SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

🚨🚨🚨Deprecate evaluation_strategy to eval_strategy🚨🚨🚨 (#30190)

Browse Source 
* Alias

* Note alias

* Tests and src

* Rest

* Clean

* Change typing?

* Fix tests

* Deprecation versions
This commit is contained in:
Zach Mueller
2024-04-18 12:49:43 -04:00
committed by GitHub
parent c86d020ead
commit 60d5f8f9f0
116 changed files with 214 additions and 203 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
docs/source/de/training.md
View File

@@ -128,12 +128,12 @@ Rufen Sie [`~evaluate.compute`] auf `metric` auf, um die Genauigkeit Ihrer Vorhe
... return metric.compute(predictions=predictions, references=labels)
```
Wenn Sie Ihre Bewertungsmetriken während der Feinabstimmung überwachen möchten, geben Sie den Parameter `evaluation_strategy` in Ihren Trainingsargumenten an, um die Bewertungsmetrik am Ende jeder Epoche zu ermitteln:
Wenn Sie Ihre Bewertungsmetriken während der Feinabstimmung überwachen möchten, geben Sie den Parameter `eval_strategy` in Ihren Trainingsargumenten an, um die Bewertungsmetrik am Ende jeder Epoche zu ermitteln:
```py
>>> from transformers import TrainingArguments, Trainer
>>> training_args = TrainingArguments(output_dir="test_trainer", evaluation_strategy="epoch")
>>> training_args = TrainingArguments(output_dir="test_trainer", eval_strategy="epoch")
```
### Trainer

2
docs/source/en/model_memory_anatomy.md
View File

@@ -145,7 +145,7 @@ arguments:
```py
default_args = {
"output_dir": "tmp",
"evaluation_strategy": "steps",
"eval_strategy": "steps",
"num_train_epochs": 1,
"log_level": "error",
"report_to": "none",

2
docs/source/en/tasks/asr.md
View File

@@ -270,7 +270,7 @@ At this point, only three steps remain:
... gradient_checkpointing=True,
... fp16=True,
... group_by_length=True,
... evaluation_strategy="steps",
... eval_strategy="steps",
... per_device_eval_batch_size=8,
... save_steps=1000,
... eval_steps=1000,

2
docs/source/en/tasks/audio_classification.md
View File

@@ -221,7 +221,7 @@ At this point, only three steps remain:
```py
>>> training_args = TrainingArguments(
... output_dir="my_awesome_mind_model",
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... save_strategy="epoch",
... learning_rate=3e-5,
... per_device_train_batch_size=32,

2
docs/source/en/tasks/document_question_answering.md
View File

@@ -399,7 +399,7 @@ In this case the `output_dir` will also be the name of the repo where your model
... num_train_epochs=20,
... save_steps=200,
... logging_steps=50,
... evaluation_strategy="steps",
... eval_strategy="steps",
... learning_rate=5e-5,
... save_total_limit=2,
... remove_unused_columns=False,

2
docs/source/en/tasks/image_captioning.md
View File

@@ -196,7 +196,7 @@ training_args = TrainingArguments(
per_device_eval_batch_size=32,
gradient_accumulation_steps=2,
save_total_limit=3,
evaluation_strategy="steps",
eval_strategy="steps",
eval_steps=50,
save_strategy="steps",
save_steps=50,

2
docs/source/en/tasks/image_classification.md
View File

@@ -302,7 +302,7 @@ At this point, only three steps remain:
>>> training_args = TrainingArguments(
... output_dir="my_awesome_food_model",
... remove_unused_columns=False,
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... save_strategy="epoch",
... learning_rate=5e-5,
... per_device_train_batch_size=16,

2
docs/source/en/tasks/knowledge_distillation_for_image_classification.md
View File

@@ -112,7 +112,7 @@ training_args = TrainingArguments(
fp16=True,
logging_dir=f"{repo_name}/logs",
logging_strategy="epoch",
evaluation_strategy="epoch",
eval_strategy="epoch",
save_strategy="epoch",
load_best_model_at_end=True,
metric_for_best_model="accuracy",

2
docs/source/en/tasks/language_modeling.md
View File

@@ -249,7 +249,7 @@ At this point, only three steps remain:
```py
>>> training_args = TrainingArguments(
... output_dir="my_awesome_eli5_clm-model",
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... learning_rate=2e-5,
... weight_decay=0.01,
... push_to_hub=True,

