File size: 11,245 Bytes
4887b09 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 |
import argparse
import sys
import os
import torch
import torch.nn as nn
from torch.utils.data import DataLoader
import numpy as np
import pandas as pd
from tqdm import tqdm
from pathlib import Path
from transformers import AutoTokenizer
# Add src to path if running from root
sys.path.append(os.path.join(os.path.dirname(__file__)))
from dataset import ProteinTaxonomyDataset
from model import TaxonomyAwareESM
from asymmetric_loss import load_ia_weights
def evaluate_gpu(model, dataloader, ic_weights, device, thresholds=None, pred_output_path=None, metrics_output_path=None):
"""
Calculates Weighted F-max and S-min using GPU streaming to avoid OOM.
(Copied from train.py)
"""
model.eval()
if thresholds is None:
thresholds = torch.linspace(0, 1, 101, device=device)
# Initialize accumulators for each threshold
sum_prec = torch.zeros(len(thresholds), device=device)
sum_rec = torch.zeros(len(thresholds), device=device)
sum_ru = torch.zeros(len(thresholds), device=device) # Remaining Uncertainty (Weighted FN)
sum_mi = torch.zeros(len(thresholds), device=device) # Misinformation (Weighted FP)
total_samples = 0
# Prepare Prediction Output
f_pred = None
if pred_output_path:
os.makedirs(os.path.dirname(pred_output_path), exist_ok=True)
f_pred = open(pred_output_path, 'w')
idx_to_go = {v: k for k, v in dataloader.dataset.go_to_idx.items()}
with torch.no_grad():
for batch in tqdm(dataloader, desc="GPU Eval"):
input_ids = batch['input_ids'].to(device)
attention_mask = batch['attention_mask'].to(device)
tax_vector = batch['tax_vector'].to(device)
labels = batch['labels'].to(device) # (B, NumClasses)
entry_ids = batch['entry_id']
# --- ID HANDLING ---
if isinstance(entry_ids, str):
entry_ids = [entry_ids]
if not isinstance(entry_ids, (list, tuple)):
if isinstance(entry_ids, torch.Tensor):
entry_ids = entry_ids.tolist()
else:
entry_ids = list(entry_ids)
# 1. Forward
logits = model(input_ids, attention_mask, tax_vector)
probs = torch.sigmoid(logits) # (B, NumClasses)
# Save Predictions
if f_pred:
probs_cpu = probs.cpu().numpy()
for i, entry_id in enumerate(entry_ids):
indices = np.where(probs_cpu[i] > 0.01)[0]
for idx in indices:
term = idx_to_go[idx]
score = probs_cpu[i][idx]
f_pred.write(f"{entry_id}\t{term}\t{score:.4f}\n")
# 2. Ground Truth IC
# labels * weights
true_ic = (labels * ic_weights).sum(dim=1) # (B,)
true_ic = torch.maximum(true_ic, torch.tensor(1e-9, device=device))
# 3. Thresholding & Metrics Broadcasting
# (B, 1, C) >= (1, T, 1) -> (B, T, C)
probs_unsqueezed = probs.unsqueeze(1)
thresholds_unsqueezed = thresholds.view(1, -1, 1)
pred_binary = (probs_unsqueezed >= thresholds_unsqueezed).float()
labels_unsqueezed = labels.unsqueeze(1) # (B, 1, C)
ic_weights_unsqueezed = ic_weights.view(1, 1, -1) # (1, 1, C)
# intersection_ic (TP) shape: (B, T)
intersection_ic = (pred_binary * labels_unsqueezed * ic_weights_unsqueezed).sum(dim=2)
# pred_ic (TP + FP) shape: (B, T)
pred_ic = (pred_binary * ic_weights_unsqueezed).sum(dim=2)
# Precision: TP / Pred
precision = intersection_ic / (pred_ic + 1e-9)
# Recall: TP / True
recall = intersection_ic / (true_ic.view(-1, 1) + 1e-9)
# RU (False Negative): (True - TP) -> (B, T)
ru = true_ic.view(-1, 1) - intersection_ic
ru = torch.clamp(ru, min=0.0)
# MI (False Positive): (Pred - TP) -> (B, T)
mi = pred_ic - intersection_ic
mi = torch.clamp(mi, min=0.0)
# Accumulate Sums
sum_prec += precision.sum(dim=0)
sum_rec += recall.sum(dim=0)
sum_ru += ru.sum(dim=0)
sum_mi += mi.sum(dim=0)
total_samples += input_ids.size(0)
del logits, probs, pred_binary, intersection_ic, pred_ic, ru, mi
if f_pred:
f_pred.close()
print(f"Saved predictions to {pred_output_path}")
# Compute Averages
avg_prec = sum_prec / total_samples
avg_rec = sum_rec / total_samples
avg_ru = sum_ru / total_samples
avg_mi = sum_mi / total_samples
# F-max
f1_scores = 2 * avg_prec * avg_rec / (avg_prec + avg_rec + 1e-9)
best_fmax = f1_scores.max().item()
best_t_idx = f1_scores.argmax().item()
best_threshold_f = thresholds[best_t_idx].item()
# S-min
s_scores = torch.sqrt(avg_ru**2 + avg_mi**2)
min_s = s_scores.min().item()
best_s_idx = s_scores.argmin().item()
best_threshold_s = thresholds[best_s_idx].item()
metrics = {
'fmax_w': best_fmax,
'threshold_fmax': best_threshold_f,
'smin': min_s,
'threshold_smin': best_threshold_s,
}
# Save Metrics Detail
if metrics_output_path:
metrics_data = {
'threshold': thresholds.cpu().numpy(),
'precision': avg_prec.cpu().numpy(),
'recall': avg_rec.cpu().numpy(),
'f1': f1_scores.cpu().numpy(),
'ru': avg_ru.cpu().numpy(),
'mi': avg_mi.cpu().numpy(),
