple GPUs.
-r|--rank_offset # DEFAULT: 0 - Ignores the first n GPUs visible to the script. Useful when running multiple experiments on a single node.
-o|--allow_disk_offload # DEFAULT: False - Whether to allow dynamic disk offloading of memory blocks or not (see the our paper's Appendix for more details). In single-GPU instances this will offload the representative tokens to CPU memory as well.
If you find EM-LLM useful, please cite the following paper:
@inproceedings{fountas2025humaninspired,
title={Human-inspired Episodic Memory for Infinite Context {LLM}s},
author={Zafeirios Fountas and Martin Benfeghoul and Adnan Oomerjee and Fenia Christopoulou and Gerasimos Lampouras and Haitham Bou Ammar and Jun Wang},
booktitle={The Thirteenth International Conference on Learning Representations},
year={2025},
url={https://openreview.net/forum?id=BI2int5SAC}
}