stage best fine adapt wce model

config.json

@@ -0,0 +1,46 @@
+{
+  "_name_or_path": "Helsinki-NLP/opus-mt-tc-big-en-fr",
+  "activation_dropout": 0.0,
+  "activation_function": "relu",
+  "architectures": [
+    "MarianMTModel"
+  ],
+  "attention_dropout": 0.0,
+  "bad_words_ids": [
+    [
+      53016
+    ]
+  ],
+  "bos_token_id": 0,
+  "classifier_dropout": 0.0,
+  "d_model": 1024,
+  "decoder_attention_heads": 16,
+  "decoder_ffn_dim": 4096,
+  "decoder_layerdrop": 0.0,
+  "decoder_layers": 6,
+  "decoder_start_token_id": 53016,
+  "decoder_vocab_size": 53017,
+  "dropout": 0.1,
+  "encoder_attention_heads": 16,
+  "encoder_ffn_dim": 4096,
+  "encoder_layerdrop": 0.0,
+  "encoder_layers": 6,
+  "eos_token_id": 43311,
+  "forced_eos_token_id": 43311,
+  "init_std": 0.02,
+  "is_encoder_decoder": true,
+  "max_length": 512,
+  "max_position_embeddings": 1024,
+  "model_type": "marian",
+  "normalize_embedding": false,
+  "num_beams": 4,
+  "num_hidden_layers": 6,
+  "pad_token_id": 53016,
+  "scale_embedding": true,
+  "share_encoder_decoder_embeddings": true,
+  "static_position_embeddings": true,
+  "torch_dtype": "float32",
+  "transformers_version": "4.28.0",
+  "use_cache": true,
+  "vocab_size": 53017
+}

generation_config.json

@@ -0,0 +1,15 @@
+{
+  "bad_words_ids": [
+    [
+      53016
+    ]
+  ],
+  "bos_token_id": 0,
+  "decoder_start_token_id": 53016,
+  "eos_token_id": 43311,
+  "forced_eos_token_id": 43311,
+  "max_length": 512,
+  "num_beams": 4,
+  "pad_token_id": 53016,
+  "transformers_version": "4.28.0"
+}

optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9bad2786a29505981210f9db60b02ae4ad60b27b27929e497a0fc079e6aecd03
+size 1845323269

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5a78f6cd743c0c49e5829acd1fdf09f3c4cad3e6f4ecd386b4676b1252d623ee
+size 922885701

rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f6d9584ca19c759f55fb6696ae63326aafdc3654fc0110f946604c56d2d2cbf8
+size 14575

scaler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:695fca40e3d626312836daafa1a572cd18b4c73565d62502f1c29f5688163d29
+size 557

scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b2d280d247d94ac98ee856efffc708e91babbdb3306480c0c30168fc32146e59
+size 627

special_tokens_map.json

@@ -0,0 +1,5 @@
+{
+  "eos_token": "</s>",
+  "pad_token": "<pad>",
+  "unk_token": "<unk>"
+}

tokenizer_config.json

@@ -0,0 +1,12 @@
+{
+  "clean_up_tokenization_spaces": true,
+  "eos_token": "</s>",
+  "model_max_length": 512,
+  "pad_token": "<pad>",
+  "separate_vocabs": false,
+  "source_lang": "en",
+  "sp_model_kwargs": {},
+  "target_lang": "fr",
+  "tokenizer_class": "MarianTokenizer",
+  "unk_token": "<unk>"
+}

