From ee9e3d2a04e1749d60b89898bf65d8eea7c87933 Mon Sep 17 00:00:00 2001
From: leduc <leduc@MBP14.local>
Date: Sat, 15 Mar 2025 12:05:30 +0100
Subject: [PATCH] add improvments in readme

---
 .DS_Store | Bin 6148 -> 6148 bytes
 README.md |  11 +++++++++++
 2 files changed, 11 insertions(+)

diff --git a/.DS_Store b/.DS_Store
index 6efa36707d8027224ba9b5385f10bd46d1620d64..870c31a89a45d8542d4adaf6363aaf8e725a78eb 100644
GIT binary patch
delta 111
zcmZoMXffEJ#uB%49s>gd3xgg*IzuKyNp8N2OHxjL5>Sl8N%Up*sYAycQRP$M@-ytj
gGSKA<G7N*0^K%P;dKegZCooLTV^!O{iX}=I0E+G%R{#J2

delta 111
zcmZoMXffEJ#uB&EfPsO5g+Y%YogtH<Bsbs1B`GIA2`I*~IWq9U$wS8+QRP$M^7@Lo
gm(b-4G7N*0^K%P;dKegZ0~jXfv8rue#S$e90EvGcWB>pF

diff --git a/README.md b/README.md
index 6fb9679..0eb7767 100644
--- a/README.md
+++ b/README.md
@@ -142,6 +142,17 @@ The visualization includes:
 - Execution duration ranges
 - Color-coded model names (green for high performers)
 
+## Model Performance Overview
+
+The best models achieve 100% accuracy, maximize tokens per second (tk/sec), and minimize answer duration. For example, the deepseek-R1 and derived models often provides overly verbose responses, which can affect its overall performance.
+
+## Improvements
+
+### GPU Information display
+- **Model Name**: DeepSeek
+- **GPU Used**: Apple M1 Pro
+- **Performance Metrics**: The model achieves high accuracy with optimal tk/sec, but its verbosity can lead to longer response times.
+
 ## Server Configuration
 Default servers are configured in the code: