Delete gine.py

gine.py

@@ -1,961 +0,0 @@
-import os
-import json
-import time
-import shutil
-
-import sys
-import csv
-
-# Increase max CSV field size limit
-csv.field_size_limit(sys.maxsize)
-
-import torch
-import torch.nn as nn
-import torch.nn.functional as F
-import numpy as np
-import pandas as pd
-from sklearn.model_selection import train_test_split
-from torch.utils.data import Dataset, DataLoader
-
-from transformers import TrainingArguments, Trainer
-from transformers.trainer_callback import TrainerCallback
-from sklearn.metrics import accuracy_score, f1_score, mean_squared_error, mean_absolute_error
-
-# PyG
-from torch_geometric.nn import GINEConv
-
-# ---------------------------
-# Configuration / Constants
-# ---------------------------
-P_MASK = 0.15
-# Manual max atomic number (user requested)
-MAX_ATOMIC_Z = 85
-# Mask token id
-MASK_ATOM_ID = MAX_ATOMIC_Z + 1
-
-USE_LEARNED_WEIGHTING = True
-
-# GINE / embedding hyperparams requested
-NODE_EMB_DIM = 300   # node embedding dimension
-EDGE_EMB_DIM = 300   # edge embedding dimension
-NUM_GNN_LAYERS = 5
-
-# Other hyperparams
-K_ANCHORS = 6
-OUTPUT_DIR = "./gin_output_5M"
-BEST_MODEL_DIR = os.path.join(OUTPUT_DIR, "best")
-os.makedirs(OUTPUT_DIR, exist_ok=True)
-
-# Data reading settings
-csv_path = "polymer_structures_unified_processed.csv"
-TARGET_ROWS = 5000000
-CHUNKSIZE = 50000
-
-# ---------------------------
-# Helper functions
-# ---------------------------
-
-def safe_get(d: dict, key: str, default=None):
-    return d[key] if (isinstance(d, dict) and key in d) else default
-
-def build_adj_list(edge_index, num_nodes):
-    adj = [[] for _ in range(num_nodes)]
-    if edge_index is None or edge_index.numel() == 0:
-        return adj
-    # edge_index shape [2, E]
-    src = edge_index[0].tolist()
-    dst = edge_index[1].tolist()
-    for u, v in zip(src, dst):
-        # ensure indices are within range
-        if 0 <= u < num_nodes and 0 <= v < num_nodes:
-            adj[u].append(v)
-    return adj
-
-def shortest_path_lengths_hops(edge_index, num_nodes):
-    """
-    Compute all-pairs shortest path lengths in hops (BFS per node).
-    Returns an (num_nodes, num_nodes) numpy array with int distances; unreachable -> large number (e.g., num_nodes+1)
-    """
-    adj = build_adj_list(edge_index, num_nodes)
-    INF = num_nodes + 1
-    dist_mat = np.full((num_nodes, num_nodes), INF, dtype=np.int32)
-    for s in range(num_nodes):
-        # BFS
-        q = [s]
-        dist_mat[s, s] = 0
-        head = 0
-        while head < len(q):
-            u = q[head]; head += 1
-            for v in adj[u]:
-                if dist_mat[s, v] == INF:
-                    dist_mat[s, v] = dist_mat[s, u] + 1
-                    q.append(v)
-    return dist_mat
-
-def match_edge_attr_to_index(edge_index: torch.Tensor, edge_attr: torch.Tensor, target_dim: int = 3):
-    """
-    Ensure edge_attr has shape [E_index, D]. Handles common mismatches:
-      - If edge_attr is empty/None -> returns zeros of shape [E_index, target_dim].
-      - If edge_attr.size(0) == edge_index.size(1) -> return as-is.
-      - If edge_attr.size(0) * 2 == edge_index.size(1) -> duplicate (common when features only for undirected edges).
-      - Otherwise repeat/truncate edge_attr to match E_index (safe fallback).