trainer_state.json

@@ -0,0 +1,176 @@
+{
+  "best_metric": 43.6202,
+  "best_model_checkpoint": "opus_big_adapt_wce_gloss_train-sampled_precision_2_ubweight_1.5/checkpoint-80000",
+  "epoch": 1.9716574245224892,
+  "global_step": 80000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9753635243376465e-05,
+      "loss": 0.1779,
+      "step": 8000
+    },
+    {
+      "epoch": 0.2,
+      "eval_bleu": 42.2604,
+      "eval_gen_len": 33.6012,
+      "eval_loss": 0.0981723740696907,
+      "eval_runtime": 83.525,
+      "eval_samples_per_second": 12.487,
+      "eval_steps_per_second": 0.79,
+      "step": 8000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.950723967960567e-05,
+      "loss": 0.1471,
+      "step": 16000
+    },
+    {
+      "epoch": 0.39,
+      "eval_bleu": 42.5679,
+      "eval_gen_len": 34.2627,
+      "eval_loss": 0.09737617522478104,
+      "eval_runtime": 183.3828,
+      "eval_samples_per_second": 5.688,
+      "eval_steps_per_second": 0.36,
+      "step": 16000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9260874922982132e-05,
+      "loss": 0.143,
+      "step": 24000
+    },
+    {
+      "epoch": 0.59,
+      "eval_bleu": 42.4537,
+      "eval_gen_len": 33.7315,
+      "eval_loss": 0.09751108288764954,
+      "eval_runtime": 82.4552,
+      "eval_samples_per_second": 12.649,
+      "eval_steps_per_second": 0.8,
+      "step": 24000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.901447935921134e-05,
+      "loss": 0.1392,
+      "step": 32000
+    },
+    {
+      "epoch": 0.79,
+      "eval_bleu": 42.7064,
+      "eval_gen_len": 33.7919,
+      "eval_loss": 0.096822090446949,
+      "eval_runtime": 102.3142,
+      "eval_samples_per_second": 10.194,
+      "eval_steps_per_second": 0.645,
+      "step": 32000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8768114602587803e-05,
+      "loss": 0.1377,
+      "step": 40000
+    },
+    {
+      "epoch": 0.99,
+      "eval_bleu": 42.9066,
+      "eval_gen_len": 34.1697,
+      "eval_loss": 0.09617254137992859,
+      "eval_runtime": 93.7017,
+      "eval_samples_per_second": 11.131,
+      "eval_steps_per_second": 0.704,
+      "step": 40000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8521749845964266e-05,
+      "loss": 0.1226,
+      "step": 48000
+    },
+    {
+      "epoch": 1.18,
+      "eval_bleu": 42.551,
+      "eval_gen_len": 34.1774,
+      "eval_loss": 0.09693969786167145,
+      "eval_runtime": 93.2047,
+      "eval_samples_per_second": 11.19,
+      "eval_steps_per_second": 0.708,
+      "step": 48000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8275415896487988e-05,
+      "loss": 0.1229,
+      "step": 56000
+    },
+    {
+      "epoch": 1.38,
+      "eval_bleu": 42.3153,
+      "eval_gen_len": 33.9128,
+      "eval_loss": 0.09742604196071625,
+      "eval_runtime": 95.1569,
+      "eval_samples_per_second": 10.961,
+      "eval_steps_per_second": 0.694,
+      "step": 56000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.802905113986445e-05,
+      "loss": 0.1235,
+      "step": 64000
+    },
+    {
+      "epoch": 1.58,
+      "eval_bleu": 42.8615,
+      "eval_gen_len": 34.024,
+      "eval_loss": 0.09685970097780228,
+      "eval_runtime": 97.6575,
+      "eval_samples_per_second": 10.68,
+      "eval_steps_per_second": 0.676,
+      "step": 64000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.778268638324091e-05,
+      "loss": 0.123,
+      "step": 72000
+    },
+    {
+      "epoch": 1.77,
+      "eval_bleu": 43.0325,
+      "eval_gen_len": 34.3768,
+      "eval_loss": 0.09707798808813095,
+      "eval_runtime": 111.6142,
+      "eval_samples_per_second": 9.345,
+      "eval_steps_per_second": 0.591,
+      "step": 72000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7536352433764633e-05,
+      "loss": 0.1227,
+      "step": 80000
+    },
+    {
+      "epoch": 1.97,
+      "eval_bleu": 43.6202,
+      "eval_gen_len": 33.8667,
+      "eval_loss": 0.09590897709131241,
+      "eval_runtime": 87.3368,
+      "eval_samples_per_second": 11.942,
+      "eval_steps_per_second": 0.756,
+      "step": 80000
+    }
+  ],
+  "max_steps": 649200,
+  "num_train_epochs": 16,
+  "total_flos": 2.5844916958632346e+17,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4e6f256a7f65903529b828f2530f16b3c838ce03efa8b0870fb213318cb6fbfe
+size 3835