evaluate_auc: adapt to new utils.load_model signature (infer checkpoint from config)

Added a `load_model` function to `utils.py` to allow loading of trained models from configuration and state dictionary files.

The `train_iter.py` script was also modified, likely to incorporate or test this new functionality.

.gitignore

@@ -5,7 +5,7 @@
 __pycache__/
 
 # Model checkpoints
-best_model_checkpoint.pt
+*.pt
 
 # Large data files
 ukb_delphi.txt

config_n_embd_120_n_layer_12_n_head_12.json

@@ -0,0 +1,18 @@
+{
+    "n_layer": 12,
+    "n_embd": 120,
+    "n_head": 12,
+    "max_epoch": 200,
+    "batch_size": 128,
+    "lr_initial": 0.0006,
+    "lr_final": 6e-05,
+    "weight_decay": 0.2,
+    "warmup_epochs": 10,
+    "early_stopping_patience": 10,
+    "pdrop": 0.0,
+    "token_pdrop": 0.0,
+    "betas": [
+        0.9,
+        0.99
+    ]
+}

config_n_embd_256_n_layer_16_n_head_16.json

@@ -0,0 +1,18 @@
+{
+    "n_layer": 16,
+    "n_embd": 256,
+    "n_head": 16,
+    "max_epoch": 200,
+    "batch_size": 128,
+    "lr_initial": 0.0006,
+    "lr_final": 6e-05,
+    "weight_decay": 0.2,
+    "warmup_epochs": 10,
+    "early_stopping_patience": 10,
+    "pdrop": 0.0,
+    "token_pdrop": 0.0,
+    "betas": [
+        0.9,
+        0.99
+    ]
+}