-    """
-    E_idx = edge_index.size(1) if (edge_index is not None and edge_index.numel() > 0) else 0
-    if E_idx == 0:
-        return torch.zeros((0, target_dim), dtype=torch.float)
-    if edge_attr is None or edge_attr.numel() == 0:
-        return torch.zeros((E_idx, target_dim), dtype=torch.float)
-    E_attr = edge_attr.size(0)
-    if E_attr == E_idx:
-        # already matches
-        if edge_attr.size(1) != target_dim:
-            # pad/truncate feature dimension to target_dim
-            D = edge_attr.size(1)
-            if D < target_dim:
-                pad = torch.zeros((E_attr, target_dim - D), dtype=torch.float, device=edge_attr.device)
-                return torch.cat([edge_attr, pad], dim=1)
-            else:
-                return edge_attr[:, :target_dim]
-        return edge_attr
-    # common case: features provided for undirected edges while edge_index contains both directions
-    if E_attr * 2 == E_idx:
-        try:
-            return torch.cat([edge_attr, edge_attr], dim=0)
-        except Exception:
-            # fallback to repeat below
-            pass
-    # fallback: repeat/truncate edge_attr to fit E_idx
-    reps = (E_idx + E_attr - 1) // E_attr
-    edge_rep = edge_attr.repeat(reps, 1)[:E_idx]
-    if edge_rep.size(1) != target_dim:
-        D = edge_rep.size(1)
-        if D < target_dim:
-            pad = torch.zeros((E_idx, target_dim - D), dtype=torch.float, device=edge_rep.device)
-            edge_rep = torch.cat([edge_rep, pad], dim=1)
-        else:
-            edge_rep = edge_rep[:, :target_dim]
-    return edge_rep
-
-# ---------------------------
-# 1. Load Data from `graph` column (chunked)
-# ---------------------------
-node_atomic_lists = []
-node_chirality_lists = []
-node_charge_lists = []
-edge_index_lists = []
-edge_attr_lists = []
-num_nodes_list = []
-rows_read = 0
-
-for chunk in pd.read_csv(csv_path, engine="python", chunksize=CHUNKSIZE):
-    for idx, row in chunk.iterrows():
-        # Prefer 'graph' column JSON (string) per user request
-        graph_field = None
-        if "graph" in row and not pd.isna(row["graph"]):
-            try:
-                graph_field = json.loads(row["graph"])
-            except Exception:
-                # If already parsed or other format
-                try:
-                    graph_field = row["graph"]
-                except Exception:
-                    graph_field = None
-        else:
-            # If no graph column, skip (user requested to use graph column)
-            continue
-
-        if graph_field is None:
-            continue
-
-        # NODE FEATURES
-        node_features = safe_get(graph_field, "node_features", None)
-        if not node_features:
-            # skip graphs without node_features
-            continue
-
-        atomic_nums = []
-        chirality_vals = []
-        formal_charges = []
-
-        for nf in node_features:
-            # atomic number
-            an = safe_get(nf, "atomic_num", None)
-            if an is None:
-                # try alternate keys
-                an = safe_get(nf, "atomic_number", 0)
-            # chirality (use 0 default)
-            ch = safe_get(nf, "chirality", 0)
-            # formal charge (use 0 default)
-            fc = safe_get(nf, "formal_charge", 0)
-            atomic_nums.append(int(an))
-            chirality_vals.append(float(ch))
-            formal_charges.append(float(fc))
-
-        n_nodes = len(atomic_nums)
-
-        # EDGE INDICES & FEATURES
-        edge_indices_raw = safe_get(graph_field, "edge_indices", None)
-        edge_features_raw = safe_get(graph_field, "edge_features", None)
-
-        if edge_indices_raw is None:
-            # try adjacency_matrix to infer edges
-            adj_mat = safe_get(graph_field, "adjacency_matrix", None)
-            if adj_mat:
-                # adjacency_matrix is list of lists
-                srcs = []
-                dsts = []
-                for i, row_adj in enumerate(adj_mat):
-                    for j, val in enumerate(row_adj):
-                        if val:
-                            srcs.append(i)
-                            dsts.append(j)
-                edge_index = torch.tensor([srcs, dsts], dtype=torch.long)
-                # no edge features available -> create zeros matching edges
-                E = edge_index.size(1)
-                edge_attr = torch.zeros((E, 3), dtype=torch.float)
-            else:
-                # no edges found -> skip this graph (GINE requires edges)
-                continue
-        else:
-            # edge_indices_raw expected like [[u,v], [u2,v2], ...] or [[u1,u2,...],[v1,v2,...]]
-            if isinstance(edge_indices_raw, list) and len(edge_indices_raw) > 0 and isinstance(edge_indices_raw[0], list):
-                # Could be list of pairs or list of lists
-                if all(len(pair) == 2 and isinstance(pair[0], int) for pair in edge_indices_raw):
-                    # list of pairs
-                    srcs = [int(p[0]) for p in edge_indices_raw]
-                    dsts = [int(p[1]) for p in edge_indices_raw]
-                elif isinstance(edge_indices_raw[0][0], int):
-                    # Possibly already in [[srcs],[dsts]] format
-                    try:
-                        srcs = [int(x) for x in edge_indices_raw[0]]
-                        dsts = [int(x) for x in edge_indices_raw[1]]
-                    except Exception:
-                        # fallback
-                        srcs = []
-                        dsts = []
-                else:
-                    srcs = []
-                    dsts = []
-            else:
-                srcs = []
-                dsts = []
-
-            if len(srcs) == 0:
-                # fallback: skip graph
-                continue
-
-            edge_index = torch.tensor([srcs, dsts], dtype=torch.long)
-
-            # Build edge_attr matrix with 3 features: bond_type, stereo, is_conjugated (as float)
-            if edge_features_raw and isinstance(edge_features_raw, list):
-                bond_types = []
-                stereos = []
-                is_conjs = []
-                for ef in edge_features_raw:
-                    bt = safe_get(ef, "bond_type", 0)
-                    st = safe_get(ef, "stereo", 0)
-                    ic = safe_get(ef, "is_conjugated", False)
-                    bond_types.append(float(bt))
-                    stereos.append(float(st))
-                    is_conjs.append(float(1.0 if ic else 0.0))
-                edge_attr = torch.tensor(np.stack([bond_types, stereos, is_conjs], axis=1), dtype=torch.float)
-            else:
-                # no edge features -> zeros
-                E = edge_index.size(1)
-                edge_attr = torch.zeros((E, 3), dtype=torch.float)
-
-        # Ensure edge_attr length matches edge_index (fix common mismatches)
-        edge_attr = match_edge_attr_to_index(edge_index, edge_attr, target_dim=3)
-
-        # NOTE: we explicitly DO NOT parse or use coordinates (geometry) anywhere.
