Training in progress, epoch 1

config.json

@@ -0,0 +1,48 @@
+{
+  "_name_or_path": "facebook/wmt19-de-en",
+  "activation_dropout": 0.0,
+  "activation_function": "relu",
+  "architectures": [
+    "FSMTForConditionalGeneration"
+  ],
+  "attention_dropout": 0.1,
+  "bos_token_id": 0,
+  "d_model": 1024,
+  "decoder": {
+    "bos_token_id": 2,
+    "model_type": "fsmt_decoder",
+    "vocab_size": 42024
+  },
+  "decoder_attention_heads": 16,
+  "decoder_ffn_dim": 4096,
+  "decoder_layerdrop": 0,
+  "decoder_layers": 6,
+  "decoder_start_token_id": 2,
+  "dropout": 0.2,
+  "encoder_attention_heads": 16,
+  "encoder_ffn_dim": 8192,
+  "encoder_layerdrop": 0,
+  "encoder_layers": 6,
+  "eos_token_id": 2,
+  "forced_eos_token_id": 2,
+  "init_std": 0.02,
+  "is_encoder_decoder": true,
+  "langs": [
+    "de",
+    "en"
+  ],
+  "length_penalty": 1.1,
+  "max_length": 200,
+  "max_position_embeddings": 1024,
+  "model_type": "fsmt",
+  "num_beams": 5,
+  "num_hidden_layers": 6,
+  "pad_token_id": 1,
+  "scale_embedding": true,
+  "src_vocab_size": 42024,
+  "tgt_vocab_size": 42024,
+  "tie_word_embeddings": true,
+  "torch_dtype": "float32",
+  "transformers_version": "4.37.2",
+  "use_cache": true
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f759d278376d01632df4cae53ee1536ef2ef90859752f1cc0a891b939ea52c70
+size 1087419504

runs/Mar18_19-30-00_gpusvr021320/events.out.tfevents.1710757951.gpusvr021320.753078.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e6d23736c4d8662101f7d7f36fe3ef724429e6022b9492fdb72c5ba2ead487da
+size 342

runs/Mar18_19-44-31_gpusvr021320/events.out.tfevents.1710758843.gpusvr021320.756466.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9ef96e9ec4c2a3a87f9f699b9d26d6d58250a965920d6e4b41dcecdefd54c8de
+size 342

runs/Mar18_20-08-17_gpusvr021320/events.out.tfevents.1710760258.gpusvr021320.765054.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:39ffcff5b51f5dfc9f7da8eb14998513bb32f4c5534a695f4644e4cda95d61db
+size 342

runs/Mar18_20-11-50_gpusvr021320/events.out.tfevents.1710760369.gpusvr021320.766285.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4308f4ca63e5a2637a9bce29b612c43a783e07c7fb0d5cb9e6a960a33d7e0385
+size 342

runs/Mar18_20-13-16_gpusvr021320/events.out.tfevents.1710760539.gpusvr021320.767049.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cd34cfbf253d6a0f153306ffc552ccc8d28d65d1e89bba6c4193c84b8a0559e7
+size 342

runs/Mar18_20-30-53_gpusvr021320/events.out.tfevents.1710761638.gpusvr021320.773013.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:347d2ae304796e66304c0be0630158025e43fb889359a1a1d246a4cfc009081e
+size 342

runs/Mar18_20-40-40_gpusvr021320/events.out.tfevents.1710762051.gpusvr021320.773013.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:389a085d25fd48844cf8327cc84d1cd7aaf116dffe2866bd3aa5716335693053
+size 4184

runs/Mar18_20-42-29_gpusvr021320/events.out.tfevents.1710762152.gpusvr021320.783960.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9bc21b32e632e9709e1031b2dcc81a64d5b4fa0a375dcd5f673dfb9ea05c1528
+size 4184

runs/Mar18_20-45-16_gpusvr021320/events.out.tfevents.1710762319.gpusvr021320.801134.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:09731ec86686ef03cd5d3bdbe30491315819cc8f2dcbc02271e01f69e4b77c02
+size 6732

special_tokens_map.json

@@ -0,0 +1,6 @@
+{
+  "bos_token": "<s>",
+  "pad_token": "<pad>",
+  "sep_token": "</s>",
+  "unk_token": "<unk>"
+}

tokenizer_config.json

@@ -0,0 +1,48 @@
+{
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": true,
+  "do_lower_case": false,
+  "langs": [
+    "de",
+    "en"
+  ],
+  "model_max_length": 1024,
+  "pad_token": "<pad>",
+  "sep_token": "</s>",
+  "tokenizer_class": "FSMTTokenizer",
+  "unk_token": "<unk>"
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c1de7d234c264ccb30ba3dc1e4be1bdfe33901ec174695a8a61d0b2798bbc7ff
+size 4475