Upload fine-tuned model - BLEU: 0.1128

.gitattributes

@@ -45,3 +45,5 @@ checkpoint-128/source.spm filter=lfs diff=lfs merge=lfs -text
 checkpoint-128/target.spm filter=lfs diff=lfs merge=lfs -text
 checkpoint-96/source.spm filter=lfs diff=lfs merge=lfs -text
 checkpoint-96/target.spm filter=lfs diff=lfs merge=lfs -text
+checkpoint-80/source.spm filter=lfs diff=lfs merge=lfs -text
+checkpoint-80/target.spm filter=lfs diff=lfs merge=lfs -text

README.md

@@ -40,13 +40,13 @@ The model was fine-tuned on a combination of:
 ## Performance
 
 ### Test Set (General Translation)
-- **BLEU:** 0.1277
-- **chrF:** 32.30
+- **BLEU:** 0.1128
+- **chrF:** 30.62
 - **Improvement over baseline:** +0.0%
 
 ### Domain Evaluation (Call Transcriptions)
 - **Domain BLEU:** 0.0000
-- **Domain chrF:** 1.86
+- **Domain chrF:** 1.68
 - **Domain COMET-QE:** 0.0000
 
 *Domain metrics are evaluated on real 10-minute call transcriptions from child helplines.*
@@ -88,7 +88,7 @@ print(translation)
 
 **Base Model:** Helsinki-NLP/opus-mt-mul-en  
 **Training Epochs:** 8  
-**Batch Size:** 550 (effective: 550)  
+**Batch Size:** 650 (effective: 650)  
 **Learning Rate:** 3e-05  
 **Optimizer:** AdamW  
 **Hardware:** NVIDIA GPU with FP16 mixed precision

baseline_results.json

@@ -1,10 +1,10 @@
 {
-  "eval_loss": 18.773775100708008,
-  "eval_model_preparation_time": 0.0027,
+  "eval_loss": 18.773773193359375,
+  "eval_model_preparation_time": 0.0022,
   "eval_bleu": 0.0317523275253792,
   "eval_chrf": 19.994361360713377,
-  "eval_runtime": 18.4416,
-  "eval_samples_per_second": 19.521,
+  "eval_runtime": 18.6114,
+  "eval_samples_per_second": 19.343,
   "eval_steps_per_second": 0.054,
   "eval_comet": 0.0341746505332392,
   "eval_comet_std": 0.1127765594779826,

checkpoint-80/config.json

@@ -0,0 +1,54 @@
+{
+  "activation_dropout": 0.0,
+  "activation_function": "swish",
+  "add_bias_logits": false,
+  "add_final_layer_norm": false,
+  "architectures": [
+    "MarianMTModel"
+  ],
+  "attention_dropout": 0.0,
+  "classif_dropout": 0.0,
+  "classifier_dropout": 0.0,
+  "d_model": 512,
+  "decoder_attention_heads": 8,
+  "decoder_ffn_dim": 2048,
+  "decoder_layerdrop": 0.0,
+  "decoder_layers": 6,
+  "decoder_start_token_id": 64171,
+  "decoder_vocab_size": 64172,
+  "dropout": 0.1,
+  "dtype": "float32",
+  "encoder_attention_heads": 8,
+  "encoder_ffn_dim": 2048,
+  "encoder_layerdrop": 0.0,
+  "encoder_layers": 6,
+  "eos_token_id": 0,
+  "extra_pos_embeddings": 64172,
+  "forced_eos_token_id": 0,
+  "id2label": {
+    "0": "LABEL_0",
+    "1": "LABEL_1",
+    "2": "LABEL_2"
+  },
+  "init_std": 0.02,
+  "is_encoder_decoder": true,
+  "label2id": {
+    "LABEL_0": 0,
+    "LABEL_1": 1,
+    "LABEL_2": 2
+  },
+  "max_length": null,
+  "max_position_embeddings": 512,
+  "model_type": "marian",
+  "normalize_before": false,
+  "normalize_embedding": false,
+  "num_beams": null,
+  "num_hidden_layers": 6,
+  "pad_token_id": 64171,
+  "scale_embedding": true,
+  "share_encoder_decoder_embeddings": true,
+  "static_position_embeddings": true,
+  "transformers_version": "4.57.1",
+  "use_cache": true,
+  "vocab_size": 64172
+}

checkpoint-80/generation_config.json

@@ -0,0 +1,17 @@
+{
+  "bad_words_ids": [
+    [
+      64171
+    ]
+  ],
+  "decoder_start_token_id": 64171,
+  "eos_token_id": [
+    0
+  ],
+  "forced_eos_token_id": 0,
+  "max_length": 512,
+  "num_beams": 6,
+  "pad_token_id": 64171,
+  "renormalize_logits": true,
+  "transformers_version": "4.57.1"
+}

checkpoint-80/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c9a591da8c0cc7e5c7dfa0f493c1d41940fc909de2e5f184478f47d7fd81f90b
+size 308263984