-
-        # Save lists
-        node_atomic_lists.append(torch.tensor(atomic_nums, dtype=torch.long))
-        node_chirality_lists.append(torch.tensor(chirality_vals, dtype=torch.float))
-        node_charge_lists.append(torch.tensor(formal_charges, dtype=torch.float))
-        edge_index_lists.append(edge_index)
-        edge_attr_lists.append(edge_attr)
-        num_nodes_list.append(n_nodes)
-
-        rows_read += 1
-        if rows_read >= TARGET_ROWS:
-            break
-    if rows_read >= TARGET_ROWS:
-        break
-
-if len(node_atomic_lists) == 0:
-    raise RuntimeError("No graphs were parsed from the CSV 'graph' column. Check input file and format.")
-
-print(f"Parsed {len(node_atomic_lists)} graphs (using 'graph' column). Using manual max atomic Z = {MAX_ATOMIC_Z}")
-
-# ---------------------------
-# 2. Train/Val Split
-# ---------------------------
-indices = list(range(len(node_atomic_lists)))
-train_idx, val_idx = train_test_split(indices, test_size=0.2, random_state=42)
-
-def subset(l, idxs):
-    return [l[i] for i in idxs]
-
-train_atomic = subset(node_atomic_lists, train_idx)
-train_chirality = subset(node_chirality_lists, train_idx)
-train_charge = subset(node_charge_lists, train_idx)
-train_edge_index = subset(edge_index_lists, train_idx)
-train_edge_attr = subset(edge_attr_lists, train_idx)
-train_num_nodes = subset(num_nodes_list, train_idx)
-
-val_atomic = subset(node_atomic_lists, val_idx)
-val_chirality = subset(node_chirality_lists, val_idx)
-val_charge = subset(node_charge_lists, val_idx)
-val_edge_index = subset(edge_index_lists, val_idx)
-val_edge_attr = subset(edge_attr_lists, val_idx)
-val_num_nodes = subset(num_nodes_list, val_idx)
-
-# ---------------------------
-# Compute class weights (for weighted CE)
-# ---------------------------
-num_classes = MASK_ATOM_ID + 1
-counts = np.ones((num_classes,), dtype=np.float64)
-for z in train_atomic:
-    vals = z.cpu().numpy().astype(int)
-    for v in vals:
-        if 0 <= v < num_classes:
-            counts[v] += 1.0
-freq = counts / counts.sum()
-inv_freq = 1.0 / (freq + 1e-12)
-class_weights = inv_freq / inv_freq.mean()
-class_weights = torch.tensor(class_weights, dtype=torch.float)
-class_weights[MASK_ATOM_ID] = 1.0
-
-# ---------------------------
-# 3. Dataset and Collator (build MLM masks + invariant distance targets using hop counts only)
-# ---------------------------
-class PolymerDataset(Dataset):
-    def __init__(self, atomic_list, chirality_list, charge_list, edge_index_list, edge_attr_list, num_nodes_list):
-        self.atomic_list = atomic_list
-        self.chirality_list = chirality_list
-        self.charge_list = charge_list
-        self.edge_index_list = edge_index_list
-        self.edge_attr_list = edge_attr_list
-        self.num_nodes_list = num_nodes_list
-
-    def __len__(self):
-        return len(self.atomic_list)
-
-    def __getitem__(self, idx):
-        return {
-            "z": self.atomic_list[idx],                     # [n_nodes]
-            "chirality": self.chirality_list[idx],         # [n_nodes] float
-            "formal_charge": self.charge_list[idx],        # [n_nodes]
-            "edge_index": self.edge_index_list[idx],       # [2, E]
-            "edge_attr": self.edge_attr_list[idx],         # [E, 3]
-            "num_nodes": int(self.num_nodes_list[idx])     # int
-        }
-
-def collate_batch(batch):
-    """
-    Builds a batched structure from a list of graph dicts.
-    Returns:
-      - z: [N_total] long (atomic numbers possibly masked)
-      - chirality: [N_total] float
-      - formal_charge: [N_total] float
-      - edge_index: [2, E_total] long (node indices offset per graph)
-      - edge_attr: [E_total, 3] float
-      - batch: [N_total] long mapping node->graph idx
-      - labels_z: [N_total] long (-100 for unselected)
-      - labels_dists: [N_total, K_ANCHORS] float (hop counts)
-      - labels_dists_mask: [N_total, K_ANCHORS] bool
-    Distance targets:
-      - Shortest-path hop distances computed from edge_index for every graph.
