stage best simple lsp 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:fe39336a53d22b211b3a4841695d4cce6332444a3686f8389e10d6d9dc99009d
+size 1845323269

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1ea1e94da9ffb4d90daa475eaaa5ce642b286bbfdf77f14a08fcaf28b89470a6
+size 922885701

rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d740139fa350b619d6aff082c4218907935095a30da095a0f135e90771865fe1
+size 14511

scaler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5a578d70e3908236a21ccec16fb40db33675969b64cf45918f62cdd714c0bd2d
+size 557

scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d841877dae97f3484921bd550ae3afec1339fe3ea6e45d722c6eee90ccd95ea9
+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": 43.3065,
+  "best_model_checkpoint": "opus_big_lsp_adapt_wce_bands_4_ubweight_2.0/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": 0.1822,
+      "step": 8000
+    },
+    {
+      "epoch": 0.2,
+      "eval_bleu": 41.6724,
+      "eval_gen_len": 33.535,
+      "eval_loss": 0.09890133887529373,
+      "eval_runtime": 90.9254,
+      "eval_samples_per_second": 11.471,
+      "eval_steps_per_second": 0.726,
+      "step": 8000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.950723967960567e-05,
+      "loss": 0.1511,
+      "step": 16000
+    },
+    {
+      "epoch": 0.39,
+      "eval_bleu": 42.4546,
+      "eval_gen_len": 33.838,
+      "eval_loss": 0.09774070978164673,
+      "eval_runtime": 188.0787,
+      "eval_samples_per_second": 5.546,
+      "eval_steps_per_second": 0.351,
+      "step": 16000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9260874922982132e-05,
+      "loss": 0.1468,
+      "step": 24000
+    },
+    {
+      "epoch": 0.59,
+      "eval_bleu": 42.5654,
+      "eval_gen_len": 34.9655,
+      "eval_loss": 0.09777849912643433,
+      "eval_runtime": 89.4604,
+      "eval_samples_per_second": 11.659,
+      "eval_steps_per_second": 0.738,
+      "step": 24000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.901447935921134e-05,
+      "loss": 0.1428,
+      "step": 32000
+    },
+    {
+      "epoch": 0.79,
+      "eval_bleu": 42.2326,
+      "eval_gen_len": 34.0834,
+      "eval_loss": 0.0976356640458107,
+      "eval_runtime": 96.433,
+      "eval_samples_per_second": 10.816,
+      "eval_steps_per_second": 0.684,
+      "step": 32000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.876814540973506e-05,
+      "loss": 0.1413,
+      "step": 40000
+    },
+    {
+      "epoch": 0.99,
+      "eval_bleu": 42.0941,
+      "eval_gen_len": 34.8849,
+      "eval_loss": 0.09630727022886276,
+      "eval_runtime": 99.7469,
+      "eval_samples_per_second": 10.456,
+      "eval_steps_per_second": 0.662,
+      "step": 40000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.852178065311152e-05,
+      "loss": 0.1258,
+      "step": 48000
+    },
+    {
+      "epoch": 1.18,
+      "eval_bleu": 42.7386,
+      "eval_gen_len": 33.6261,
+      "eval_loss": 0.09742352366447449,
+      "eval_runtime": 76.9301,
+      "eval_samples_per_second": 13.558,
+      "eval_steps_per_second": 0.858,
+      "step": 48000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8275415896487988e-05,
+      "loss": 0.1261,
+      "step": 56000
+    },
+    {
+      "epoch": 1.38,
+      "eval_bleu": 42.7075,
+      "eval_gen_len": 34.2646,
+      "eval_loss": 0.09748394042253494,
+      "eval_runtime": 96.146,
+      "eval_samples_per_second": 10.848,
+      "eval_steps_per_second": 0.686,
+      "step": 56000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.802905113986445e-05,
+      "loss": 0.1266,
+      "step": 64000
+    },
+    {
+      "epoch": 1.58,
+      "eval_bleu": 41.5564,
+      "eval_gen_len": 35.6529,
+      "eval_loss": 0.0969226285815239,
+      "eval_runtime": 104.3825,
+      "eval_samples_per_second": 9.992,
+      "eval_steps_per_second": 0.632,
+      "step": 64000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.778268638324091e-05,
+      "loss": 0.1261,
+      "step": 72000
+    },
+    {
+      "epoch": 1.77,
+      "eval_bleu": 42.379,
+      "eval_gen_len": 34.3078,
+      "eval_loss": 0.09717730432748795,
+      "eval_runtime": 151.7683,
+      "eval_samples_per_second": 6.872,
+      "eval_steps_per_second": 0.435,
+      "step": 72000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7536321626617377e-05,
+      "loss": 0.1258,
+      "step": 80000
+    },
+    {
+      "epoch": 1.97,
+      "eval_bleu": 43.3065,
+      "eval_gen_len": 33.5158,
+      "eval_loss": 0.096554696559906,
+      "eval_runtime": 85.4527,
+      "eval_samples_per_second": 12.206,
+      "eval_steps_per_second": 0.772,
+      "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 sha256:24528e49ec48597ebc8cc2f9e9a72af7d42aba681888ff615eea651807a5315a
+size 3771