Updated embeddings.py

config.json

@@ -1,23 +1,14 @@
 {
-  "add_cross_attention": false,
   "architectures": [
     "DualBertForMaskedLM"
   ],
   "attention_probs_dropout_prob": 0.1,
-  "auto_map": {
-    "AutoConfig": "config.DualBertConfig",
-    "AutoModelForMaskedLM": "model.DualBertForMaskedLM"
-  },
-  "bos_token_id": null,
   "classifier_dropout": null,
-  "dtype": "float32",
-  "eos_token_id": null,
   "hidden_act": "gelu",
   "hidden_dropout_prob": 0.1,
   "hidden_size": 512,
   "initializer_range": 0.02,
   "intermediate_size": 2048,
-  "is_decoder": false,
   "layer_norm_eps": 1e-12,
   "max_position_embeddings": 256,
   "model_type": "dual_bert",
@@ -25,12 +16,16 @@
   "num_hidden_layers": 6,
   "pad_token_id": 0,
   "position_embedding_type": "absolute",
-  "tie_word_embeddings": true,
-  "transformers_version": "5.0.0",
+  "torch_dtype": "float32",
+  "transformers_version": "4.48.0",
   "type_vocab_size": 2,
   "use_cache": true,
   "vocab_char_size": 156,
   "vocab_size": 156,
   "vocab_word_size": 50000,
-  "word_char_emb_dim": 192
-}
+  "word_char_emb_dim": 192,
+  "auto_map": {
+    "AutoConfig": "config.DualBertConfig",
+    "AutoModelForMaskedLM": "model.DualBertForMaskedLM"
+  }
+}

embeddings.py

@@ -21,7 +21,7 @@ class DualEmbeddings(nn.Module):
         self.dropout = nn.Dropout(config.hidden_dropout_prob)
 
         self.register_buffer(
-            "position_ids", torch.arange(config.max_position_embeddings).unsqueeze(0), persistent=False
+            "position_ids", torch.arange(config.max_position_embeddings).unsqueeze(0), persistent=True
         )
 
     def forward(self, input_ids, word_ids):

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dbc4d3b130b1eaf4dc0facc41db0e2eb6945741f82154084f146706be7c6d040
-size 115901848
+oid sha256:804d554654bf3ec17609d0794f24ac71ca2c311e4dd2be9253672e958f12e1c6
+size 115899720