-    """
-    all_z = []
-    all_ch = []
-    all_fc = []
-    all_labels_z = []
-    all_labels_dists = []
-    all_labels_dists_mask = []
-    batch_idx = []
-    edge_index_list_batched = []
-    edge_attr_list_batched = []
-    node_offset = 0
-    total_nodes = 0
-    total_edges = 0
-
-    for i, g in enumerate(batch):
-        z = g["z"]                       # tensor [n]
-        n = z.size(0)
-        if n == 0:
-            continue
-
-        chir = g["chirality"]
-        fc = g["formal_charge"]
-        edge_index = g["edge_index"]
-        edge_attr = g["edge_attr"]
-
-        # Mask selection
-        is_selected = torch.rand(n) < P_MASK
-        if is_selected.all():
-            is_selected[torch.randint(0, n, (1,))] = False
-
-        labels_z = torch.full((n,), -100, dtype=torch.long)
-        labels_dists = torch.zeros((n, K_ANCHORS), dtype=torch.float)
-        labels_dists_mask = torch.zeros((n, K_ANCHORS), dtype=torch.bool)
-        labels_z[is_selected] = z[is_selected]
-
-        # BERT-style corruption on atomic numbers
-        z_masked = z.clone()
-        if is_selected.any():
-            sel_idx = torch.nonzero(is_selected).squeeze(-1)
-            rand_atomic = torch.randint(1, MAX_ATOMIC_Z + 1, (sel_idx.size(0),), dtype=torch.long)
-            probs = torch.rand(sel_idx.size(0))
-            mask_choice = probs < 0.8
-            rand_choice = (probs >= 0.8) & (probs < 0.9)
-            if mask_choice.any():
-                z_masked[sel_idx[mask_choice]] = MASK_ATOM_ID
-            if rand_choice.any():
-                z_masked[sel_idx[rand_choice]] = rand_atomic[rand_choice]
-            # keep_choice -> do nothing
-
-        # Build invariant distance targets using hop distances only
-        visible_idx = torch.nonzero(~is_selected).squeeze(-1)
-        if visible_idx.numel() == 0:
-            visible_idx = torch.arange(n, dtype=torch.long)
-
-        # compute hop distances via BFS using edge_index
-        ei = edge_index.clone()
-        num_nodes_local = n
-        dist_mat = shortest_path_lengths_hops(ei, num_nodes_local)  # numpy int matrix
-        for a in torch.nonzero(is_selected).squeeze(-1).tolist():
-            # distances to visible nodes
-            vis = visible_idx.numpy()
-            if vis.size == 0:
-                continue
-            dists = dist_mat[a, vis].astype(np.float32)
-            # filter unreachable (INF = n+1)
-            valid_mask = dists <= num_nodes_local
-            if not valid_mask.any():
-                continue
-            dists_valid = dists[valid_mask]
-            vis_valid = vis[valid_mask]
-            # choose smallest hop distances
-            k = min(K_ANCHORS, dists_valid.size)
-            idx_sorted = np.argsort(dists_valid)[:k]
-            selected_vals = dists_valid[idx_sorted]
-            labels_dists[a, :k] = torch.tensor(selected_vals, dtype=torch.float)
-            labels_dists_mask[a, :k] = True
-
-        # Append node-level tensors to batched lists
-        all_z.append(z_masked)
-        all_ch.append(chir)
-        all_fc.append(fc)
-        all_labels_z.append(labels_z)
-        all_labels_dists.append(labels_dists)
-        all_labels_dists_mask.append(labels_dists_mask)
-        batch_idx.append(torch.full((n,), i, dtype=torch.long))
-
-        # Offset edge indices and append
-        if edge_index is not None and edge_index.numel() > 0:
-            ei_offset = edge_index + node_offset
-            edge_index_list_batched.append(ei_offset)
-            # edge_attr already matched earlier; still ensure shapes here for safety
-            edge_attr_matched = match_edge_attr_to_index(edge_index, edge_attr, target_dim=3)
-            edge_attr_list_batched.append(edge_attr_matched)
-            total_edges += edge_index.size(1)
-
-        node_offset += n
-        total_nodes += n
-
-    if len(all_z) == 0:
-        # Return empty structured batch
-        return {
-            "z": torch.tensor([], dtype=torch.long),
-            "chirality": torch.tensor([], dtype=torch.float),
-            "formal_charge": torch.tensor([], dtype=torch.float),
-            "edge_index": torch.tensor([[], []], dtype=torch.long),
-            "edge_attr": torch.tensor([], dtype=torch.float).reshape(0, 3),
-            "batch": torch.tensor([], dtype=torch.long),
-            "labels_z": torch.tensor([], dtype=torch.long),
-            "labels_dists": torch.tensor([], dtype=torch.float).reshape(0, K_ANCHORS),
-            "labels_dists_mask": torch.tensor([], dtype=torch.bool).reshape(0, K_ANCHORS)
-        }
-
-    z_batch = torch.cat(all_z, dim=0)
-    chir_batch = torch.cat(all_ch, dim=0)
-    fc_batch = torch.cat(all_fc, dim=0)
-    labels_z_batch = torch.cat(all_labels_z, dim=0)
-    labels_dists_batch = torch.cat(all_labels_dists, dim=0)
-    labels_dists_mask_batch = torch.cat(all_labels_dists_mask, dim=0)
-    batch_batch = torch.cat(batch_idx, dim=0)
-
-    if len(edge_index_list_batched) > 0:
