File size: 4,460 Bytes
22c1c50 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 | #!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Benchmark Evaluation Script for Model Text Similarity
=========================================================
Compares generated results with reference texts using text embeddings.
"""
import numpy as np
import requests
import subprocess
import os
from pathlib import Path
from typing import List
MODEL_NAME = "grok-4.1-fast"
MODEL_RESULTS_PATH = "/path/to/your/model_generate_results_dir/%s/" % MODEL_NAME
TEST_DATA_PATH = "/path/to/your/DATA_PATH/test/"
class TextSimilarityCalculator:
def __init__(self, model_name="bge-m3", ollama_host="http://localhost:11434"):
self.model_name = model_name
self.ollama_host = ollama_host
def get_embedding(self, text: str) -> List[float]:
try:
response = requests.get(f"{self.ollama_host}/api/tags")
if response.status_code != 200:
return None
payload = {"model": self.model_name, "prompt": text, "stream": False}
response = requests.post(f"{self.ollama_host}/api/embeddings", json=payload, timeout=30)
if response.status_code == 200:
return response.json().get("embedding", [])
return None
except:
return None
def cosine_similarity(self, vec1: List[float], vec2: List[float]) -> float:
if not vec1 or not vec2:
return 0.0
vec1, vec2 = np.array(vec1), np.array(vec2)
norm1, norm2 = np.linalg.norm(vec1), np.linalg.norm(vec2)
if norm1 == 0 or norm2 == 0:
return 0.0
return np.dot(vec1, vec2) / (norm1 * norm2)
def calculate_similarity(self, text1: str, text2: str) -> float:
embedding1, embedding2 = self.get_embedding(text1), self.get_embedding(text2)
if embedding1 is None or embedding2 is None:
return 0.0
return float(self.cosine_similarity(embedding1, embedding2))
def check_ollama_installation(self):
try:
result = subprocess.run(["ollama", "--version"], capture_output=True, text=True)
if result.returncode == 0:
result = subprocess.run(["ollama", "list"], capture_output=True, text=True)
return self.model_name in result.stdout
return False
except:
return False
def find_matching_txt_files(ref_dir, test_dir):
matches = []
ref_txt_files = list(Path(ref_dir).glob("*.txt"))
for txt_path in Path(test_dir).rglob("*.txt"):
txt_name = txt_path.name
matching_ref = [ref for ref in ref_txt_files if ref.name == txt_name]
if matching_ref:
for ref_file in matching_ref:
matches.append((ref_file, txt_path))
return matches
def read_file_content(file_path):
try:
with open(file_path, 'r', encoding='utf-8') as f:
return f.read().strip()
except:
return ""
def main():
matches = find_matching_txt_files(MODEL_RESULTS_PATH, TEST_DATA_PATH)
if not matches:
print("No matching txt files found")
return
print(f"Found {len(matches)} matching txt file pairs")
print("-" * 50)
calculator = TextSimilarityCalculator()
if not calculator.check_ollama_installation():
print("Ollama environment check failed")
return
similarities = []
for i, (ref_path, test_path) in enumerate(matches, 1):
ref_content = read_file_content(ref_path)
test_content = read_file_content(test_path)
if not ref_content or not test_content:
print(f"File {ref_path.name}: Skipped (empty content)")
continue
similarity = calculator.calculate_similarity(ref_content, test_content)
similarities.append(similarity)
print(f"Pair {i}: {ref_path.name}")
print(f" Reference file: {ref_path}")
print(f" Target file: {test_path}")
print(f" Similarity: {similarity:.4f}")
print("-" * 30)
if similarities:
avg_similarity = np.mean(similarities)
print("=" * 50)
print(f"Total file pairs: {len(similarities)}")
print(f"Average similarity: {avg_similarity:.4f}")
else:
print("No valid file pairs for similarity calculation")
if __name__ == "__main__":
main()
|