evaluate_auc.py

@@ -0,0 +1,496 @@
+import torch
+from tqdm import tqdm
+import pandas as pd
+import numpy as np
+import argparse
+from utils import load_model, get_batch, PatientEventDataset
+from pathlib import Path
+from joblib import Parallel, delayed
+
+
+def auc(x1, x2):
+    n1 = len(x1)
+    n2 = len(x2)
+    R1 = np.concatenate([x1, x2]).argsort().argsort()[:n1].sum() + n1
+    U1 = n1 * n2 + 0.5 * n1 * (n1 + 1) - R1
+    if n1 == 0 or n2 == 0:
+        return np.nan
+    return U1 / n1 / n2
+
+
+def get_common_diseases(delphi_labels, filter_min_total=100):
+    chapters_of_interest = [
+        "I. Infectious Diseases",
+        "II. Neoplasms",
+        "III. Blood & Immune Disorders",
+        "IV. Metabolic Diseases",
+        "V. Mental Disorders",
+        "VI. Nervous System Diseases",
+        "VII. Eye Diseases",
+        "VIII. Ear Diseases",
+        "IX. Circulatory Diseases",
+        "X. Respiratory Diseases",
+        "XI. Digestive Diseases",
+        "XII. Skin Diseases",
+        "XIII. Musculoskeletal Diseases",
+        "XIV. Genitourinary Diseases",
+        "XV. Pregnancy & Childbirth",
+        "XVI. Perinatal Conditions",
+        "XVII. Congenital Abnormalities",
+        "Death",
+    ]
+    labels_df = delphi_labels[
+        delphi_labels["ICD-10 Chapter (short)"].isin(chapters_of_interest) * (delphi_labels["count"] > filter_min_total)
+    ]
+    return labels_df["index"].tolist()
+
+
+def optimized_bootstrapped_auc_gpu(case, control, n_bootstrap=1000):
+    """
+    Computes bootstrapped AUC estimates using PyTorch on CUDA.
+
+    Parameters:
+        case: 1D tensor of scores for positive cases
+        control: 1D tensor of scores for controls
+        n_bootstrap: Number of bootstrap replicates
+
+    Returns:
+        Tensor of shape (n_bootstrap,) containing AUC for each bootstrap replicate
+    """
+    if not torch.cuda.is_available():
+        raise RuntimeError("CUDA is not available. This function requires a GPU.")
+
+    # Convert inputs to CUDA tensors
+    if not torch.is_tensor(case):
+        case = torch.tensor(case, device="cuda", dtype=torch.float32)
+    else:
+        case = case.to("cuda", dtype=torch.float32)
+
+    if not torch.is_tensor(control):
+        control = torch.tensor(control, device="cuda", dtype=torch.float32)
+    else:
+        control = control.to("cuda", dtype=torch.float32)
+
+    n_case = case.size(0)
+    n_control = control.size(0)
+    total = n_case + n_control
+
+    # Generate bootstrap samples
+    boot_idx_case = torch.randint(0, n_case, (n_bootstrap, n_case), device="cuda")
+    boot_idx_control = torch.randint(0, n_control, (n_bootstrap, n_control), device="cuda")
+
+    boot_case = case[boot_idx_case]
+    boot_control = control[boot_idx_control]
+
+    combined = torch.cat([boot_case, boot_control], dim=1)
+
+    # Mask to identify case entries
+    mask = torch.zeros((n_bootstrap, total), dtype=torch.bool, device="cuda")
+    mask[:, :n_case] = True
+
+    # Compute ranks and AUC
+    ranks = combined.argsort(dim=1).argsort(dim=1)
+    case_ranks_sum = torch.sum(ranks.float() * mask.float(), dim=1)
+    min_case_rank_sum = n_case * (n_case - 1) / 2.0
+    U = case_ranks_sum - min_case_rank_sum
+    aucs = U / (n_case * n_control)
+    return aucs.cpu().tolist()
+
+
+# AUC comparison adapted from
+# https://github.com/Netflix/vmaf/
+def compute_midrank(x):
+    """Computes midranks.
+    Args:
+       x - a 1D numpy array
+    Returns:
+       array of midranks
+    """
+    J = np.argsort(x)
+    Z = x[J]
+    N = len(x)
+    T = np.zeros(N, dtype=np.float32)
+    i = 0
+    while i < N:
+        j = i
+        while j < N and Z[j] == Z[i]:
+            j += 1
+        T[i:j] = 0.5 * (i + j - 1)
+        i = j
+    T2 = np.empty(N, dtype=np.float32)
+    # Note(kazeevn) +1 is due to Python using 0-based indexing
+    # instead of 1-based in the AUC formula in the paper
+    T2[J] = T + 1
+    return T2
+
+
+def fastDeLong(predictions_sorted_transposed, label_1_count):
+    """
+    The fast version of DeLong's method for computing the covariance of
+    unadjusted AUC.
+    Args:
+       predictions_sorted_transposed: a 2D numpy.array[n_classifiers, n_examples]
+          sorted such as the examples with label "1" are first
+    Returns:
+       (AUC value, DeLong covariance)
+    Reference:
+     @article{sun2014fast,
+       title={Fast Implementation of DeLong's Algorithm for
+              Comparing the Areas Under Correlated Receiver Operating Characteristic Curves},
+       author={Xu Sun and Weichao Xu},
+       journal={IEEE Signal Processing Letters},
+       volume={21},
+       number={11},
+       pages={1389--1393},
+       year={2014},
+       publisher={IEEE}
+     }
+    """
+    # Short variables are named as they are in the paper
+    m = label_1_count
+    n = predictions_sorted_transposed.shape[1] - m
+    positive_examples = predictions_sorted_transposed[:, :m]
+    negative_examples = predictions_sorted_transposed[:, m:]
+    k = predictions_sorted_transposed.shape[0]
+
+    tx = np.empty([k, m], dtype=np.float32)
+    ty = np.empty([k, n], dtype=np.float32)
+    tz = np.empty([k, m + n], dtype=np.float32)
+    for r in range(k):
+        tx[r, :] = compute_midrank(positive_examples[r, :])
+        ty[r, :] = compute_midrank(negative_examples[r, :])
+        tz[r, :] = compute_midrank(predictions_sorted_transposed[r, :])
+    aucs = tz[:, :m].sum(axis=1) / m / n - float(m + 1.0) / 2.0 / n
+    v01 = (tz[:, :m] - tx[:, :]) / n
+    v10 = 1.0 - (tz[:, m:] - ty[:, :]) / m
+    sx = np.cov(v01)
+    sy = np.cov(v10)
+    delongcov = sx / m + sy / n
+    return aucs, delongcov
+
+
+def compute_ground_truth_statistics(ground_truth):
+    assert np.array_equal(np.unique(ground_truth), [0, 1])
+    order = (-ground_truth).argsort()
+    label_1_count = int(ground_truth.sum())
+    return order, label_1_count
+
+
+def get_auc_delong_var(healthy_scores, diseased_scores):
+    """
+    Computes ROC AUC value and variance using DeLong's method
+
+    Args:
+        healthy_scores: Values for class 0 (healthy/controls)
+        diseased_scores: Values for class 1 (diseased/cases)
+    Returns:
+        AUC value and variance
+    """
+    # Create ground truth labels (1 for diseased, 0 for healthy)
+    ground_truth = np.array([1] * len(diseased_scores) + [0] * len(healthy_scores))
+    predictions = np.concatenate([diseased_scores, healthy_scores])
+
+    # Compute statistics needed for DeLong method
+    order, label_1_count = compute_ground_truth_statistics(ground_truth)
+    predictions_sorted_transposed = predictions[np.newaxis, order]
+
+    # Calculate AUC and covariance
+    aucs, delongcov = fastDeLong(predictions_sorted_transposed, label_1_count)
+    assert len(aucs) == 1, "There is a bug in the code, please forward this to the developers"
+
+    return aucs[0], delongcov
+
+
+def get_calibration_auc(j, k, d, p, offset=365.25, age_groups=range(45, 80, 5), precomputed_idx=None, n_bootstrap=1, use_delong=False):
+    age_step = age_groups[1] - age_groups[0]
+
+    # Indexes of cases with disease k
+    wk = np.where(d[2] == k)
+
+    if len(wk[0]) < 2:
+        return None
+
+    # For controls, we need to exclude cases with disease k
+    wc = np.where((d[2] != k) & (~(d[2] == k).any(-1))[..., None])
+
+    wall = (np.concatenate([wk[0], wc[0]]), np.concatenate([wk[1], wc[1]]))  # All cases and controls
+
+    # We need to take into account the offset t and use the tokens for prediction that are at least t before the event
+    if precomputed_idx is None:
+        pred_idx = (d[1][wall[0]] <= d[3][wall].reshape(-1, 1) - offset).sum(1) - 1
+    else:
+        pred_idx = precomputed_idx[wall]  # It's actually much faster to precompute this
+
+    valid_indices = pred_idx != -1
+    pred_idx = pred_idx[valid_indices]
+    wall = (wall[0][valid_indices], wall[1][valid_indices])
+    
+    z = d[1][(wall[0], pred_idx)]  # Times of the tokens for prediction
+    zk = d[3][wall]  # Target times
+
+    x = p[..., j][(wall[0], pred_idx)]
+
+    p_idx = wall[0]
+
+    out = []
+
+    for i, aa in enumerate(age_groups):
+        a = (z / 365.25 >= aa) & (z / 365.25 < aa + age_step)
+        
+        if not np.any(a):
+            continue
+
+        selected_groups = p_idx[a]
+        _, unique_indices = np.unique(selected_groups, return_index=True)
+        
+        a_filtered = a[a]
+        a_filtered[:] = False
+        a_filtered[unique_indices] = True
+        a[a] = a_filtered
+
+        is_case = np.zeros_like(x, dtype=bool)
+        is_case[:len(wk[0])] = True
+        
+        control = x[~is_case & a]
+        case = x[is_case & a]
+
+        if len(control) == 0 or len(case) == 0:
+            continue
+
+        if use_delong:
+            auc_value_delong, auc_variance_delong = get_auc_delong_var(control, case)
+            auc_delong_dict = {"auc_delong": auc_value_delong, "auc_variance_delong": auc_variance_delong}
+        else:
+            auc_delong_dict = {}
+
+        if n_bootstrap > 1:
+            aucs_bootstrapped = optimized_bootstrapped_auc_gpu(case, control, n_bootstrap)
+
+        for bootstrap_idx in range(n_bootstrap):
+            y = auc_value_delong if n_bootstrap == 1 else aucs_bootstrapped[bootstrap_idx]
+            out_item = {
+                "token": k,
+                "auc": y,
+                "age": aa,
+                "n_healthy": len(control),
+                "n_diseased": len(case),
+            }
+            out.append(out_item | auc_delong_dict)
+            if n_bootstrap > 1:
+                out_item["bootstrap_idx"] = bootstrap_idx
+    return out
+
+
+def process_chunk(disease_chunk_idx, diseases_chunk, d100k, p100k, pred_idx_precompute, age_groups, offset, n_bootstrap):
+    all_aucs = []
+    for sex, sex_idx in [("female", 2), ("male", 3)]:
+        sex_mask = ((d100k[0] == sex_idx).sum(1) > 0).cpu().detach().numpy()
+        p_sex = p100k[sex_mask]
+        d100k_sex = [d_.cpu().detach().numpy()[sex_mask] for d_ in d100k]
+        precomputed_idx_subset = pred_idx_precompute[sex_mask].cpu().detach().numpy()
+        for j, k in tqdm(
+            list(enumerate(diseases_chunk)), desc=f"Processing diseases in chunk {disease_chunk_idx}, {sex}"
+        ):
+            out = get_calibration_auc(
+                j,
+                k,
+                d100k_sex,
+                p_sex,
+                age_groups=age_groups,
+                offset=offset,
+                precomputed_idx=precomputed_idx_subset,
+                n_bootstrap=n_bootstrap,
+                use_delong=True,
+            )
+            if out is None:
+                continue
+            for out_item in out:
+                out_item["sex"] = sex
+                all_aucs.append(out_item)
+    return all_aucs
+
+
+# New internal function that performs the AUC evaluation pipeline.
+def evaluate_auc_pipeline(
+    model,
+    d100k,
+    output_path,
+    delphi_labels,
+    diseases_of_interest=None,
+    filter_min_total=100,
+    disease_chunk_size=200,
+    age_groups=np.arange(40, 80, 5),
+    offset=0.1,
+    batch_size=256,
+    device="cpu",
+    seed=1337,
+    n_bootstrap=1,
+    meta_info={},
+    n_jobs=-1,
+):
+    """
+    Runs the AUC evaluation pipeline.
+
+    Args:
+        model (torch.nn.Module): The loaded model set to eval().
+        d100k (tuple): Data batch from get_batch.
+        delphi_labels (pd.DataFrame): DataFrame with label info (token names, etc. "delphi_labels_chapters_colours_icd.csv").
+        output_path (str | None): Directory where CSV files will be written. If None, files will not be saved.
+        diseases_of_interest (np.ndarray or list, optional): If provided, these disease indices are used.
+        filter_min_total (int): Minimum total token count to include a token.
+        disease_chunk_size (int): Maximum chunk size for processing diseases.
+        age_groups (np.ndarray): Age groups to use in calibration.
+        offset (float): Offset used in get_calibration_auc.
+        batch_size (int): Batch size for model forwarding.
+        device (str): Device identifier.
+        seed (int): Random seed for reproducibility.
+        n_bootstrap (int): Number of bootstrap samples. (1 for no bootstrap)
+        n_jobs (int): Number of parallel jobs to run.
+    Returns:
+        tuple: (df_auc_unpooled, df_auc, df_both) DataFrames.
+    """
+
+    assert n_bootstrap > 0, "n_bootstrap must be greater than 0"
+
+    # Set random seeds
+    torch.manual_seed(seed)
+    torch.cuda.manual_seed(seed)
+
+    # Get common diseases
+    if diseases_of_interest is None:
+        diseases_of_interest = get_common_diseases(delphi_labels, filter_min_total)
+
+    # Split diseases into chunks for processing
+    num_chunks = (len(diseases_of_interest) + disease_chunk_size - 1) // disease_chunk_size
+    diseases_chunks = np.array_split(diseases_of_interest, num_chunks)
+
+    # Precompute prediction indices for calibration
+    pred_idx_precompute = (d100k[1][:, :, np.newaxis] < d100k[3][:, np.newaxis, :] - offset).sum(1) - 1
+
+    p100k = []
+    model.to(device)
+    with torch.no_grad():
+        for dd in tqdm(
+            zip(*[torch.split(x, batch_size) for x in d100k]),
+            desc=f"Model inference",
+            total=d100k[0].shape[0] // batch_size + 1,
+        ):
+            dd = [x.to(device) for x in dd]
+            outputs = model(dd[0], dd[1]).cpu().detach().numpy()
+            p100k.append(outputs.astype("float16"))
+    p100k = np.vstack(p100k)
+    
+    results = Parallel(n_jobs=n_jobs)(
+        delayed(process_chunk)(
+            disease_chunk_idx, diseases_chunk, d100k, p100k[:, :, diseases_chunk], pred_idx_precompute, age_groups, offset, n_bootstrap
+        )
+        for disease_chunk_idx, diseases_chunk in enumerate(diseases_chunks)
+    )
+
+    all_aucs = [item for sublist in results for item in sublist]
+
+    df_auc_unpooled = pd.DataFrame(all_aucs)
+
+    for key, value in meta_info.items():
+        df_auc_unpooled[key] = value
+
+    delphi_labels_subset = delphi_labels[['index', 'ICD-10 Chapter (short)', 'name', 'color', 'count']]
+    df_auc_unpooled_merged = df_auc_unpooled.merge(delphi_labels_subset, left_on="token", right_on="index", how="inner")
+
+    def aggregate_age_brackets_delong(group):
+        # For normal distributions, when averaging n of them:
+        # The variance of the sum is the sum of variances
+        # The variance of the average is the sum of variances divided by n^2
+        n = len(group)
+        mean = group['auc_delong'].mean()
+        # Since we're taking the average, divide combined variance by n^2
+        var = group['auc_variance_delong'].sum() / (n**2)
+        return pd.Series({
+            'auc': mean,
+            'auc_variance_delong': var,
+            'n_samples': n, 
+            'n_diseased': group['n_diseased'].sum(),
+            'n_healthy': group['n_healthy'].sum(),
+        })
+
+    print('Using DeLong method to calculate AUC confidence intervals..')
+    
+    df_auc = df_auc_unpooled.groupby(["token"]).apply(aggregate_age_brackets_delong).reset_index()
+    df_auc_merged = df_auc.merge(delphi_labels, left_on="token", right_on="index", how="inner")
+    
+    if output_path is not None:
+        Path(output_path).mkdir(exist_ok=True, parents=True)
+        df_auc_merged.to_csv(f"{output_path}/df_both.csv", index=False)
+        df_auc_unpooled_merged.to_csv(f"{output_path}/df_auc_unpooled.csv", index=False)
+
+    return df_auc_unpooled_merged, df_auc_merged
+
+
+def main():
+    parser = argparse.ArgumentParser(description="Evaluate AUC")
+    parser.add_argument("--model_name", type=str, default="n_embd_256_n_layer_16_n_head_16", help="Model checkpoint name")
+    parser.add_argument("--dataset_subset_size", type=int, default=-1, help="Dataset subset size for evaluation")
+    parser.add_argument("--n_bootstrap", type=int, default=1, help="Number of bootstrap samples")
+    parser.add_argument("--offset", type=float, default=365.25, help="Offset in days for prediction")
+    # Optional filtering/chunking parameters:
+    parser.add_argument("--filter_min_total", type=int, default=100, help="Minimum total count to filter tokens")
+    parser.add_argument("--disease_chunk_size", type=int, default=200, help="Chunk size for processing diseases")
+    parser.add_argument("--n_jobs", type=int, default=-1, help="Number of parallel jobs to run")
+    args = parser.parse_args()
+
+    model_name = args.model_name
+    output_path = f'auc_evaluation_{model_name}'
+    dataset_subset_size = args.dataset_subset_size
+    offset = args.offset
+
+    # Create output folder if it doesn't exist.
+    Path(output_path).mkdir(exist_ok=True, parents=True)
+
+    device = "cuda" if torch.cuda.is_available() else "cpu"
+    seed = 1337
+
+    # Load model checkpoint and initialize model.
+    model = load_model(
+        config_path=f'config_{model_name}.json',
+        device=device,
+    )
+    model.eval()
+    model = model.to(device)
+
+    # Load training and validation data.
+
+    
+    val_data_path = 'ukb_real_val.bin'
+    
+    val_data_arr = np.memmap(val_data_path, dtype=np.uint32, mode='r').reshape(-1, 3)
+    block_length = 128    
+    val_dataset = PatientEventDataset(val_data_arr, block_length)
+
+    if dataset_subset_size == -1:
+        dataset_subset_size = len(val_dataset)
+
+    # Get a subset batch for evaluation.
+    d100k = get_batch(val_dataset, slice(dataset_subset_size))
+
+    # Load labels (external) to be passed in.
+    delphi_labels = pd.read_csv("delphi_labels_chapters_colours_icd.csv")
+
+    # Call the internal evaluation function.
+    df_auc_unpooled, df_auc_merged = evaluate_auc_pipeline(
+        model,
+        d100k,
+        output_path,
+        delphi_labels,
+        diseases_of_interest=None,
+        filter_min_total=args.filter_min_total,
+        disease_chunk_size=args.disease_chunk_size,
+        device=device,
+        seed=seed,
+        offset=offset,
+        n_bootstrap=args.n_bootstrap,
+        n_jobs=args.n_jobs,
+    )
+
+
+if __name__ == "__main__":
+    main()