checkpoint-80/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:623177cbea03a4e4588e245e9e1d0f5e9f7696c2f16618301fa5eaec047250a7
+size 616171979

checkpoint-80/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c0c4f4291309307efda89e70fd2c39641c317bcec1f58f1d31cb6ab7ca07788a
+size 14645

checkpoint-80/scaler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:582c843e3d12d7dc6963fc03450e40afe0584d4ac6255f1b6648493911399102
+size 1383

checkpoint-80/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2249357a514228703ea3f694c4da545ead2df07caff7cf6b49b66e6ff739d683
+size 1465

checkpoint-80/source.spm

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c4a99ea3602b29fbf901ade8b93a45efa3d7c64eab8fc5fa812383efa327a87d
+size 706917

checkpoint-80/special_tokens_map.json

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

checkpoint-80/target.spm

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c6dce5fa58fcd7dde9e81e279b8c075bf42ee558278f73d6fb48e342029d7f19
+size 791194

checkpoint-80/tokenizer_config.json

@@ -0,0 +1,39 @@
+{
+  "added_tokens_decoder": {
+    "0": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "64171": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "</s>",
+  "extra_special_tokens": {},
+  "model_max_length": 512,
+  "pad_token": "<pad>",
+  "separate_vocabs": false,
+  "source_lang": "mul",
+  "sp_model_kwargs": {},
+  "target_lang": "eng",
+  "tokenizer_class": "MarianTokenizer",
+  "unk_token": "<unk>"
+}

checkpoint-80/trainer_state.json

@@ -0,0 +1,50 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 8.0,
+  "eval_steps": 1000,
+  "global_step": 80,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 5.0,
+      "grad_norm": 3.22662353515625,
+      "learning_rate": 1.1753405790928458e-05,
+      "loss": 6.3402,
+      "step": 50
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 80,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 8,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "EarlyStoppingCallback": {
+      "args": {
+        "early_stopping_patience": 5,
+        "early_stopping_threshold": 0.001
+      },
+      "attributes": {
+        "early_stopping_patience_counter": 0
+      }
+    },
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1138408789377024.0,
+  "train_batch_size": 650,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-80/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:955d0ef591cda55faf435097bd52609fbd96d29eca11d908443cbd38a2c8ea1e
+size 5969

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4e1c9e6b6073e8a0fc83c1a11432fbadf83a5c32033b36823091975ede4c568f
+oid sha256:c9a591da8c0cc7e5c7dfa0f493c1d41940fc909de2e5f184478f47d7fd81f90b
 size 308263984

test_results.json

@@ -1,16 +1,16 @@
 {
-  "eval_loss": 3.5031542778015137,
-  "eval_model_preparation_time": 0.0027,
-  "eval_bleu": 0.1276835459626037,
-  "eval_chrf": 32.29660565619225,
-  "eval_runtime": 4.0246,
-  "eval_samples_per_second": 89.449,
-  "eval_steps_per_second": 0.248,
+  "eval_loss": 3.5755770206451416,
+  "eval_model_preparation_time": 0.0022,
+  "eval_bleu": 0.11276124860018161,
+  "eval_chrf": 30.62377139965732,
+  "eval_runtime": 9.8882,
+  "eval_samples_per_second": 36.407,
+  "eval_steps_per_second": 0.101,
   "epoch": 8.0,
-  "eval_comet": 0.08929400351530098,
-  "eval_comet_std": 0.1904605300394193,
-  "domain_bleu": 8.850691462561237e-18,
-  "domain_chrf": 1.8573826171296468,
-  "domain_comet": 4.175902431597933e-05,
-  "domain_comet_std": 3.194240118007998e-07
+  "eval_comet": 0.07703527456770745,
+  "eval_comet_std": 0.1713326675596412,
+  "domain_bleu": 1.3107807363788414e-19,
+  "domain_chrf": 1.6767435894193803,
+  "domain_comet": 4.171086184214801e-05,
+  "domain_comet_std": 3.703383228661324e-07
 }

train_results.json

@@ -1,8 +1,8 @@
 {
-  "train_runtime": 62.295,
-  "train_samples_per_second": 784.14,
-  "train_steps_per_second": 1.541,
+  "train_runtime": 62.7594,
+  "train_samples_per_second": 778.338,
+  "train_steps_per_second": 1.275,
   "total_flos": 1138408789377024.0,
-  "train_loss": 5.05125633875529,
+  "train_loss": 5.31331958770752,
   "epoch": 8.0
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:eb0b9e675407ca1a9ee1a08d6b6b650e16ca2de15fd3b2c37137b42aedec39fa
+oid sha256:955d0ef591cda55faf435097bd52609fbd96d29eca11d908443cbd38a2c8ea1e
 size 5969

translation_config.json

@@ -30,7 +30,7 @@
   },
   "training_config": {
     "learning_rate": 3e-05,
-    "batch_size": 550,
+    "batch_size": 650,
     "gradient_accumulation_steps": 1,
     "num_epochs": 8,
     "max_length": 300,