From e0201ed007d98ce14c98c56f7fd3553fcb6390e6 Mon Sep 17 00:00:00 2001
From: Zuellni <123005779+Zuellni@users.noreply.github.com>
Date: Tue, 19 Sep 2023 10:05:02 +0200
Subject: [PATCH] Add a note on model downloading

---
 README.md | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/README.md b/README.md
index 4a0228f..01e98ee 100644
--- a/README.md
+++ b/README.md
@@ -20,9 +20,12 @@ pip install https://github.com/jllllll/exllama/releases/download/0.0.17/exllama-
 Comes with the following nodes:
 
 ### Loader
-Used to load 4-bit GPTQ Llama/2 models. You can find a lot of them over at [Hugging Face](https://huggingface.co/TheBloke).  
+Used to load 4-bit GPTQ Llama/2 models. You can find a lot of them over at [Hugging Face](https://huggingface.co/TheBloke).
+
+You should either clone the model repository or download all the files in it manually, then point to the directory in `model_dir`. The `model.safetensors` file on its own is not enough to work.
+
 ExLlama allocates [memory](https://github.com/turboderp/exllama/issues/259) according to `max_seq_len`. Lowering it is a good way to save on GPU RAM.  
-It's currently not possible to [offload](https://github.com/turboderp/exllama/issues/177) the model to CPU RAM.
+It's currently not possible to [offload](https://github.com/turboderp/exllama/issues/177) the models to CPU RAM.
 
 ### Generator
 Generates a `string` based on the given `prompt` for use with other nodes.