models.py

@@ -1,7 +1,7 @@
 import torch
 import torch.nn as nn
 from torch.nn import functional as F
-from typing import Tuple
+from typing import Tuple, Optional
 
 # =============================================================================
 # 1. Component Modules (Building Blocks)
@@ -85,6 +85,39 @@ class AgeSinusoidalEncoding(nn.Module):
         output[:, :, 1::2] = torch.sin(args)
         return output
 
+
+class LearnableAgeEncoding(nn.Module):
+    """Combines fixed sinusoidal age encodings with a learnable MLP projection."""
+
+    def __init__(self, base_dim: int, hidden_dim: Optional[int] = None, final_dim: Optional[int] = None, dropout: float = 0.0):
+        super().__init__()
+        self.base_dim = base_dim
+        self.final_dim = final_dim or base_dim
+
+        hidden_dim = hidden_dim or base_dim
+        if hidden_dim <= 0:
+            raise ValueError("hidden_dim must be a positive integer.")
+        if self.final_dim <= 0:
+            raise ValueError("final_dim must be a positive integer.")
+
+        self.sinusoidal = AgeSinusoidalEncoding(base_dim)
+
+        mlp_layers = [
+            nn.Linear(base_dim, hidden_dim),
+            nn.GELU(),
+        ]
+        if dropout > 0.0:
+            mlp_layers.append(nn.Dropout(dropout))
+        mlp_layers.append(nn.Linear(hidden_dim, self.final_dim))
+
+        self.mlp = nn.Sequential(*mlp_layers)
+
+    def forward(self, t: torch.Tensor) -> torch.Tensor:
+        sin_embed = self.sinusoidal(t)
+        flat_embed = sin_embed.reshape(-1, self.base_dim)
+        projected = self.mlp(flat_embed)
+        return projected.reshape(*sin_embed.shape[:-1], self.final_dim)
+
 class PiecewiseLinearEncoder(nn.Module):
     """
     Encodes continuous variables using piecewise linear encoding.
@@ -177,9 +210,10 @@ class TimeAwareGPT2(nn.Module):
     A time-aware GPT-2 model with custom temporal features.
     """
 
-    def __init__(self, vocab_size: int, n_embd: int, n_layer: int, n_head: int, pdrop: float, token_pdrop: float):
+    def __init__(self, vocab_size: int, n_embd: int, n_layer: int, n_head: int, pdrop: float, token_pdrop: float, ignore_tokens: list[int] = None):
         super().__init__()
         self.token_pdrop = token_pdrop
+        self.ignore_tokens = ignore_tokens if ignore_tokens is not None else []
 
         self.wte = nn.Embedding(vocab_size, n_embd)
         self.age_encoder = AgeSinusoidalEncoding(n_embd)
@@ -234,94 +268,71 @@ class TimeAwareGPT2(nn.Module):
         """
         return sum(p.numel() for p in self.parameters() if p.requires_grad) / 1e6
 
-class CovariateAwareGPT2(nn.Module):
+    @torch.no_grad()
+    def generate(self, x, t, max_new_tokens=100, max_age=85*365.25, no_repeat=True, termination_tokens=None, top_k=None):
         """
-    Extends TimeAwareGPT2 to incorporate static and time-varying covariates.
+        Take a conditioning sequence of indices x (LongTensor of shape (b,t)) and complete
+        the sequence max_new_tokens times, feeding the predictions back into the model each time.
+        Most likely you'll want to make sure to be in model.eval() mode of operation for this.
         """
+        self.eval()
 
-    def __init__(self, vocab_size: int, n_embd: int, n_layer: int, n_head: int,
+        if termination_tokens is None:
-                 pdrop: float, token_pdrop: float, num_bins: int):
+            termination_tokens = [1269]
-        """
-        Initializes the CovariateAwareGPT2 model.
         