2
docs/source/en/tasks/masked_language_modeling.md
View File

@@ -238,7 +238,7 @@ At this point, only three steps remain:
```py
>>> training_args = TrainingArguments(
... output_dir="my_awesome_eli5_mlm_model",
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... learning_rate=2e-5,
... num_train_epochs=3,
... weight_decay=0.01,

2
docs/source/en/tasks/multiple_choice.md
View File

@@ -265,7 +265,7 @@ At this point, only three steps remain:
```py
>>> training_args = TrainingArguments(
... output_dir="my_awesome_swag_model",
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... save_strategy="epoch",
... load_best_model_at_end=True,
... learning_rate=5e-5,

2
docs/source/en/tasks/question_answering.md
View File

@@ -218,7 +218,7 @@ At this point, only three steps remain:
```py
>>> training_args = TrainingArguments(
... output_dir="my_awesome_qa_model",
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... learning_rate=2e-5,
... per_device_train_batch_size=16,
... per_device_eval_batch_size=16,

2
docs/source/en/tasks/semantic_segmentation.md
View File

@@ -535,7 +535,7 @@ At this point, only three steps remain:
... per_device_train_batch_size=2,
... per_device_eval_batch_size=2,
... save_total_limit=3,
... evaluation_strategy="steps",
... eval_strategy="steps",
... save_strategy="steps",
... save_steps=20,
... eval_steps=20,

2
docs/source/en/tasks/sequence_classification.md
View File

@@ -187,7 +187,7 @@ At this point, only three steps remain:
... per_device_eval_batch_size=16,
... num_train_epochs=2,
... weight_decay=0.01,
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... save_strategy="epoch",
... load_best_model_at_end=True,
... push_to_hub=True,

2
docs/source/en/tasks/summarization.md
View File

@@ -202,7 +202,7 @@ At this point, only three steps remain:
```py
>>> training_args = Seq2SeqTrainingArguments(
... output_dir="my_awesome_billsum_model",
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... learning_rate=2e-5,
... per_device_train_batch_size=16,
... per_device_eval_batch_size=16,

2
docs/source/en/tasks/text-to-speech.md
View File

@@ -477,7 +477,7 @@ only look at the loss:
... max_steps=4000,
... gradient_checkpointing=True,
... fp16=True,
... evaluation_strategy="steps",
... eval_strategy="steps",
... per_device_eval_batch_size=2,
... save_steps=1000,
... eval_steps=1000,

2
docs/source/en/tasks/token_classification.md
View File

@@ -290,7 +290,7 @@ At this point, only three steps remain:
... per_device_eval_batch_size=16,
... num_train_epochs=2,
... weight_decay=0.01,
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... save_strategy="epoch",
... load_best_model_at_end=True,
... push_to_hub=True,

2
docs/source/en/tasks/translation.md
View File

@@ -209,7 +209,7 @@ At this point, only three steps remain:
```py
>>> training_args = Seq2SeqTrainingArguments(
... output_dir="my_awesome_opus_books_model",
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... learning_rate=2e-5,
... per_device_train_batch_size=16,
... per_device_eval_batch_size=16,

2
docs/source/en/tasks/video_classification.md
View File

@@ -354,7 +354,7 @@ Most of the training arguments are self-explanatory, but one that is quite impor
>>> args = TrainingArguments(
... new_model_name,
... remove_unused_columns=False,
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... save_strategy="epoch",
... learning_rate=5e-5,
... per_device_train_batch_size=batch_size,

2
docs/source/en/trainer.md
View File

@@ -62,7 +62,7 @@ training_args = TrainingArguments(
per_device_eval_batch_size=16,
num_train_epochs=2,
weight_decay=0.01,
evaluation_strategy="epoch",
eval_strategy="epoch",
save_strategy="epoch",
load_best_model_at_end=True,
push_to_hub=True,

4
docs/source/en/training.md
View File

@@ -128,12 +128,12 @@ Call [`~evaluate.compute`] on `metric` to calculate the accuracy of your predict
... return metric.compute(predictions=predictions, references=labels)
```
If you'd like to monitor your evaluation metrics during fine-tuning, specify the `evaluation_strategy` parameter in your training arguments to report the evaluation metric at the end of each epoch:
If you'd like to monitor your evaluation metrics during fine-tuning, specify the `eval_strategy` parameter in your training arguments to report the evaluation metric at the end of each epoch:
```py
>>> from transformers import TrainingArguments, Trainer
>>> training_args = TrainingArguments(output_dir="test_trainer", evaluation_strategy="epoch")
>>> training_args = TrainingArguments(output_dir="test_trainer", eval_strategy="epoch")
```
### Trainer

