Training in progress, step 500

config.json

@@ -1,29 +1,27 @@
 {
-  "_name_or_path": "vinai/phobert-base",
+  "_name_or_path": "keepitreal/mini-phobert-v3.1",
   "architectures": [
-    "RobertaForMultipleChoice"
+    "MathModel"
   ],
   "attention_probs_dropout_prob": 0.1,
   "bos_token_id": 0,
   "classifier_dropout": null,
   "eos_token_id": 2,
-  "gradient_checkpointing": false,
   "hidden_act": "gelu",
   "hidden_dropout_prob": 0.1,
-  "hidden_size": 768,
+  "hidden_size": 256,
   "initializer_range": 0.02,
-  "intermediate_size": 3072,
-  "layer_norm_eps": 1e-05,
-  "max_position_embeddings": 258,
+  "intermediate_size": 512,
+  "layer_norm_eps": 1e-12,
+  "max_position_embeddings": 512,
   "model_type": "roberta",
-  "num_attention_heads": 12,
-  "num_hidden_layers": 12,
+  "num_attention_heads": 4,
+  "num_hidden_layers": 6,
   "pad_token_id": 1,
   "position_embedding_type": "absolute",
-  "tokenizer_class": "PhobertTokenizer",
   "torch_dtype": "float32",
   "transformers_version": "4.29.1",
-  "type_vocab_size": 1,
+  "type_vocab_size": 2,
   "use_cache": true,
   "vocab_size": 64001
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:11375c58ecfbf7112546ff880623d7b7eb25c8c3554f70aa8299e00c746ed9de
-size 540065138
+oid sha256:c9068cb97547cfd616a3090c0d78428cc2541145ced7abaebed4ef8d090853ce
+size 79019445

runs/Nov24_18-06-27_Na.local/1700824164.622946/events.out.tfevents.1700824164.Na.local.91827.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ffa1c124dac3a8d21e48aac61fc8c16b50ee95f4880939661451aa6e49a7d121
+size 5846

runs/Nov24_18-06-27_Na.local/events.out.tfevents.1700824164.Na.local.91827.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bcf3a7d31b83168c5d2ff27536377cf46f9079a6cfb75d436ac8609216a17b72
+size 4106

runs/Nov24_18-35-11_Na.local/1700825714.31274/events.out.tfevents.1700825714.Na.local.2790.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:11bda1f9a7a152e6bcd8cf97894eac7af96be7364a923256418a7e06c539827c
+size 5846

runs/Nov24_18-35-11_Na.local/events.out.tfevents.1700825714.Na.local.2790.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:35afc801d7e138b429a5d5ce775b0e156343f2e133a75da43632a0d48eaf853b
+size 6557

runs/Nov24_18-48-01_Na.local/1700826484.87286/events.out.tfevents.1700826484.Na.local.7714.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b89f4d525b12d9147b7cd91f0f1553bce52115cff13cc74f8d04e09358158598
+size 5846

runs/Nov24_18-48-01_Na.local/events.out.tfevents.1700826484.Na.local.7714.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:89f7a5478ef0984f70726520d93850eca5bd12bc2da8c7641f84e281482f09f7
+size 7360

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:713c3e5307fa1e353b73af2ac257c1cbd01f09efb5a481b6b0bef349b27a6443
-size 4344
+oid sha256:d9d242cfe84cccea01341578945d9774a5e1edc65dc93075b66b5a17828236db
+size 3899