From c4a0a370afc61f8714ac57444d81956495ad8f22 Mon Sep 17 00:00:00 2001
From: randaller <randaller@users.noreply.github.com>
Date: Sat, 4 Mar 2023 13:53:03 +0300
Subject: [PATCH] Update README.md

---
 README.md | 15 ---------------
 1 file changed, 15 deletions(-)

diff --git a/README.md b/README.md
index cfe6106..7ec544a 100755
--- a/README.md
+++ b/README.md
@@ -27,21 +27,6 @@ Run it:
 python example-cpu.py
 ```
 
-### Inference
-The provided `example.py` can be run on a single or multi-gpu node with `torchrun` and will output completions for two pre-defined prompts. Using `TARGET_FOLDER` as defined in `download.sh`:
-```
-torchrun --nproc_per_node MP example.py --ckpt_dir $TARGET_FOLDER/model_size --tokenizer_path $TARGET_FOLDER/tokenizer.model
-```
-
-Different models require different MP values:
-
-|  Model | MP |
-|--------|----|
-| 7B     | 1  |
-| 13B    | 2  |
-| 33B    | 4  |
-| 65B    | 8  |
-
 ### FAQ
 - [1. The download.sh script doesn't work on default bash in MacOS X](FAQ.md#1)
 - [2. Generations are bad!](FAQ.md#2)