-        Args:
+        termination_tokens = torch.tensor(termination_tokens, dtype=torch.int64, device=x.device)
-            vocab_size (int): Size of the event vocabulary.
+        mask_time = -10000
-            n_embd (int): Embedding dimensionality.
-            n_layer (int): Number of transformer layers.
-            n_head (int): Number of attention heads.
-            pdrop (float): Dropout probability for layers.
-            token_pdrop (float): Dropout probability for input token embeddings.
-            num_bins (int): Number of bins for the PiecewiseLinearEncoder.
-        """
-        super().__init__()
-        self.token_pdrop = token_pdrop
 
-        self.wte = nn.Embedding(vocab_size, n_embd)
+        for _ in range(max_new_tokens):
-        self.age_encoder = AgeSinusoidalEncoding(n_embd)
+            logits = self(x, t)
-        self.drop = nn.Dropout(pdrop)
+            logits = logits[:, -1, :]
-        self.blocks = nn.ModuleList([Block(n_embd, n_head, pdrop) for _ in range(n_layer)])
-        self.n_embd = n_embd
-        self.cov_encoder = PiecewiseLinearEncoder(num_bins=num_bins, embedding_dim=n_embd)
             
-        self.ln_f = nn.LayerNorm(2 * n_embd)
+            if self.ignore_tokens:
-        self.head = nn.Sequential(
+                logits[:, self.ignore_tokens] = -torch.inf
-            nn.Linear(2 * n_embd, n_embd),
+
-            nn.GELU(),
+            if no_repeat:
-            nn.Linear(n_embd, vocab_size)
+                fill = x.clone()
+                fill[fill == 1] = 0
+                logits = logits.scatter(1, fill, -torch.inf)
+            
+            t_next_dist = torch.clamp(-torch.exp(-logits) * torch.rand(logits.shape, device=x.device).log(), min=0, max=365*80)
+            t_next_val, idx_next = t_next_dist.min(1)
+            
+            idx_next = idx_next.unsqueeze(1)
+            age_next = t[:, -1].unsqueeze(1) + t_next_val.unsqueeze(1)
+            
+            x = torch.cat((x, idx_next), dim=1)
+            t = torch.cat((t, age_next), dim=1)
+            
+            if torch.logical_or(torch.isin(x, termination_tokens).any(-1), age_next.squeeze() > max_age).all():
+                break
+        
+        pad = (torch.cumsum(torch.cumsum(torch.isin(x, termination_tokens), 1).bool().int(), 1) > 1) + (t > max_age)
+
+        final_logits = self(x, t)
+        x[pad] = 0
+        t[pad] = mask_time
+
+        if no_repeat:
+            fill = x.clone()
+            fill[fill == 1] = 0
+            final_logits = torch.stack([final_logits[:,j].scatter(1, fill[:,:j+1], -torch.inf) for j in range(fill.shape[1])]).transpose(0,1)
+
+        return x, t, final_logits
+
+
+class TimeAwareGPT2Learnable(TimeAwareGPT2):
+    """Variant of TimeAwareGPT2 that uses LearnableAgeEncoding for temporal features."""
+
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        self.age_encoder = LearnableAgeEncoding(
+            base_dim=self.n_embd,
+            hidden_dim=2 * self.n_embd,
+            final_dim=self.n_embd,
         )
 
-    def forward(self, x: torch.Tensor, t: torch.Tensor, cov: torch.Tensor, cov_t: torch.Tensor) -> torch.Tensor:
-        """
-        Forward pass for the CovariateAwareGPT2 model.
 
-        Args:
-            x (torch.Tensor): Event sequence tensor of shape (B, L).
-            t (torch.Tensor): Time sequence tensor of shape (B, L).
-            cov (torch.Tensor): Covariate tensor of shape (B, N).
-            cov_t (torch.Tensor): Covariate time tensor of shape (B).
-
-        Returns:
-            torch.Tensor: Logits of shape (B, L, vocab_size).
-        """
-        B, L = x.size()
-
-        cov_encoded = self.cov_encoder(cov).sum(dim=1).unsqueeze(1)
-        cov_t_encoded = self.age_encoder(t - cov_t.unsqueeze(1))
-        cov_embed = cov_encoded + cov_t_encoded
-
-        token_embeddings = self.wte(x)
-        if self.training and self.token_pdrop > 0:
-            drop_mask = torch.rand(token_embeddings.shape[:2], device=token_embeddings.device) < self.token_pdrop
-            token_embeddings[drop_mask] = 0.0
-        
-        pos_embeddings = self.age_encoder(t.float())
-        seq_embed = self.drop(token_embeddings + pos_embeddings)
-
-        t_i = t.unsqueeze(-1)
-        t_j = t.unsqueeze(1)
-        time_mask = (t_j < t_i)
-        padding_mask = (x != 0).unsqueeze(1)
-        combined_mask = time_mask & padding_mask
-        is_row_all_zero = ~combined_mask.any(dim=-1)
-        is_not_padding = (x != 0)
-        force_self_attention = is_row_all_zero & is_not_padding
-        combined_mask.diagonal(dim1=-2, dim2=-1)[force_self_attention] = True
-
-        block_output = seq_embed
-        for block in self.blocks:
-            block_output = block(block_output, custom_mask=combined_mask)
-
-        integrated_embed = torch.cat([block_output, cov_embed], dim=-1)
-
-        final_output = self.ln_f(integrated_embed)
-        logits = self.head(final_output)
-        return logits
-
-    def get_num_params(self) -> float:
-        """
-        Returns the number of trainable parameters in the model in millions.
-        """
-        return sum(p.numel() for p in self.parameters() if p.requires_grad) / 1e6
 
 # =============================================================================
 # 3. Loss Function
@@ -367,8 +378,12 @@ class CombinedLoss(nn.Module):
         per_element_ce = F.cross_entropy(logits_for_ce, x, reduction='none')
         loss_ce = per_element_ce[mask].mean()
 
-        intensity = torch.sum(torch.exp(logits), dim=2)
+        # Survival loss based on exponential log-likelihood
-        per_element_survival = -(torch.log(intensity + 1e-8) - intensity * t)
+        t_min = 0.1
-        loss_survival = per_element_survival[mask].mean()
+        lse = torch.logsumexp(logits, dim=-1)
+        lse = -torch.log(torch.exp(-lse) + t_min)
+        ldt = -torch.log(t + t_min)
+        loss_dt = -(lse - torch.exp(lse - ldt))
+        loss_survival = loss_dt[mask].mean()
 
         return loss_ce, loss_survival

requirements.txt

@@ -2,3 +2,4 @@ torch
 numpy
 tqdm
 matplotlib
+joblib

train.py

@@ -6,8 +6,10 @@ import numpy as np
 import math
 import tqdm
 import matplotlib.pyplot as plt
+import json
+import argparse
 
-from models import TimeAwareGPT2, CombinedLoss
+from models import TimeAwareGPT2, TimeAwareGPT2Learnable, CombinedLoss
 from utils import PatientEventDataset
 
 # --- Configuration ---
@@ -23,14 +25,17 @@ class TrainConfig:
     n_head = 12
     pdrop = 0.1
     token_pdrop = 0.1
+    model_name = 'TimeAwareGPT2'
 
     # Training parameters
     max_epoch = 200
     batch_size = 128
     lr_initial = 6e-4
     lr_final = 6e-5
+    weight_decay = 2e-1
     warmup_epochs = 10
     early_stopping_patience = 10
+    betas = (0.9, 0.99)
     
     # Loss parameters
     # 0 = padding, 1 = "no event"
@@ -41,10 +46,50 @@ class TrainConfig:
 
 # --- Main Training Script ---
 def main():
-    config = TrainConfig()
+    parser = argparse.ArgumentParser(description='Train a Time-Aware GPT-2 model.')
+    parser.add_argument('--n_layer', type=int, default=12, help='Number of transformer layers.')
+    parser.add_argument('--n_embd', type=int, default=120, help='Embedding dimension.')
+    parser.add_argument('--n_head', type=int, default=12, help='Number of attention heads.')
+    parser.add_argument('--max_epoch', type=int, default=200, help='Maximum number of training epochs.')
+    parser.add_argument('--batch_size', type=int, default=128, help='Batch size for training.')
+    parser.add_argument('--lr_initial', type=float, default=6e-4, help='Initial learning rate.')
+    parser.add_argument('--lr_final', type=float, default=6e-5, help='Final learning rate.')
+    parser.add_argument('--weight_decay', type=float, default=2e-1, help='Weight decay for the optimizer.')
+    parser.add_argument('--warmup_epochs', type=int, default=10, help='Number of warmup epochs.')
+    parser.add_argument('--early_stopping_patience', type=int, default=10, help='Patience for early stopping.')
+    parser.add_argument('--pdrop', type=float, default=0.1, help='Dropout probability.')
+    parser.add_argument('--token_pdrop', type=float, default=0.1, help='Token dropout probability.')
+    parser.add_argument('--betas', type=float, nargs=2, default=[0.9, 0.99], help='AdamW betas.')
+    parser.add_argument('--model', type=str, choices=['TimeAwareGPT2', 'TimeAwareGPT2Learnable'], default='TimeAwareGPT2', help='Model architecture to train.')
 