-        edge_index_batched = torch.cat(edge_index_list_batched, dim=1)
-        edge_attr_batched = torch.cat(edge_attr_list_batched, dim=0)
-    else:
-        edge_index_batched = torch.tensor([[], []], dtype=torch.long)
-        edge_attr_batched = torch.tensor([], dtype=torch.float).reshape(0, 3)
-
-    return {
-        "z": z_batch,
-        "chirality": chir_batch,
-        "formal_charge": fc_batch,
-        "edge_index": edge_index_batched,
-        "edge_attr": edge_attr_batched,
-        "batch": batch_batch,
-        "labels_z": labels_z_batch,
-        "labels_dists": labels_dists_batch,
-        "labels_dists_mask": labels_dists_mask_batch
-    }
-
-train_dataset = PolymerDataset(train_atomic, train_chirality, train_charge, train_edge_index, train_edge_attr, train_num_nodes)
-val_dataset   = PolymerDataset(val_atomic, val_chirality, val_charge, val_edge_index, val_edge_attr, val_num_nodes)
-
-train_loader = DataLoader(train_dataset, batch_size=16, shuffle=True, collate_fn=collate_batch)
-val_loader   = DataLoader(val_dataset, batch_size=8, shuffle=False, collate_fn=collate_batch)
-
-# ---------------------------
-# 4. Model Definition (GINE-based masked model)
-# ---------------------------
-class GineBlock(nn.Module):
-    def __init__(self, node_dim):
-        super().__init__()
-        # MLP used by GINEConv: map (node_dim) -> node_dim
-        self.mlp = nn.Sequential(
-            nn.Linear(node_dim, node_dim),
-            nn.ReLU(),
-            nn.Linear(node_dim, node_dim)
-        )
-        self.conv = GINEConv(self.mlp)
-        self.bn = nn.BatchNorm1d(node_dim)
-        self.act = nn.ReLU()
-
-    def forward(self, x, edge_index, edge_attr):
-        # GINEConv accepts edge_attr; edge_attr should be same dim as x (or handled in MLP inside)
-        x = self.conv(x, edge_index, edge_attr)
-        x = self.bn(x)
-        x = self.act(x)
-        return x
-
-class MaskedGINE(nn.Module):
-    def __init__(self,
-                 node_emb_dim=NODE_EMB_DIM,
-                 edge_emb_dim=EDGE_EMB_DIM,
-                 num_layers=NUM_GNN_LAYERS,
-                 max_atomic_z=MAX_ATOMIC_Z,
-                 class_weights=None):
-        super().__init__()
-        self.node_emb_dim = node_emb_dim
-        self.edge_emb_dim = edge_emb_dim
-        self.max_atomic_z = max_atomic_z
-
-        # Embedding for atomic numbers (including MASK token)
-        num_embeddings = MASK_ATOM_ID + 1
-        self.atom_emb = nn.Embedding(num_embeddings=num_embeddings, embedding_dim=node_emb_dim, padding_idx=None)
-
-        # Small MLP to map numeric node attributes (chirality, formal_charge) -> node_emb_dim
-        self.node_attr_proj = nn.Sequential(
-            nn.Linear(2, node_emb_dim),
-            nn.ReLU(),
-            nn.Linear(node_emb_dim, node_emb_dim)
-        )
-
-        # Edge encoder: maps 3-dim raw edge features -> edge_emb_dim
-        self.edge_encoder = nn.Sequential(
-            nn.Linear(3, edge_emb_dim),
-            nn.ReLU(),
-            nn.Linear(edge_emb_dim, edge_emb_dim)
-        )
-
-        # Project edge_emb -> node_emb_dim if needed (registered in __init__ to avoid dynamic creation)
-        if edge_emb_dim != node_emb_dim:
-            self._edge_to_node_proj = nn.Linear(edge_emb_dim, node_emb_dim)
-        else:
-            self._edge_to_node_proj = None
-
-        # GINE layers
-        self.gnn_layers = nn.ModuleList([GineBlock(node_emb_dim) for _ in range(num_layers)])
-
-        # Heads
-        num_classes_local = MASK_ATOM_ID + 1
-        self.atom_head = nn.Linear(node_emb_dim, num_classes_local)
-        self.coord_head = nn.Linear(node_emb_dim, K_ANCHORS)
-
-        # Learned uncertainty weighting
-        if USE_LEARNED_WEIGHTING:
-            self.log_var_z = nn.Parameter(torch.zeros(1))
-            self.log_var_pos = nn.Parameter(torch.zeros(1))
-        else:
-            self.log_var_z = None
-            self.log_var_pos = None
-
-        if class_weights is not None:
-            self.register_buffer("class_weights", class_weights)
-        else:
-            self.class_weights = None
-
-    def forward(self, z, chirality, formal_charge, edge_index, edge_attr,
-                batch=None, labels_z=None, labels_dists=None, labels_dists_mask=None):
-        """
-        z: [N] long (atomic numbers or MASK_ATOM_ID)
-        chirality: [N] float
-        formal_charge: [N] float
-        edge_index: [2, E] long (global batched indices)
-        edge_attr: [E, 3] float
-        batch: [N] long mapping nodes->graph idx
-        labels_*: optional supervision targets as in collate_batch
-        """
-        if batch is None:
-            batch = torch.zeros(z.size(0), dtype=torch.long, device=z.device)
-
-        # Node embedding
-        atom_embedding = self.atom_emb(z)  # [N, node_emb_dim]
-        node_attr = torch.stack([chirality, formal_charge], dim=1)  # [N,2]
-        node_attr_emb = self.node_attr_proj(node_attr.to(atom_embedding.device))  # [N, node_emb_dim]
