From ce81f268450c572798df9ddb14f367c116326d35 Mon Sep 17 00:00:00 2001
From: Colin <>
Date: Tue, 12 Aug 2025 11:08:16 +0800
Subject: [PATCH] Fix dataset cache check with vocab size.

---
 wit/meaning/dataset.py         | 10 ++++------
 wit/meaning/meaning_dataset.py |  2 +-
 2 files changed, 5 insertions(+), 7 deletions(-)

diff --git a/wit/meaning/dataset.py b/wit/meaning/dataset.py
index b196a81..55ce971 100644
--- a/wit/meaning/dataset.py
+++ b/wit/meaning/dataset.py
@@ -37,9 +37,8 @@ def InitDataset(config):
         seed = c.seed
 
         path = "./data/"
-        conf_name = (
-            f"_s{start}_e{end}_s{size}_ms{c.min_subitem}_maxs{c.max_subitem}_stride{c.stride}_tree{c.with_tree}.pt"
-        )
+        conf_name = f"_s{start}_e{end}_s{size}_ms{c.min_subitem}_maxs{c.max_subitem}_seed{seed}"
+        conf_name = conf_name + f"_vocab{vocab}_stride{c.stride}_tree{c.with_tree}.pt"
         trainfile = path + f"MeaningDataset_train" + conf_name
         valfile = path + f"MeaningDataset_val" + conf_name
         if not os.path.exists(path):
@@ -101,9 +100,8 @@ def InitValDataset(config):
         seed = c.seed
 
         path = "./data/"
-        conf_name = (
-            f"_s{start}_e{end}_s{size}_ms{c.min_subitem}_maxs{c.max_subitem}_stride{c.stride}_tree{c.with_tree}.pt"
-        )
+        conf_name = f"_s{start}_e{end}_s{size}_ms{c.min_subitem}_maxs{c.max_subitem}_seed{seed}"
+        conf_name = conf_name + f"_vocab{vocab}_stride{c.stride}_tree{c.with_tree}.pt"
         valfile = path + f"MeaningDataset_val" + conf_name
         if not os.path.exists(path):
             os.mkdir(path)
diff --git a/wit/meaning/meaning_dataset.py b/wit/meaning/meaning_dataset.py
index 8c6f501..2044cdd 100644
--- a/wit/meaning/meaning_dataset.py
+++ b/wit/meaning/meaning_dataset.py
@@ -50,7 +50,7 @@ class MeaningMap:
         path = "./data/"
         file = "structured_language_" + str(size) + "_" + str(vocab_size)
         file += "_" + str(max_subitem) + "_" + str(min_subitem)
-        file += "_" + str(stride) + "_" + str(with_tree)
+        file += "_" + str(stride) + "_" + str(with_tree) + "_" + str(seed)
         file_prop = path + file + "_prop.npy"
         file_data = path + file + "_data.npy"
         file_level = path + file + "_level.npy"