's': s_scores.cpu().numpy()
}
pd.DataFrame(metrics_data).to_csv(metrics_output_path, sep='\t', index=False)
print(f"Saved detailed metrics to {metrics_output_path}")
return metrics
def main():
parser = argparse.ArgumentParser()
parser.add_argument("--data_path", type=str, required=True, help="Path to mounted dataset")
parser.add_argument("--output_dir", type=str, default="outputs", help="Directory for checkpoints and predictions")
parser.add_argument("--checkpoint", type=str, default="latest_model.pth", help="Checkpoint filename to load (in output_dir)")
parser.add_argument("--epoch", type=int, default=3, help="Epoch to associate with predictions")
parser.add_argument("--esm_model_name", type=str, default="facebook/esm2_t33_650M_UR50D", help="ESM model name")
parser.add_argument("--force_novel", action="store_true", help="Force re-evaluation of novel dataset")
args = parser.parse_args()
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
print(f"Using device: {device}")
output_dir = Path(args.output_dir)
data_path = Path(args.data_path)
# Paths
val_novel_fasta = data_path / "validation_superset" / "validation_novel" / "validation_novel.fasta"
val_novel_term = data_path / "validation_superset" / "validation_novel" / "validation_novel_terms.tsv"
val_homolog_fasta = data_path / "validation_superset" / "validation_homolog" / "validation_homolog.fasta"
val_homolog_term = data_path / "validation_superset" / "validation_homolog" / "validation_homolog_terms.tsv"
species_vec = data_path / "taxon_embedding" / "species_vectors.tsv"
# GO Vocab
go_vocab_path = "src/go_terms.json"
if not os.path.exists(go_vocab_path):
go_vocab_path = "go_terms.json"
ia_path = data_path / "IA.tsv"
go_matrix_path = data_path / "go_info" / "go_ancestor_matrix.npz"
go_mapping_path = data_path / "go_info" / "go_term_mappings.pkl"
# Load Tokenizer
print("Loading tokenizer...")
tokenizer = AutoTokenizer.from_pretrained(args.esm_model_name)
# Initialize Dummy Dataset to get vocab sizes and mappings
# We need to reuse the same vocab as training
print("Initializing mapping dataset...")
dummy_dataset = ProteinTaxonomyDataset(
val_novel_fasta, val_novel_term, species_vec, go_vocab_path, max_len=1024, esm_tokenizer=tokenizer,
go_matrix_path=str(go_matrix_path), go_mapping_path=str(go_mapping_path)
)
print("Initializing Model...")
model = TaxonomyAwareESM(
num_classes=dummy_dataset.num_classes,
pretrained_model_name=args.esm_model_name,
use_lora=True,
lora_rank=512, # Assuming standard rank from training
vocab_sizes=dummy_dataset.vocab_sizes
).to(device)
# Load Checkpoint
ckpt_path = output_dir / args.checkpoint
if not ckpt_path.exists():
print(f"Error: Checkpoint not found at {ckpt_path}")
return
print(f"Loading checkpoint from {ckpt_path}...")
checkpoint = torch.load(ckpt_path, map_location=device)
model.load_state_dict(checkpoint['model_state_dict'])
print(f"Model loaded (Epoch {checkpoint['epoch']})")
# Load IC Weights
print("Loading IC weights...")
ic_weights = load_ia_weights(
str(ia_path) if ia_path.exists() else "IA.tsv",
dummy_dataset.go_to_idx,
dummy_dataset.num_classes
).to(device)
# --- EVALUATION ---
# 1. Novel Dataset
novel_preds_path = output_dir / f"gpu_preds_novel_epoch_{args.epoch}.tsv"
novel_metrics_path = output_dir / f"metrics_novel_epoch_{args.epoch}.tsv"
if novel_metrics_path.exists() and not args.force_novel:
print(f"Novel metrics already exist at {novel_metrics_path}. Skipping.")
with open(novel_metrics_path, 'r') as f:
print(f.read())
else:
print("Evaluating Novel Dataset...")
val_novel_loader = DataLoader(dummy_dataset, batch_size=32, shuffle=False, num_workers=4, pin_memory=True)
metrics_novel = evaluate_gpu(
model, val_novel_loader, ic_weights, device,
pred_output_path=novel_preds_path,
metrics_output_path=novel_metrics_path
)
print("Novel Metrics:", metrics_novel)
# 2. Homolog Dataset
homolog_preds_path = output_dir / f"gpu_preds_homolog_epoch_{args.epoch}.tsv"
homolog_metrics_path = output_dir / f"metrics_homolog_epoch_{args.epoch}.tsv"
print("Evaluating Homolog Dataset...")
val_homolog_dataset = ProteinTaxonomyDataset(
val_homolog_fasta, val_homolog_term, species_vec, go_vocab_path, max_len=1024, esm_tokenizer=tokenizer,
go_matrix_path=str(go_matrix_path), go_mapping_path=str(go_mapping_path)
)
val_homolog_loader = DataLoader(val_homolog_dataset, batch_size=32, shuffle=False, num_workers=4, pin_memory=True)
metrics_homolog = evaluate_gpu(
model, val_homolog_loader, ic_weights, device,
pred_output_path=homolog_preds_path,
metrics_output_path=homolog_metrics_path
)
print("Homolog Metrics:", metrics_homolog)
print("Evaluation All Done.")
if __name__ == "__main__":
main()
|