2
docs/source/es/tasks/asr.md
View File

@@ -260,7 +260,7 @@ En este punto, solo quedan tres pasos:
... gradient_checkpointing=True,
... fp16=True,
... group_by_length=True,
... evaluation_strategy="steps",
... eval_strategy="steps",
... per_device_eval_batch_size=8,
... save_steps=1000,
... eval_steps=1000,

2
docs/source/es/tasks/image_captioning.md
View File

@@ -188,7 +188,7 @@ training_args = TrainingArguments(
per_device_eval_batch_size=32,
gradient_accumulation_steps=2,
save_total_limit=3,
evaluation_strategy="steps",
eval_strategy="steps",
eval_steps=50,
save_strategy="steps",
save_steps=50,

2
docs/source/es/tasks/image_classification.md
View File

@@ -143,7 +143,7 @@ Al llegar a este punto, solo quedan tres pasos:
>>> training_args = TrainingArguments(
... output_dir="./results",
... per_device_train_batch_size=16,
... evaluation_strategy="steps",
... eval_strategy="steps",
... num_train_epochs=4,
... fp16=True,
... save_steps=100,

4
docs/source/es/tasks/language_modeling.md
View File

@@ -232,7 +232,7 @@ A este punto, solo faltan tres pasos:
```py
>>> training_args = TrainingArguments(
... output_dir="./results",
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... learning_rate=2e-5,
... weight_decay=0.01,
... )
@@ -338,7 +338,7 @@ A este punto, solo faltan tres pasos:
```py
>>> training_args = TrainingArguments(
... output_dir="./results",
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... learning_rate=2e-5,
... num_train_epochs=3,
... weight_decay=0.01,

2
docs/source/es/tasks/multiple_choice.md
View File

@@ -212,7 +212,7 @@ En este punto, solo quedan tres pasos:
```py
>>> training_args = TrainingArguments(
... output_dir="./results",
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... learning_rate=5e-5,
... per_device_train_batch_size=16,
... per_device_eval_batch_size=16,

2
docs/source/es/tasks/question_answering.md
View File

@@ -182,7 +182,7 @@ En este punto, solo quedan tres pasos:
```py
>>> training_args = TrainingArguments(
... output_dir="./results",
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... learning_rate=2e-5,
... per_device_train_batch_size=16,
... per_device_eval_batch_size=16,

2
docs/source/es/tasks/summarization.md
View File

@@ -140,7 +140,7 @@ En este punto, solo faltan tres pasos:
```py
>>> training_args = Seq2SeqTrainingArguments(
... output_dir="./results",
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... learning_rate=2e-5,
... per_device_train_batch_size=16,
... per_device_eval_batch_size=16,

2
docs/source/es/trainer.md
View File

@@ -60,7 +60,7 @@ training_args = TrainingArguments(
per_device_eval_batch_size=16,
num_train_epochs=2,
weight_decay=0.01,
evaluation_strategy="epoch",
eval_strategy="epoch",
save_strategy="epoch",
load_best_model_at_end=True,
push_to_hub=True,

4
docs/source/es/training.md
View File

@@ -120,12 +120,12 @@ Define la función `compute` en `metric` para calcular el accuracy de tus predic
... return metric.compute(predictions=predictions, references=labels)
```
Si quieres controlar tus métricas de evaluación durante el fine-tuning, especifica el parámetro `evaluation_strategy` en tus argumentos de entrenamiento para que el modelo tenga en cuenta la métrica de evaluación al final de cada época:
Si quieres controlar tus métricas de evaluación durante el fine-tuning, especifica el parámetro `eval_strategy` en tus argumentos de entrenamiento para que el modelo tenga en cuenta la métrica de evaluación al final de cada época:
```py
>>> from transformers import TrainingArguments
>>> training_args = TrainingArguments(output_dir="test_trainer", evaluation_strategy="epoch")
>>> training_args = TrainingArguments(output_dir="test_trainer", eval_strategy="epoch")
```
### Trainer

2
docs/source/it/migration.md
View File

@@ -167,7 +167,7 @@ Per quanto riguarda la classe `Trainer`:
- Il metodo `is_world_master` di `Trainer` è deprecato a favore di `is_world_process_zero`.
Per quanto riguarda la classe `TrainingArguments`:
- L'argomento `evaluate_during_training` di `TrainingArguments` è deprecato a favore di `evaluation_strategy`.
- L'argomento `evaluate_during_training` di `TrainingArguments` è deprecato a favore di `eval_strategy`.
Per quanto riguarda il modello Transfo-XL:
- L'attributo di configurazione `tie_weight` di Transfo-XL diventa `tie_words_embeddings`.