-
-        x = atom_embedding + node_attr_emb  # combine categorical and numeric node features
-
-        # Edge embedding
-        if edge_attr is None or edge_attr.numel() == 0:
-            E = 0 if edge_attr is None else edge_attr.size(0)
-            edge_emb = torch.zeros((E, self.edge_emb_dim), dtype=torch.float, device=x.device)
-        else:
-            edge_emb = self.edge_encoder(edge_attr.to(x.device))  # [E, edge_emb_dim]
-
-        # For GINEConv, edge_attr should match node feature dim used inside GINE (GINE uses the provided nn to process x_j + edge_attr)
-        # Project edge_emb -> node_emb_dim if dims differ (registered in __init__)
-        if self._edge_to_node_proj is not None:
-            edge_for_conv = self._edge_to_node_proj(edge_emb)
-        else:
-            edge_for_conv = edge_emb
-
-        # Run GNN layers
-        h = x
-        for layer in self.gnn_layers:
-            h = layer(h, edge_index.to(h.device), edge_for_conv)
-
-        logits = self.atom_head(h)      # [N, num_classes]
-        dists_pred = self.coord_head(h)  # [N, K_ANCHORS]
-
-        # Compute loss if labels provided
-        if labels_z is not None and labels_dists is not None and labels_dists_mask is not None:
-            mask = labels_z != -100
-            if mask.sum() == 0:
-                return torch.tensor(0.0, device=z.device)
-
-            logits_masked = logits[mask]
-            dists_pred_masked = dists_pred[mask]
-            labels_z_masked = labels_z[mask]
-            labels_dists_masked = labels_dists[mask]
-            labels_dists_mask_mask = labels_dists_mask[mask]
-
-            # classification loss
-            if self.class_weights is not None:
-                loss_z = F.cross_entropy(logits_masked, labels_z_masked.to(logits_masked.device), weight=self.class_weights.to(logits_masked.device))
-            else:
-                loss_z = F.cross_entropy(logits_masked, labels_z_masked.to(logits_masked.device))
-
-            # distance loss (only where mask true)
-            if labels_dists_mask_mask.any():
-                preds = dists_pred_masked[labels_dists_mask_mask]
-                trues = labels_dists_masked[labels_dists_mask_mask].to(preds.device)
-                loss_pos = F.mse_loss(preds, trues, reduction="mean")
-            else:
-                loss_pos = torch.tensor(0.0, device=z.device)
-
-            if USE_LEARNED_WEIGHTING:
-                lz = torch.exp(-self.log_var_z) * loss_z + self.log_var_z
-                lp = torch.exp(-self.log_var_pos) * loss_pos + self.log_var_pos
-                loss = 0.5 * (lz + lp)
-            else:
-                alpha = 1.0
-                loss = loss_z + alpha * loss_pos
-
-            return loss
-
-        return logits, dists_pred
-
-# Instantiate model
-model = MaskedGINE(node_emb_dim=NODE_EMB_DIM,
-                   edge_emb_dim=EDGE_EMB_DIM,
-                   num_layers=NUM_GNN_LAYERS,
-                   max_atomic_z=MAX_ATOMIC_Z,
-                   class_weights=class_weights)
-
-device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
-model.to(device)
-
-# ---------------------------
-# 5. Training Setup (Hugging Face Trainer)
-# ---------------------------
-training_args = TrainingArguments(
-    output_dir=OUTPUT_DIR,
-    overwrite_output_dir=True,
-    num_train_epochs=25,
-    per_device_train_batch_size=16,
-    per_device_eval_batch_size=8,
-    gradient_accumulation_steps=4,
-    eval_strategy="epoch",
-    logging_steps=500,
-    learning_rate=1e-4,
-    weight_decay=0.01,
-    fp16=torch.cuda.is_available(),
-    save_strategy="no",
-    disable_tqdm=False,
-    logging_first_step=True,
-    report_to=[],
-    dataloader_num_workers=4,
-)
-
-class ValLossCallback(TrainerCallback):
-    def __init__(self, trainer_ref=None):
-        self.best_val_loss = float("inf")
-        self.epochs_no_improve = 0
-        self.patience = 10
-        self.best_epoch = None
-        self.trainer_ref = trainer_ref
-
-    def on_epoch_end(self, args, state, control, **kwargs):
-        epoch_num = int(state.epoch)
-        train_loss = next((x["loss"] for x in reversed(state.log_history) if "loss" in x), None)
-        print(f"\n=== Epoch {epoch_num}/{args.num_train_epochs} ===")
-        if train_loss is not None:
-            print(f"Train Loss: {train_loss:.4f}")
-
-    def on_evaluate(self, args, state, control, metrics=None, **kwargs):
-        epoch_num = int(state.epoch) + 1
-        if self.trainer_ref is None:
-            print(f"[Eval] Epoch {epoch_num} - metrics (trainer_ref missing): {metrics}")
-            return
-
-        metric_val_loss = None
-        if metrics is not None:
-            metric_val_loss = metrics.get("eval_loss")
-
-        model_eval = self.trainer_ref.model
-        model_eval.eval()
-        device_local = next(model_eval.parameters()).device if any(p.numel() > 0 for p in model_eval.parameters()) else torch.device("cpu")
-
-        preds_z_all = []