-    model_filename = f"best_model_n_embd_{config.n_embd}_n_layer_{config.n_layer}_n_head_{config.n_head}.pt"
+    args = parser.parse_args()
-    checkpoint_filename = f"best_model_checkpoint_n_embd_{config.n_embd}_n_layer_{config.n_layer}_n_head_{config.n_head}.pt"
+
+    config = TrainConfig()
+    config.n_layer = args.n_layer
+    config.n_embd = args.n_embd
+    config.n_head = args.n_head
+    config.max_epoch = args.max_epoch
+    config.batch_size = args.batch_size
+    config.lr_initial = args.lr_initial
+    config.lr_final = args.lr_final
+    config.weight_decay = args.weight_decay
+    config.warmup_epochs = args.warmup_epochs
+    config.early_stopping_patience = args.early_stopping_patience
+    config.pdrop = args.pdrop
+    config.token_pdrop = args.token_pdrop
+    config.betas = tuple(args.betas)
+    config.model_name = args.model
+
+    model_suffix = f"{config.model_name}_n_embd_{config.n_embd}_n_layer_{config.n_layer}_n_head_{config.n_head}"
+    model_filename = f"best_model_{model_suffix}.pt"
+    checkpoint_filename = f"best_model_checkpoint_{model_suffix}.pt"
+
+    # --- 0. Save Configuration ---
+    config_filename = f"config_n_embd_{config.n_embd}_n_layer_{config.n_layer}_n_head_{config.n_head}.json"
+    config_dict = {k: v for k, v in vars(config).items() if not k.startswith('__')}
+    with open(config_filename, 'w') as f:
+        json.dump(config_dict, f, indent=4)
+    print(f"Configuration saved to {config_filename}")
 
     # --- 1. Data Loading ---
     print(f"Loading data from {config.train_data_path} and {config.val_data_path}...")
@@ -63,7 +108,12 @@ def main():
 
     # --- 2. Model, Optimizer, and Loss Initialization ---
     print(f"Initializing model on {config.device}...")
-    model = TimeAwareGPT2(
+    model_cls = {
+        'TimeAwareGPT2': TimeAwareGPT2,
+        'TimeAwareGPT2Learnable': TimeAwareGPT2Learnable,
+    }[config.model_name]
+
+    model = model_cls(
         vocab_size=vocab_size,
         n_embd=config.n_embd,
         n_layer=config.n_layer,
@@ -75,7 +125,7 @@ def main():
     print(f"Model initialized with {model.get_num_params():.2f}M trainable parameters.")
 
     loss_fn = CombinedLoss(config.ignored_token_ids)
-    optimizer = AdamW(model.parameters(), lr=config.lr_initial)
+    optimizer = AdamW(model.parameters(), lr=config.lr_initial, weight_decay=config.weight_decay, betas=config.betas)
 
     # --- 3. Training Loop ---
     best_val_loss = float('inf')
@@ -193,7 +243,7 @@ def main():
         print("\nTraining finished. No best model to save as validation loss never improved.")
 
     # --- Save losses to a txt file ---
-    losses_filename = f"losses_n_embd_{config.n_embd}_n_layer_{config.n_layer}_n_head_{config.n_head}.txt"
+    losses_filename = f"losses_{model_suffix}.txt"
     with open(losses_filename, 'w') as f:
         f.write("epoch,train_loss_ce,train_loss_surv,train_loss_total,val_loss_ce,val_loss_surv,val_loss_total\n")
         for i in range(len(train_losses_total)):

