From 676a3f19d8255c03bdc27ad502e69e3a08fae9f6 Mon Sep 17 00:00:00 2001
From: Al-Tahir_Roman <romaaltahir@icloud.com>
Date: Tue, 21 Apr 2026 20:38:44 +0000
Subject: [PATCH] =?UTF-8?q?=D0=9E=D0=B1=D0=BD=D0=BE=D0=B2=D0=B8=D1=82?=
 =?UTF-8?q?=D1=8C=20README.md?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 README.md | 16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)

diff --git a/README.md b/README.md
index 6afc938..256dec3 100644
--- a/README.md
+++ b/README.md
@@ -40,16 +40,16 @@
   - Не требует эталонного изображения (reference-free).  
   - Является общепринятым стандартом в сообществе.
 - **Источник**: [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) (Radford et al.)
+- [library for testing](https://github.com/Taited/clip-score/)
 
 ### Детали реализации
 
-```python
-# Псевдокод расчета CLIP-score
-from transformers import CLIPProcessor, CLIPModel
+```bash
+# код расчета CLIP-score
+pip install transformers=4.25.1
+pip install torch
+pip install clip-score
 
-model = CLIPModel.from_pretrained("openai/clip-vit-base-patch32")
-processor = CLIPProcessor.from_pretrained("openai/clip-vit-base-patch32")
+python -m clip_score .\images .\texts
 
-inputs = processor(text=prompt, images=image, return_tensors="pt", padding=True)
-outputs = model(**inputs)
-score = outputs.logits_per_image.softmax(dim=1).item()  # или чистое косинусное сходство
\ No newline at end of file
+# CLIP Score:  0.3308749198913574
\ No newline at end of file