-        true_z_all = []
-        pred_dists_all = []
-        true_dists_all = []
-        total_loss = 0.0
-        n_batches = 0
-
-        logits_masked_list = []
-        labels_masked_list = []
-
-        with torch.no_grad():
-            for batch in val_loader:
-                z = batch["z"].to(device_local)
-                chir = batch["chirality"].to(device_local)
-                fc = batch["formal_charge"].to(device_local)
-                edge_index = batch["edge_index"].to(device_local)
-                edge_attr = batch["edge_attr"].to(device_local)
-                batch_idx = batch["batch"].to(device_local)
-                labels_z = batch["labels_z"].to(device_local)
-                labels_dists = batch["labels_dists"].to(device_local)
-                labels_dists_mask = batch["labels_dists_mask"].to(device_local)
-
-                try:
-                    loss = model_eval(z, chir, fc, edge_index, edge_attr, batch_idx, labels_z, labels_dists, labels_dists_mask)
-                except Exception as e:
-                    loss = None
-
-                if isinstance(loss, torch.Tensor):
-                    total_loss += loss.item()
-                    n_batches += 1
-
-                logits, dists_pred = model_eval(z, chir, fc, edge_index, edge_attr, batch_idx)
-
-                mask = labels_z != -100
-                if mask.sum().item() == 0:
-                    continue
-
-                logits_masked_list.append(logits[mask])
-                labels_masked_list.append(labels_z[mask])
-
-                pred_z = torch.argmax(logits[mask], dim=-1)
-                true_z = labels_z[mask]
-
-                # flatten valid distances
-                pred_d = dists_pred[mask][labels_dists_mask[mask]]
-                true_d = labels_dists[mask][labels_dists_mask[mask]]
-
-                if pred_d.numel() > 0:
-                    pred_dists_all.extend(pred_d.cpu().tolist())
-                    true_dists_all.extend(true_d.cpu().tolist())
-
-                preds_z_all.extend(pred_z.cpu().tolist())
-                true_z_all.extend(true_z.cpu().tolist())
-
-        avg_val_loss = metric_val_loss if metric_val_loss is not None else ((total_loss / n_batches) if n_batches > 0 else float("nan"))
-
-        accuracy = accuracy_score(true_z_all, preds_z_all) if len(true_z_all) > 0 else 0.0
-        f1 = f1_score(true_z_all, preds_z_all, average="weighted") if len(true_z_all) > 0 else 0.0
-        rmse = np.sqrt(mean_squared_error(true_dists_all, pred_dists_all)) if len(true_dists_all) > 0 else 0.0
-        mae = mean_absolute_error(true_dists_all, pred_dists_all) if len(true_dists_all) > 0 else 0.0
-
-        if len(logits_masked_list) > 0:
-            all_logits_masked = torch.cat(logits_masked_list, dim=0)
-            all_labels_masked = torch.cat(labels_masked_list, dim=0)
-            cw = getattr(model_eval, "class_weights", None)
-            if cw is not None:
-                cw_device = cw.to(device_local)
-                try:
-                    loss_z_all = F.cross_entropy(all_logits_masked, all_labels_masked, weight=cw_device)
-                except Exception:
-                    loss_z_all = F.cross_entropy(all_logits_masked, all_labels_masked)
-            else:
-                loss_z_all = F.cross_entropy(all_logits_masked, all_labels_masked)
-            try:
-                perplexity = float(torch.exp(loss_z_all).cpu().item())
-            except Exception:
-                perplexity = float(np.exp(float(loss_z_all.cpu().item())))
-        else:
-            perplexity = float("nan")
-
-        print(f"\n--- Evaluation after Epoch {epoch_num} ---")
-        print(f"Validation Loss: {avg_val_loss:.4f}")
-        print(f"Validation Accuracy: {accuracy:.4f}")
-        print(f"Validation F1 (weighted): {f1:.4f}")
-        print(f"Validation RMSE (distances): {rmse:.4f}")
-        print(f"Validation MAE  (distances): {mae:.4f}")
-        print(f"Validation Perplexity (classification head): {perplexity:.4f}")
-
-        # Save best model by val loss
-        if avg_val_loss is not None and not (isinstance(avg_val_loss, float) and np.isnan(avg_val_loss)) and avg_val_loss < self.best_val_loss - 1e-6:
-            self.best_val_loss = avg_val_loss
-            self.best_epoch = int(state.epoch)
-            self.epochs_no_improve = 0
-            os.makedirs(BEST_MODEL_DIR, exist_ok=True)
-            try:
-                torch.save(self.trainer_ref.model.state_dict(), os.path.join(BEST_MODEL_DIR, "pytorch_model.bin"))
-                print(f"Saved new best model (epoch {epoch_num}) to {os.path.join(BEST_MODEL_DIR, 'pytorch_model.bin')}")
-            except Exception as e:
-                print(f"Failed to save best model at epoch {epoch_num}: {e}")
-        else:
-            self.epochs_no_improve += 1
-
-        if self.epochs_no_improve >= self.patience:
-            print(f"Early stopping after {self.patience} epochs with no improvement.")