train_dpp.py

@@ -1,400 +0,0 @@
-# train.py (DDP-ready)
-import os
-import math
-import argparse
-import numpy as np
-import tqdm
-import matplotlib.pyplot as plt
-
-import torch
-import torch.nn as nn
-import torch.distributed as dist
-from torch.optim import Adam
-from torch.utils.data import DataLoader, DistributedSampler
-
-from models import TimeAwareGPT2, CombinedLoss
-from utils import PatientEventDataset
-
-
-# --- Configuration ---
-class TrainConfig:
-    # Data parameters
-    train_data_path = 'ukb_real_train.bin'
-    val_data_path = 'ukb_real_val.bin'
-    block_length = 24  # Sequence length
-
-    # Model parameters
-    n_embd = 256
-    n_layer = 8
-    n_head = 8
-    pdrop = 0.1
-    token_pdrop = 0.1
-
-    # Training parameters
-    max_epoch = 200
-    batch_size = 128
-    lr_initial = 6e-4
-    lr_final = 6e-5
-    warmup_epochs = 10
-    early_stopping_patience = 5
-
-    # Loss parameters
-    # 0 = padding, 1 = "no event"
-    ignored_token_ids = [0, 1]
-
-    # System parameters (device 将在 main() 内按 local_rank 动态设置)
-    device = 'cuda' if torch.cuda.is_available() else 'cpu'
-
-
-def setup_distributed(backend: str = "nccl"):
-    """
-    如果由 torchrun 启动且 WORLD_SIZE>1，则初始化分布式。
-    返回 (is_distributed, world_size, rank, local_rank)
-    """
-    world_size = int(os.environ.get("WORLD_SIZE", "1"))
-    is_distributed = world_size > 1
-    if is_distributed:
-        if not dist.is_initialized():
-            dist.init_process_group(backend=backend, init_method="env://")
-        rank = dist.get_rank()
-        local_rank = int(os.environ.get("LOCAL_RANK", "0"))
-        torch.cuda.set_device(local_rank)
-    else:
-        rank = 0
-        local_rank = 0
-    return is_distributed, world_size, rank, local_rank
-
-
-def cleanup_distributed():
-    if dist.is_available() and dist.is_initialized():
-        dist.destroy_process_group()
-
-
-def all_reduce_mean(value: float, device, world_size: int):
-    """
-    value 是 Python float（本进程的和/均值），返回所有进程平均后的 float。
-    """
-    tensor = torch.tensor([value], dtype=torch.float32, device=device)
-    dist.all_reduce(tensor, op=dist.ReduceOp.SUM)
-    tensor /= world_size
-    return float(tensor.item())
-
-
-def main():
-    parser = argparse.ArgumentParser()
-    parser.add_argument("--backend", type=str, default="nccl", choices=["nccl", "gloo", "mpi"])
-    parser.add_argument("--seed", type=int, default=42)
-    args = parser.parse_args()
-
-    # 分布式初始化
-    is_dist, world_size, rank, local_rank = setup_distributed(args.backend)
-
-    # 基本环境
-    torch.manual_seed(args.seed + rank)
-    np.random.seed(args.seed + rank)
-    torch.backends.cudnn.benchmark = True
-
-    config = TrainConfig()
-    device = torch.device(f"cuda:{local_rank}" if torch.cuda.is_available() else "cpu")
-    config.device = device
-
-    is_main = (rank == 0)
-
-    # --- 1. Data Loading ---
-    if is_main:
-        print(f"Loading data from {config.train_data_path} and {config.val_data_path}...")
-    train_data_arr = np.memmap(config.train_data_path, dtype=np.uint32, mode='r').reshape(-1, 3)
-    val_data_arr = np.memmap(config.val_data_path, dtype=np.uint32, mode='r').reshape(-1, 3)
-
-    # Infer vocab_size from the data (max label + 1)
-    vocab_size = int(max(train_data_arr[:, 2].max(), val_data_arr[:, 2].max())) + 1
-    if is_main:
-        print(f"Inferred vocabulary size: {vocab_size}")
-
-    train_dataset = PatientEventDataset(train_data_arr, config.block_length)
-    val_dataset = PatientEventDataset(val_data_arr, config.block_length)
-
-    # 分布式采样器
-    if is_dist:
-        train_sampler = DistributedSampler(train_dataset, num_replicas=world_size, rank=rank, shuffle=True, drop_last=False)
-        val_sampler = DistributedSampler(val_dataset, num_replicas=world_size, rank=rank, shuffle=False, drop_last=False)
-    else:
-        train_sampler = None
-        val_sampler = None
-
-    train_loader = DataLoader(
-        train_dataset,
-        batch_size=config.batch_size,
-        shuffle=(train_sampler is None),
-        sampler=train_sampler,
-        num_workers=4,
-        pin_memory=True,
-        drop_last=False,
-        persistent_workers=True if 4 > 0 else False,
-    )
-    val_loader = DataLoader(
-        val_dataset,
-        batch_size=config.batch_size,
-        shuffle=False,
-        sampler=val_sampler,
-        num_workers=4,
-        pin_memory=True,
-        drop_last=False,
-        persistent_workers=True if 4 > 0 else False,
-    )
-
-    # --- 2. Model, Optimizer, and Loss Initialization ---
-    if is_main:
-        print(f"Initializing model on {config.device}...")
-    model = TimeAwareGPT2(
-        vocab_size=vocab_size,
-        n_embd=config.n_embd,
-        n_layer=config.n_layer,
-        n_head=config.n_head,
-        pdrop=config.pdrop,
-        token_pdrop=config.token_pdrop
-    ).to(device)
-
-    if is_main and hasattr(model, "get_num_params"):
-        print(f"Model initialized with {model.get_num_params():.2f}M trainable parameters.")
-
-    loss_fn = CombinedLoss(config.ignored_token_ids)
-    optimizer = Adam(model.parameters(), lr=config.lr_initial)
-
-    # DDP 包装
-    if is_dist:
-        model = nn.parallel.DistributedDataParallel(
-            model,
-            device_ids=[local_rank] if device.type == "cuda" else None,
-            output_device=local_rank if device.type == "cuda" else None,
-            find_unused_parameters=False,
-        )
-
-    # --- 3. Training Loop ---
-    best_val_loss = float('inf')
-    patience_counter = 0
-
-    # 只在主进程收集与画图
-    train_losses_ce, train_losses_surv, train_losses_total = [], [], []
-    val_losses_ce, val_losses_surv, val_losses_total = [], [], []
-
-    if is_main:
-        print("Starting training...")
-
-    stop_training = False
-
-    for epoch in range(config.max_epoch):
-        # 设置 epoch 给分布式采样器，确保跨 epoch shuffle
-        if is_dist:
-            train_sampler.set_epoch(epoch)
-
-        # --- Learning Rate Scheduling ---
-        if epoch < config.warmup_epochs:
-            lr = config.lr_initial
-        else:
-            progress = (epoch - config.warmup_epochs) / (config.max_epoch - config.warmup_epochs)
-            lr = config.lr_final + 0.5 * (config.lr_initial - config.lr_final) * (1 + math.cos(math.pi * progress))
-        for param_group in optimizer.param_groups:
-            param_group['lr'] = lr
-
-        # --- Training Phase ---
-        if is_main:
-            pbar = tqdm.tqdm(train_loader, desc=f"Epoch {epoch+1}/{config.max_epoch} [Train]")
-        else:
-            pbar = train_loader  # 非主进程禁用 tqdm
-
-        model.train()
-        train_loss_ce_acc, train_loss_surv_acc = 0.0, 0.0
-        train_steps = 0
-
-        for batch in pbar:
-            event_seq, time_seq = batch
-            event_seq = event_seq.to(device, non_blocking=True)
-            time_seq = time_seq.to(device, non_blocking=True)
-
-            # Prepare inputs and targets
-            input_events = event_seq[:, :-1]
-            input_times = time_seq[:, :-1]
-            target_events = event_seq[:, 1:]
-            target_wait_times = (time_seq[:, 1:] - time_seq[:, :-1]).float()
-
-            # Forward pass
-            logits = model(input_events, input_times)
-            loss_ce, loss_survival = loss_fn(logits, target_events, target_wait_times)
-            loss = loss_ce + loss_survival
-
-            # Backward pass and optimization
-            optimizer.zero_grad(set_to_none=True)
-            loss.backward()
-            optimizer.step()
-
-            train_loss_ce_acc += float(loss_ce.item())
-            train_loss_surv_acc += float(loss_survival.item())
-            train_steps += 1
-
-            if is_main and isinstance(pbar, tqdm.tqdm):
-                pbar.set_postfix({'loss_ce': f'{loss_ce.item():.4f}', 'loss_surv': f'{loss_survival.item():.4f}', 'lr': f'{lr:.2e}'})
-
-        # 进程内均值
-        avg_train_loss_ce_local = train_loss_ce_acc / max(train_steps, 1)
-        avg_train_loss_surv_local = train_loss_surv_acc / max(train_steps, 1)
-
-        # 所有进程平均
-        if is_dist:
-            avg_train_loss_ce = all_reduce_mean(avg_train_loss_ce_local, device, world_size)
-            avg_train_loss_surv = all_reduce_mean(avg_train_loss_surv_local, device, world_size)
-        else:
-            avg_train_loss_ce = avg_train_loss_ce_local
-            avg_train_loss_surv = avg_train_loss_surv_local
-
-        if is_main:
-            train_losses_ce.append(avg_train_loss_ce)
-            train_losses_surv.append(avg_train_loss_surv)
-            train_losses_total.append(avg_train_loss_ce + avg_train_loss_surv)
-
-        # --- Validation Phase ---
-        if is_main:
-            pbar_val = tqdm.tqdm(val_loader, desc=f"Epoch {epoch+1}/{config.max_epoch} [Val]")
-        else:
-            pbar_val = val_loader
-
-        model.eval()
-        val_loss_ce_acc, val_loss_surv_acc = 0.0, 0.0
-        val_steps = 0
-
-        with torch.no_grad():
-            for batch in pbar_val:
-                event_seq, time_seq = batch
-                event_seq = event_seq.to(device, non_blocking=True)
-                time_seq = time_seq.to(device, non_blocking=True)
-
-                input_events = event_seq[:, :-1]
-                input_times = time_seq[:, :-1]
-                target_events = event_seq[:, 1:]
-                target_wait_times = (time_seq[:, 1:] - time_seq[:, :-1]).float()
-
-                logits = model(input_events, input_times)
-                loss_ce, loss_survival = loss_fn(logits, target_events, target_wait_times)
-
-                val_loss_ce_acc += float(loss_ce.item())
-                val_loss_surv_acc += float(loss_survival.item())
-                val_steps += 1
-
-                if is_main and isinstance(pbar_val, tqdm.tqdm):
-                    pbar_val.set_postfix({'loss_ce': f'{loss_ce.item():.4f}', 'loss_surv': f'{loss_survival.item():.4f}'})
-
-        avg_val_loss_ce_local = val_loss_ce_acc / max(val_steps, 1)
-        avg_val_loss_surv_local = val_loss_surv_acc / max(val_steps, 1)
-
-        if is_dist:
-            avg_val_loss_ce = all_reduce_mean(avg_val_loss_ce_local, device, world_size)
-            avg_val_loss_surv = all_reduce_mean(avg_val_loss_surv_local, device, world_size)
-        else:
-            avg_val_loss_ce = avg_val_loss_ce_local
-            avg_val_loss_surv = avg_val_loss_surv_local
-
-        total_val_loss = avg_val_loss_ce + avg_val_loss_surv
-
-        # 主进程打印与记录
-        if is_main:
-            print(f"Epoch {epoch+1} Summary: \n"
-                  f"  Train Loss: {avg_train_loss_ce + avg_train_loss_surv:.4f} (CE: {avg_train_loss_ce:.4f}, Surv: {avg_train_loss_surv:.4f})\n"
-                  f"  Val Loss:   {total_val_loss:.4f} (CE: {avg_val_loss_ce:.4f}, Surv: {avg_val_loss_surv:.4f})\n"
-                  f"  Learning Rate: {lr:.6f}")
-            val_losses_ce.append(avg_val_loss_ce)
-            val_losses_surv.append(avg_val_loss_surv)
-            val_losses_total.append(total_val_loss)
-
-        # --- Early Stopping Check (基于聚合后的 total_val_loss) ---
-        improved = False
-        if is_main:
-            if total_val_loss < best_val_loss:
-                best_val_loss = total_val_loss
-                patience_counter = 0
-                improved = True
-                print(f"Validation loss improved to {best_val_loss:.4f}. Saving checkpoint...")
-                # DDP: 保存 module.state_dict()
-                state_dict = model.module.state_dict() if isinstance(model, nn.parallel.DistributedDataParallel) else model.state_dict()
-                torch.save(state_dict, 'best_model_checkpoint.pt')
-            else:
-                if epoch >= config.warmup_epochs:
-                    patience_counter += 1
-                    print(f"Validation loss did not improve. Patience: {patience_counter}/{config.early_stopping_patience}")
-            stop_training = patience_counter >= config.early_stopping_patience
-
-        # 把 improved/stop 广播到所有进程，确保一致退出
-        if is_dist:
-            flag_tensor = torch.tensor([1 if stop_training else 0], device=device, dtype=torch.int32)
-            dist.broadcast(flag_tensor, src=0)
-            stop_training = bool(int(flag_tensor.item()))
-
-        if stop_training:
-            if is_main:
-                print("\nEarly stopping triggered due to no improvement in validation loss.")
-            break
-
-    # --- Save Best Model at the End (只主进程) ---
-    if is_main:
-        if best_val_loss != float('inf'):
-            print(f"\nTraining finished. Loading best model from checkpoint with validation loss {best_val_loss:.4f}.")
-            # 为了易用，这里在主进程上重新构建单卡模型加载权重再保存
-            model_single = TimeAwareGPT2(
-                vocab_size=vocab_size,
-                n_embd=config.n_embd,
-                n_layer=config.n_layer,
-                n_head=config.n_head,
-                pdrop=config.pdrop,
-                token_pdrop=config.token_pdrop
-            ).to('cpu')
-            model_single.load_state_dict(torch.load('best_model_checkpoint.pt', map_location='cpu'))
-            print("Saving final best model to best_model.pt")
-            torch.save(model_single.state_dict(), 'best_model.pt')
-        else:
-            print("\nTraining finished. No best model to save as validation loss never improved.")
-
-        # --- Plot and Save Loss Curves ---
-        num_epochs = len(train_losses_total)
-        if num_epochs > 0:
-            epochs = range(1, num_epochs + 1)
-            plt.figure(figsize=(18, 5))
-
-            # Plot CE Loss
-            plt.subplot(1, 3, 1)
-            plt.plot(epochs, train_losses_ce, label='Train CE')
-            plt.plot(epochs, val_losses_ce, label='Val CE')
-            plt.title('Cross-Entropy Loss')
-            plt.xlabel('Epochs')
-            plt.ylabel('Loss')
-            plt.legend()
-            plt.grid(True)
-
-            # Plot Survival Loss
-            plt.subplot(1, 3, 2)
-            plt.plot(epochs, train_losses_surv, label='Train Survival')
-            plt.plot(epochs, val_losses_surv, label='Val Survival')
-            plt.title('Survival Loss')
-            plt.xlabel('Epochs')
-            plt.ylabel('Loss')
-            plt.legend()
-            plt.grid(True)
-
-            # Plot Total Loss
-            plt.subplot(1, 3, 3)
-            plt.plot(epochs, train_losses_total, label='Train Total')
-            plt.plot(epochs, val_losses_total, label='Val Total')
-            plt.title('Total Loss')
-            plt.xlabel('Epochs')
-            plt.ylabel('Loss')
-            plt.legend()
-            plt.grid(True)
-
-            plt.tight_layout()
-            plt.savefig('loss_curves.png')
-            print("\nLoss curves saved to loss_curves.png")
-
-    # 清理分布式
-    cleanup_distributed()
-
-
-if __name__ == '__main__':
-    main()

