Training in progress, epoch 1

config.json

@@ -1,61 +1,46 @@
 {
-  "_name_or_path": "Helsinki-NLP/opus-mt-tr-en",
-  "_num_labels": 3,
+  "_name_or_path": "Helsinki-NLP/opus-mt-tc-big-en-tr",
   "activation_dropout": 0.0,
-  "activation_function": "swish",
-  "add_bias_logits": false,
-  "add_final_layer_norm": false,
+  "activation_function": "relu",
   "architectures": [
     "MarianMTModel"
   ],
   "attention_dropout": 0.0,
   "bad_words_ids": [
     [
-      62388
+      57059
     ]
   ],
   "bos_token_id": 0,
-  "classif_dropout": 0.0,
   "classifier_dropout": 0.0,
-  "d_model": 512,
-  "decoder_attention_heads": 8,
-  "decoder_ffn_dim": 2048,
+  "d_model": 1024,
+  "decoder_attention_heads": 16,
+  "decoder_ffn_dim": 4096,
   "decoder_layerdrop": 0.0,
   "decoder_layers": 6,
-  "decoder_start_token_id": 62388,
-  "decoder_vocab_size": 62389,
+  "decoder_start_token_id": 57059,
+  "decoder_vocab_size": 57060,
   "dropout": 0.1,
-  "encoder_attention_heads": 8,
-  "encoder_ffn_dim": 2048,
+  "encoder_attention_heads": 16,
+  "encoder_ffn_dim": 4096,
   "encoder_layerdrop": 0.0,
   "encoder_layers": 6,
-  "eos_token_id": 0,
-  "forced_eos_token_id": 0,
-  "id2label": {
-    "0": "LABEL_0",
-    "1": "LABEL_1",
-    "2": "LABEL_2"
-  },
+  "eos_token_id": 43741,
+  "forced_eos_token_id": 43741,
   "init_std": 0.02,
   "is_encoder_decoder": true,
-  "label2id": {
-    "LABEL_0": 0,
-    "LABEL_1": 1,
-    "LABEL_2": 2
-  },
   "max_length": 512,
-  "max_position_embeddings": 512,
+  "max_position_embeddings": 1024,
   "model_type": "marian",
-  "normalize_before": false,
   "normalize_embedding": false,
-  "num_beams": 6,
+  "num_beams": 4,
   "num_hidden_layers": 6,
-  "pad_token_id": 62388,
+  "pad_token_id": 57059,
   "scale_embedding": true,
   "share_encoder_decoder_embeddings": true,
   "static_position_embeddings": true,
   "torch_dtype": "float32",
   "transformers_version": "4.38.2",
   "use_cache": true,
-  "vocab_size": 62389
+  "vocab_size": 57060
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:609f709de5778280f06a01219a06fe3ad5496c5b28248f7b6004e72fb748940c
-size 304605268
+oid sha256:6cd7666db5a6307b0916cf4543db59e8b2c1d35df5f879f50a5d9d3a416a9e65
+size 939405120

runs/Apr05_03-00-32_efae8f6580e8/events.out.tfevents.1712286656.efae8f6580e8.309.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:47ee4e08a1fbf5ac903b2a6eee3b72599b4aed4890b70dfdc40e0c81bbf64007
+size 4438

runs/Apr05_03-16-00_efae8f6580e8/events.out.tfevents.1712287485.efae8f6580e8.4170.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:774a29ba5a3c1b54a8c0702bdbf2ef188dcbe0c9e2f658a5cb9b8c30c1e4b059
+size 4438

runs/Apr05_03-29-18_efae8f6580e8/events.out.tfevents.1712288247.efae8f6580e8.7524.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5bdd35641fe4b3825f9b9968c127badd643434eb6a0b21f3dab98371b378ffde
+size 4438

runs/Apr05_03-39-23_efae8f6580e8/events.out.tfevents.1712288622.efae8f6580e8.10632.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:14c0982b0093f651b0b76086574544cb003edcdfb789c349e154955f9d48d9ee
+size 4438

runs/Apr05_03-48-42_efae8f6580e8/events.out.tfevents.1712288989.efae8f6580e8.12458.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:507ef4c192fa995932a2a94fbe5d46a516ab4869775b29f74535925d7621da19
+size 5450

runs/Apr05_04-17-59_efae8f6580e8/events.out.tfevents.1712291027.efae8f6580e8.14060.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dda74131c04862091c85fd23f684c0eb7ef30c2e98715446f6b172a803722bbe
+size 5450

runs/Apr05_04-47-05_efae8f6580e8/events.out.tfevents.1712292752.efae8f6580e8.26802.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c34700fcc8acd1a093e79781c575a09900f9a3f4b386901395e942ac8042628
+size 5450

tokenizer_config.json

@@ -1,6 +1,6 @@
 {
   "added_tokens_decoder": {
-    "0": {
+    "43741": {
       "content": "</s>",
       "lstrip": false,
       "normalized": false,
@@ -8,7 +8,7 @@
       "single_word": false,
       "special": true
     },
-    "1": {
+    "52508": {
       "content": "<unk>",
       "lstrip": false,
       "normalized": false,
@@ -16,7 +16,7 @@
       "single_word": false,
       "special": true
     },
-    "62388": {
+    "57059": {
       "content": "<pad>",
       "lstrip": false,
       "normalized": false,
@@ -31,9 +31,9 @@
   "pad_token": "<pad>",
   "return_tensors": "pt",
   "separate_vocabs": false,
-  "source_lang": "tr",
+  "source_lang": "en",
   "sp_model_kwargs": {},
-  "target_lang": "en",
+  "target_lang": "tr",
   "tokenizer_class": "MarianTokenizer",
   "unk_token": "<unk>"
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1bbaf470138f6721e5b4a47155a4aaf6095c6775c23a1b2836f56198d576b30d
+oid sha256:1385c197ac163a13c8d22ccb6b6ff70386899f07d65b923cebfee40ab756aa31
 size 5048