From 8185e9c3ac51c08bf51c83e0db7b962e243e9d29 Mon Sep 17 00:00:00 2001 From: Thomas Limbacher Date: Thu, 22 Oct 2020 18:55:16 +0200 Subject: [PATCH] Bumped version number to 1.1 --- README.md | 20 ++++++++++++++++++-- 1 file changed, 18 insertions(+), 2 deletions(-) diff --git a/README.md b/README.md index 487296e..a5cfb73 100644 --- a/README.md +++ b/README.md @@ -1,6 +1,6 @@ # H-Mem: Harnessing synaptic plasticity with Hebbian Memory Networks This is the code used in the paper "[H-Mem: Harnessing synaptic plasticity with Hebbian Memory -Networks](https://www.biorxiv.org/content/10.1101/2020.07.01.180372v1)" for training H-Mem on a single-shot +Networks](https://www.biorxiv.org/content/10.1101/2020.07.01.180372v2)" for training H-Mem on a single-shot image association task and on the bAbI question-answering tasks. ![H-Mem schema](https://i.imgur.com/fK3UWaP.png) @@ -48,5 +48,21 @@ python babi_task_single.py --task_id=4 --memory_size=20 --epochs=50 --logging=1 will train the model with an associative memory of size 20 on task 4 for 50 epochs. The results will be stored in `results/`. +### Memory-dependent memorization +In our extended model we have added an 'read-before-write' step. This model will be used if the +command line argument `--read_before_write` is set to `1`. Run the following command + +```bash +python babi_task_single.py --task_id=16 --epochs=250 --read_before_write=1 +``` + +to start training on bAbI task 16 in the 10k training examples setting (note that we trained the extended +model for 250 epochs---instead of 100 epochs). You should get an accuracy of about 100% on this task. Compare +to the original model, which does not solve task 16, by running the following command + +```bash +python babi_task_single.py --task_id=16 --epochs=250 +``` + ## References -* Limbacher, T., Legenstein, R. (2020). H-Mem: Harnessing synaptic plasticity with Hebbian Memory Networks bioRxiv https://www.biorxiv.org/content/10.1101/2020.07.01.180372v1 +* Limbacher, T., Legenstein, R. (2020). H-Mem: Harnessing synaptic plasticity with Hebbian Memory Networks bioRxiv https://www.biorxiv.org/content/10.1101/2020.07.01.180372v2