train_iter.py

@@ -0,0 +1,218 @@
+import torch
+import torch.nn as nn
+from torch.optim import AdamW
+from torch.utils.data import DataLoader
+import numpy as np
+import math
+import tqdm
+import matplotlib.pyplot as plt
+import json
+import itertools
+
+from models import TimeAwareGPT2, CombinedLoss
+from utils import PatientEventDataset
+
+# --- Configuration ---
+class TrainConfig:
+    # Data parameters
+    train_data_path = 'ukb_real_train.bin'
+    val_data_path = 'ukb_real_val.bin'
+    block_length = 48  # Sequence length
+
+    # Model parameters
+    n_embd = 120
+    n_layer = 12
+    n_head = 12
+    pdrop = 0.0
+    token_pdrop = 0.0
+
+    # Training parameters
+    max_iter = 200000
+    batch_size = 128
+    lr_initial = 6e-4
+    lr_final = 6e-5
+    weight_decay = 2e-1
+    warmup_iter = 1000
+    
+    # Loss parameters
+    # 0 = padding, 1 = "no event"
+    ignored_token_ids = [0, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12]  # Example ignored token IDs
+
+    # System parameters
+    device = 'cuda' if torch.cuda.is_available() else 'cpu'
+
+# --- Main Training Script ---
+def main():
+    config = TrainConfig()
+
+    model_filename = f"best_model_n_embd_{config.n_embd}_n_layer_{config.n_layer}_n_head_{config.n_head}_iter.pt"
+
+    # --- 0. Save Configuration ---
+    config_filename = f"config_n_embd_{config.n_embd}_n_layer_{config.n_layer}_n_head_{config.n_head}_iter.json"
+    config_dict = {k: v for k, v in vars(config).items() if not k.startswith('__')}
+    with open(config_filename, 'w') as f:
+        json.dump(config_dict, f, indent=4)
+    print(f"Configuration saved to {config_filename}")
+
+    # --- 1. Data Loading ---
+    print(f"Loading data from {config.train_data_path} and {config.val_data_path}...")
+    train_data_arr = np.memmap(config.train_data_path, dtype=np.uint32, mode='r').reshape(-1, 3)
+    val_data_arr = np.memmap(config.val_data_path, dtype=np.uint32, mode='r').reshape(-1, 3)
+
+    # Infer vocab_size from the data (max label + 1)
+    vocab_size = int(max(train_data_arr[:, 2].max(), val_data_arr[:, 2].max())) + 1
+    print(f"Inferred vocabulary size: {vocab_size}")
+
+    train_dataset = PatientEventDataset(train_data_arr, config.block_length)
+    val_dataset = PatientEventDataset(val_data_arr, config.block_length)
+
+    train_loader = DataLoader(train_dataset, batch_size=config.batch_size, shuffle=True, num_workers=4, pin_memory=True)
+    val_loader = DataLoader(val_dataset, batch_size=config.batch_size, shuffle=False, num_workers=4, pin_memory=True)
+    train_iter_loader = iter(itertools.cycle(train_loader))
+
+    # --- 2. Model, Optimizer, and Loss Initialization ---
+    print(f"Initializing model on {config.device}...")
+    model = TimeAwareGPT2(
+        vocab_size=vocab_size,
+        n_embd=config.n_embd,
+        n_layer=config.n_layer,
+        n_head=config.n_head,
+        pdrop=config.pdrop,
+        token_pdrop=config.token_pdrop
+    ).to(config.device)
+
+    print(f"Model initialized with {model.get_num_params():.2f}M trainable parameters.")
+
+    loss_fn = CombinedLoss(config.ignored_token_ids)
+    optimizer = AdamW(model.parameters(), lr=config.lr_initial, weight_decay=config.weight_decay, betas=(0.9, 0.99))
+
+    # --- 3. Training Loop ---
+    
+    # Lists to store losses
+    train_losses_ce, train_losses_surv, train_losses_total = [], [], []
+    
+    print("Starting training...")
+    pbar = tqdm.tqdm(range(1, config.max_iter + 1), desc="Training")
+    for iter_num in pbar:
+        # --- Learning Rate Scheduling ---
+        if iter_num < config.warmup_iter:
+            lr = config.lr_initial
+        else:
+            progress = (iter_num - config.warmup_iter) / (config.max_iter - config.warmup_iter)
+            lr = config.lr_final + 0.5 * (config.lr_initial - config.lr_final) * (1 + math.cos(math.pi * progress))
+        
+        for param_group in optimizer.param_groups:
+            param_group['lr'] = lr
+
+        # --- Training Step ---
+        model.train()
+        
+        event_seq, time_seq = next(train_iter_loader)
+        event_seq, time_seq = event_seq.to(config.device), time_seq.to(config.device)
+
+        # Prepare inputs and targets
+        input_events = event_seq[:, :-1]
+        input_times = time_seq[:, :-1]
+        target_events = event_seq[:, 1:]
+        target_wait_times = (time_seq[:, 1:] - time_seq[:, :-1]).float()
+
+        # Forward pass
+        logits = model(input_events, input_times)
+        loss_ce, loss_survival = loss_fn(logits, target_events, target_wait_times)
+        loss = loss_ce + loss_survival
+
+        # Backward pass and optimization
+        optimizer.zero_grad()
+        loss.backward()
+        optimizer.step()
+
+        train_losses_ce.append(loss_ce.item())
+        train_losses_surv.append(loss_survival.item())
+        train_losses_total.append(loss.item())
+        
+        pbar.set_postfix({'loss_ce': f'{loss_ce.item():.4f}', 'loss_surv': f'{loss_survival.item():.4f}', 'lr': f'{lr:.2e}'})
+
+    print("\nTraining finished.")
+
+    # --- 4. Final Validation ---
+    print("Running final validation...")
+    model.eval()
+    val_loss_ce_acc, val_loss_surv_acc = 0.0, 0.0
+    val_steps = 0
+
+    with torch.no_grad():
+        pbar_val = tqdm.tqdm(val_loader, desc="Final Validation")
+        for event_seq, time_seq in pbar_val:
+            event_seq, time_seq = event_seq.to(config.device), time_seq.to(config.device)
+
+            input_events = event_seq[:, :-1]
+            input_times = time_seq[:, :-1]
+            target_events = event_seq[:, 1:]
+            target_wait_times = (time_seq[:, 1:] - time_seq[:, :-1]).float()
+
+            logits = model(input_events, input_times)
+            loss_ce, loss_survival = loss_fn(logits, target_events, target_wait_times)
+            
+            val_loss_ce_acc += loss_ce.item()
+            val_loss_surv_acc += loss_survival.item()
+            val_steps += 1
+            pbar_val.set_postfix({'loss_ce': f'{loss_ce.item():.4f}', 'loss_surv': f'{loss_survival.item():.4f}'})
+
+    avg_val_loss_ce = val_loss_ce_acc / val_steps
+    avg_val_loss_surv = val_loss_surv_acc / val_steps
+    total_val_loss = avg_val_loss_ce + avg_val_loss_surv
+
+    print(f"Final Validation Summary: \n" 
+          f"  Val Loss:   {total_val_loss:.4f} (CE: {avg_val_loss_ce:.4f}, Surv: {avg_val_loss_surv:.4f})")
+
+    # --- 5. Save Model ---
+    print(f"Saving final model to {model_filename}")
+    torch.save(model.state_dict(), model_filename)
+
+    # --- 6. Save and Plot Losses ---
+    losses_filename = f"losses_n_embd_{config.n_embd}_n_layer_{config.n_layer}_n_head_{config.n_head}_iter.txt"
+    with open(losses_filename, 'w') as f:
+        f.write("iteration,train_loss_ce,train_loss_surv,train_loss_total\n")
+        for i in range(len(train_losses_total)):
+            f.write(f"{i+1},{train_losses_ce[i]},{train_losses_surv[i]},{train_losses_total[i]}\n")
+    print(f"\nLosses saved to {losses_filename}")
+
+    # Plot and Save Loss Curves
+    iterations = range(1, len(train_losses_total) + 1)
+
+    plt.figure(figsize=(18, 5))
+
+    # Plot CE Loss
+    plt.subplot(1, 3, 1)
+    plt.plot(iterations, train_losses_ce, label='Train CE')
+    plt.title('Cross-Entropy Loss')
+    plt.xlabel('Iterations')
+    plt.ylabel('Loss')
+    plt.legend()
+    plt.grid(True)
+
+    # Plot Survival Loss
+    plt.subplot(1, 3, 2)
+    plt.plot(iterations, train_losses_surv, label='Train Survival')
+    plt.title('Survival Loss')
+    plt.xlabel('Iterations')
+    plt.ylabel('Loss')
+    plt.legend()
+    plt.grid(True)
+
+    # Plot Total Loss
+    plt.subplot(1, 3, 3)
+    plt.plot(iterations, train_losses_total, label='Train Total')
+    plt.title('Total Loss')
+    plt.xlabel('Iterations')
+    plt.ylabel('Loss')
+    plt.legend()
+    plt.grid(True)
+
+    plt.tight_layout()
+    plt.savefig('loss_curves_iter.png')
+    print("\nLoss curves saved to loss_curves_iter.png")
+
+
+if __name__ == '__main__':
+    main()