-            control.should_training_stop = True
-
-# Create callback and Trainer
-callback = ValLossCallback()
-trainer = Trainer(
-    model=model,
-    args=training_args,
-    train_dataset=train_dataset,
-    eval_dataset=val_dataset,
-    data_collator=collate_batch,
-    callbacks=[callback]
-)
-callback.trainer_ref = trainer
-
-# ---------------------------
-# 6. Run training
-# ---------------------------
-start_time = time.time()
-trainer.train()
-total_time = time.time() - start_time
-
-# ---------------------------
-# 7. Final Evaluation (on best saved model)
-# ---------------------------
-best_model_path = os.path.join(BEST_MODEL_DIR, "pytorch_model.bin")
-if os.path.exists(best_model_path):
-    try:
-        model.load_state_dict(torch.load(best_model_path, map_location=device))
-        print(f"\nLoaded best model from {best_model_path}")
-    except Exception as e:
-        print(f"\nFailed to load best model from {best_model_path}: {e}")
-
-model.eval()
-preds_z_all = []
-true_z_all = []
-pred_dists_all = []
-true_dists_all = []
-
-logits_masked_list_final = []
-labels_masked_list_final = []
-
-with torch.no_grad():
-    for batch in val_loader:
-        z = batch["z"].to(device)
-        chir = batch["chirality"].to(device)
-        fc = batch["formal_charge"].to(device)
-        edge_index = batch["edge_index"].to(device)
-        edge_attr = batch["edge_attr"].to(device)
-        batch_idx = batch["batch"].to(device)
-        labels_z = batch["labels_z"].to(device)
-        labels_dists = batch["labels_dists"].to(device)
-        labels_dists_mask = batch["labels_dists_mask"].to(device)
-
-        logits, dists_pred = model(z, chir, fc, edge_index, edge_attr, batch_idx)
-
-        mask = labels_z != -100
-        if mask.sum().item() == 0:
-            continue
-
-        logits_masked_list_final.append(logits[mask])
-        labels_masked_list_final.append(labels_z[mask])
-
-        pred_z = torch.argmax(logits[mask], dim=-1)
-        true_z = labels_z[mask]
-
-        pred_d = dists_pred[mask][labels_dists_mask[mask]]
-        true_d = labels_dists[mask][labels_dists_mask[mask]]
-
-        if pred_d.numel() > 0:
-            pred_dists_all.extend(pred_d.cpu().tolist())
-            true_dists_all.extend(true_d.cpu().tolist())
-
-        preds_z_all.extend(pred_z.cpu().tolist())
-        true_z_all.extend(true_z.cpu().tolist())
-
-accuracy = accuracy_score(true_z_all, preds_z_all) if len(true_z_all) > 0 else 0.0
-f1 = f1_score(true_z_all, preds_z_all, average="weighted") if len(true_z_all) > 0 else 0.0
-rmse = np.sqrt(mean_squared_error(true_dists_all, pred_dists_all)) if len(true_dists_all) > 0 else 0.0
-mae = mean_absolute_error(true_dists_all, pred_dists_all) if len(true_dists_all) > 0 else 0.0
-
-if len(logits_masked_list_final) > 0:
-    all_logits_masked_final = torch.cat(logits_masked_list_final, dim=0)
-    all_labels_masked_final = torch.cat(labels_masked_list_final, dim=0)
-    cw_final = getattr(model, "class_weights", None)
-    if cw_final is not None:
-        try:
-            loss_z_final = F.cross_entropy(all_logits_masked_final, all_labels_masked_final, weight=cw_final.to(device))
-        except Exception:
-            loss_z_final = F.cross_entropy(all_logits_masked_final, all_labels_masked_final)
-    else:
-        loss_z_final = F.cross_entropy(all_logits_masked_final, all_labels_masked_final)
-    try:
-        perplexity_final = float(torch.exp(loss_z_final).cpu().item())
-    except Exception:
-        perplexity_final = float(np.exp(float(loss_z_final.cpu().item())))
-else:
-    perplexity_final = float("nan")
-
-best_val_loss = callback.best_val_loss if hasattr(callback, "best_val_loss") else float("nan")
-best_epoch_num = (int(callback.best_epoch) + 1) if callback.best_epoch is not None else None
-
-print(f"\n=== Final Results (evaluated on best saved model) ===")
-print(f"Total Training Time (s): {total_time:.2f}")
-if best_epoch_num is not None:
-    print(f"Best Epoch (1-based): {best_epoch_num}")
-else:
-    print("Best Epoch: (none saved)")
-
-print(f"Best Validation Loss: {best_val_loss:.4f}")
-print(f"Validation Accuracy: {accuracy:.4f}")
-print(f"Validation F1 (weighted): {f1:.4f}")
-print(f"Validation RMSE (distances): {rmse:.4f}")
-print(f"Validation MAE  (distances): {mae:.4f}")
-print(f"Validation Perplexity (classification head): {perplexity_final:.4f}")
-
-total_params = sum(p.numel() for p in model.parameters())
-trainable_params = sum(p.numel() for p in model.parameters() if p.requires_grad)
-non_trainable_params = total_params - trainable_params
-print(f"Total Parameters: {total_params}")
-print(f"Trainable Parameters: {trainable_params}")
-print(f"Non-trainable Parameters: {non_trainable_params}")