File size: 3,772 Bytes
32c143f | 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 | import torch
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from sklearn.decomposition import PCA
import os
import sys
import json
import random
# Import model
sys.path.append(os.path.join(os.path.dirname(__file__), '..', 'src'))
from model import TaxonomyAwareESM
def visualize_phylum_embeddings(model_path, vocab_path, output_dir):
print("Visualizing Random Phylum Embeddings (Rank 1)...")
os.makedirs(output_dir, exist_ok=True)
# 1. Load Model Weights
try:
checkpoint = torch.load(model_path, map_location='cpu')
state_dict = checkpoint['model_state_dict'] if 'model_state_dict' in checkpoint else checkpoint
except Exception as e:
print(f"Error loading model: {e}")
return
# 2. Load Vocab
try:
with open(vocab_path, 'r') as f:
vocab = json.load(f)
except FileNotFoundError:
print(f"Error: Vocab file not found at {vocab_path}")
return
# Filter out <UNK> if possible, or keep if it's all there is
candidates = [name for name in vocab.keys() if name != "<UNK>"]
if not candidates:
candidates = list(vocab.keys())
# Select 4 random
if len(candidates) >= 4:
selected_names = random.sample(candidates, 4)
else:
selected_names = candidates
print(f"Warning: Only found {len(candidates)} candidates in vocab. Using all of them.")
print(f"Selected Phyla: {selected_names}")
# 3. Visualize Rank 1 (Phylum)
rank_idx = 1
key = f"tax_embeddings.{rank_idx}.weight"
if key not in state_dict:
print(f"Missing weight for rank {rank_idx}")
return
weight = state_dict[key].numpy()
# PCA
pca = PCA(n_components=2)
transformed = pca.fit_transform(weight)
# Plot
plt.figure(figsize=(12, 12))
# Background: All points
plt.scatter(transformed[:, 0], transformed[:, 1], c='lightgrey', alpha=0.5, s=20, label='Others')
# Highlight Selected
colors = ['#FF0000', '#008000', '#0000FF', '#FFA500'] # Red, DarkGreen, Blue, Orange
for i, name in enumerate(selected_names):
idx = vocab.get(name)
if idx is not None and idx < len(transformed):
x, y = transformed[idx]
color = colors[i % len(colors)]
plt.scatter(x, y, c=color, s=200, edgecolor='black', zorder=10, marker='*')
plt.annotate(name,
xy=(x, y),
xytext=(10, 10),
textcoords='offset points',
fontsize=14,
fontweight='bold',
color=color,
bbox=dict(facecolor='white', alpha=0.7, edgecolor='none'))
else:
print(f"Index for {name} ({idx}) out of bounds or not found.")
plt.title("Phylum Embedding Space (Random 4)", fontsize=16)
plt.xlabel("PC1")
plt.ylabel("PC2")
plt.grid(True, alpha=0.3)
out_file = os.path.join(output_dir, "rank_1_phylum_random_4.png")
plt.savefig(out_file, dpi=300, bbox_inches='tight')
plt.close()
print(f"Saved plot to {out_file}")
if __name__ == "__main__":
import argparse
parser = argparse.ArgumentParser()
parser.add_argument("--model_path", type=str, required=True)
parser.add_argument("--vocab_path", type=str, default="data/vocab/phylum_vocab.json")
parser.add_argument("--output_dir", type=str, default="outputs")
args = parser.parse_args()
visualize_phylum_embeddings(args.model_path, args.vocab_path, args.output_dir)
|