utils.py

@@ -1,7 +1,11 @@
+import os
 import torch
 import numpy as np
 import random
 from collections import defaultdict
+import json
+from models import TimeAwareGPT2, TimeAwareGPT2Learnable
+
 
 class PatientEventDataset(torch.utils.data.Dataset):
     """
@@ -39,17 +43,22 @@ class PatientEventDataset(torch.utils.data.Dataset):
         """
         return len(self.patient_ids)
 
-    def __getitem__(self, idx: int) -> tuple[torch.Tensor, torch.Tensor]:
+    def __getitem__(self, idx):
         """
-        Retrieves, processes, and returns a single patient's event sequence.
+        Retrieves, processes, and returns a single patient's event sequence,
+        or a list of sequences if a slice is provided.
 
         Args:
-            idx (int): The index of the patient to retrieve.
+            idx (int or slice): The index or slice of the patient(s) to retrieve.
 
         Returns:
-            A tuple of two torch.long tensors: (event_sequence, time_sequence),
+            If idx is an int, a tuple of two torch.long tensors:
-            both of shape (block_length,).
+            (event_sequence, time_sequence), both of shape (block_length,).
+            If idx is a slice, a list of such tuples.
         """
+        if isinstance(idx, slice):
+            return [self[i] for i in range(*idx.indices(len(self)))]
+
         # 1. Retrieve and Sort
         patient_id = self.patient_ids[idx]
         records = sorted(self.patient_events[patient_id], key=lambda x: x[0])
@@ -102,3 +111,146 @@ class PatientEventDataset(torch.utils.data.Dataset):
         time_tensor = torch.tensor(time_stamps, dtype=torch.long)
 
         return event_tensor, time_tensor
+
+def load_model(config_path: str, device: str = 'cpu'):
+    """
+        Load a trained model based on the training configuration, inferring the
+        checkpoint filename from the configuration.
+
+    According to train.py, models may be either 'TimeAwareGPT2' or
+    'TimeAwareGPT2Learnable'. This function:
+      - Reads the config JSON to get architecture hyperparameters
+      - Selects the model class using config.model_name (defaults to TimeAwareGPT2 if absent)
+            - Infers the checkpoint path from the config values
+            - Infers vocab_size from the checkpoint
+      - Loads weights and returns the model in eval mode on the requested device
+
+    Args:
+        config_path: Path to the JSON configuration file saved during training.
+        device: 'cpu' or 'cuda'.
+
+    Returns:
+        torch.nn.Module: Loaded model ready for inference.
+    """
+    # 1) Read config
+    with open(config_path, 'r') as f:
+        config_dict = json.load(f)
+
+    # Access config entries with attribute-style access while staying tolerant to missing keys
+    class AttrDict(dict):
+        def __getattr__(self, item):
+            try:
+                return self[item]
+            except KeyError:
+                raise AttributeError(item)
+
+    config = AttrDict(config_dict)
+
+    # 2) Decide model class (train.py supports two variants)
+    model_name = getattr(config, 'model_name', 'TimeAwareGPT2')
+    model_cls = {
+        'TimeAwareGPT2': TimeAwareGPT2,
+        'TimeAwareGPT2Learnable': TimeAwareGPT2Learnable,
+    }.get(model_name, TimeAwareGPT2)
+
+    # 3) Infer checkpoint filename from config
+    n_embd = getattr(config, 'n_embd')
+    n_layer = getattr(config, 'n_layer')
+    n_head = getattr(config, 'n_head')
+
+    # Newer naming (includes model_name) used by train.py when model_name is present
+    suffix_with_model = f"{model_name}_n_embd_{n_embd}_n_layer_{n_layer}_n_head_{n_head}"
+    ckpt_with_model = f"best_model_{suffix_with_model}.pt"
+
+    # Older naming (without model_name) matches existing repo files
+    suffix_legacy = f"n_embd_{n_embd}_n_layer_{n_layer}_n_head_{n_head}"
+    ckpt_legacy = f"best_model_{suffix_legacy}.pt"
+
+    # Prefer file that exists on disk
+    if os.path.exists(ckpt_with_model):
+        model_path = ckpt_with_model
+    elif os.path.exists(ckpt_legacy):
+        model_path = ckpt_legacy
+    else:
+        # Fall back to including model_name; if not present in config earlier, user may still have saved this way
+        model_path = ckpt_with_model
+        print(f"Warning: Could not find checkpoint on disk. Expected one of: {ckpt_with_model}, {ckpt_legacy}")
+
+    # 4) Infer vocab_size from checkpoint
+    state_preview = torch.load(model_path, map_location='cpu')
+    if 'wte.weight' in state_preview:
+        vocab_size = state_preview['wte.weight'].shape[0]
+    elif 'head.weight' in state_preview:
+        vocab_size = state_preview['head.weight'].shape[0]
+    else:
+        candidate = None
+        for k, v in state_preview.items():
+            if isinstance(v, torch.Tensor) and v.ndim == 2:
+                V = max(v.shape)
+                if candidate is None or V > candidate:
+                    candidate = V
+        if candidate is None:
+            raise ValueError("Unable to infer vocab_size from checkpoint. Unknown tensor shapes.")
+        vocab_size = candidate
+
+    # 5) Build model from config (tolerant to missing fields)
+    pdrop = getattr(config, 'pdrop', 0.1)
+    token_pdrop = getattr(config, 'token_pdrop', 0.1)
+
+    model = model_cls(
+        vocab_size=vocab_size,
+        n_embd=n_embd,
+        n_layer=n_layer,
+        n_head=n_head,
+        pdrop=pdrop,
+        token_pdrop=token_pdrop,
+    ).to(device)
+
+    # 6) Load weights
+    state_dict = torch.load(model_path, map_location=device)
+    missing, unexpected = model.load_state_dict(state_dict, strict=False)
+
+    if missing:
+        print(f"Warning: Missing keys when loading state_dict: {missing}")
+    if unexpected:
+        print(f"Warning: Unexpected keys when loading state_dict: {unexpected}")
+
+    model.eval()
+    try:
+        num_params_m = model.get_num_params()
+        print(f"Model loaded from {model_path} with {num_params_m:.2f}M parameters.")
+    except Exception:
+        pass
+    return model
+
+
+def get_batch(dataset: PatientEventDataset, batch_slice: slice) -> tuple[torch.Tensor, torch.Tensor, torch.Tensor, torch.Tensor]:
+    """
+    Retrieves a batch of data from a PatientEventDataset and prepares it for model training.
+
+    Args:
+        dataset (PatientEventDataset): The dataset to retrieve data from.
+        batch_slice (slice): The slice defining the batch of patients to retrieve.
+        ignore_tokens (list, optional): A list of token IDs to be ignored in the target events.
+                                        These tokens will be replaced with -100. Defaults to None.
+
+    Returns:
+        A tuple containing four tensors:
+        - input_events: (batch_size, sequence_length - 1)
+        - input_tims: (batch_size, sequence_length - 1)
+        - target_events: (batch_size, sequence_length - 1)
+        - target_times: (batch_size, sequence_length - 1)
+    """
+    batch_data = dataset[batch_slice]
+    
+    input_events = [item[0][:-1] for item in batch_data]
+    input_tims = [item[1][:-1] for item in batch_data]
+    target_events = [item[0][1:] for item in batch_data]
+    target_times = [item[1][1:] for item in batch_data]
+
+    input_events = torch.stack(input_events)
+    input_tims = torch.stack(input_tims)
+    target_events = torch.stack(target_events)
+    target_times = torch.stack(target_times)
+
+    return input_events, input_tims, target_events, target_times