4
docs/source/it/training.md
View File

@@ -121,12 +121,12 @@ Richiama `compute` su `metric` per calcolare l'accuratezza delle tue previsioni.
... return metric.compute(predictions=predictions, references=labels)
```
Se preferisci monitorare le tue metriche di valutazione durante il fine-tuning, specifica il parametro `evaluation_strategy` nei tuoi training arguments per restituire le metriche di valutazione ad ogni epoca di addestramento:
Se preferisci monitorare le tue metriche di valutazione durante il fine-tuning, specifica il parametro `eval_strategy` nei tuoi training arguments per restituire le metriche di valutazione ad ogni epoca di addestramento:
```py
>>> from transformers import TrainingArguments, Trainer
>>> training_args = TrainingArguments(output_dir="test_trainer", evaluation_strategy="epoch")
>>> training_args = TrainingArguments(output_dir="test_trainer", eval_strategy="epoch")
```
### Trainer

2
docs/source/ja/model_memory_anatomy.md
View File

@@ -136,7 +136,7 @@ Tue Jan 11 08:58:05 2022
```py
default_args = {
"output_dir": "tmp",
"evaluation_strategy": "steps",
"eval_strategy": "steps",
"num_train_epochs": 1,
"log_level": "error",
"report_to": "none",

2
docs/source/ja/tasks/asr.md
View File

@@ -270,7 +270,7 @@ MInDS-14 データセットのサンプリング レートは 8000kHz です (
... gradient_checkpointing=True,
... fp16=True,
... group_by_length=True,
... evaluation_strategy="steps",
... eval_strategy="steps",
... per_device_eval_batch_size=8,
... save_steps=1000,
... eval_steps=1000,

2
docs/source/ja/tasks/audio_classification.md
View File

@@ -221,7 +221,7 @@ MInDS-14 データセットのサンプリング レートは 8000khz です (
```py
>>> training_args = TrainingArguments(
... output_dir="my_awesome_mind_model",
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... save_strategy="epoch",
... learning_rate=3e-5,
... per_device_train_batch_size=32,

2
docs/source/ja/tasks/document_question_answering.md
View File

@@ -403,7 +403,7 @@ end_index 18
... num_train_epochs=20,
... save_steps=200,
... logging_steps=50,
... evaluation_strategy="steps",
... eval_strategy="steps",
... learning_rate=5e-5,
... save_total_limit=2,
... remove_unused_columns=False,

2
docs/source/ja/tasks/image_captioning.md
View File

@@ -194,7 +194,7 @@ training_args = TrainingArguments(
per_device_eval_batch_size=32,
gradient_accumulation_steps=2,
save_total_limit=3,
evaluation_strategy="steps",
eval_strategy="steps",
eval_steps=50,
save_strategy="steps",
save_steps=50,

2
docs/source/ja/tasks/image_classification.md
View File

@@ -308,7 +308,7 @@ food["test"].set_transform(preprocess_val)
>>> training_args = TrainingArguments(
... output_dir="my_awesome_food_model",
... remove_unused_columns=False,
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... save_strategy="epoch",
... learning_rate=5e-5,
... per_device_train_batch_size=16,

2
docs/source/ja/tasks/knowledge_distillation_for_image_classification.md
View File

@@ -112,7 +112,7 @@ training_args = TrainingArguments(
fp16=True,
logging_dir=f"{repo_name}/logs",
logging_strategy="epoch",
evaluation_strategy="epoch",
eval_strategy="epoch",
save_strategy="epoch",
load_best_model_at_end=True,
metric_for_best_model="accuracy",

2
docs/source/ja/tasks/language_modeling.md
View File

@@ -246,7 +246,7 @@ Apply the `group_texts` function over the entire dataset:
```py
>>> training_args = TrainingArguments(
... output_dir="my_awesome_eli5_clm-model",
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... learning_rate=2e-5,
... weight_decay=0.01,
... push_to_hub=True,

2
docs/source/ja/tasks/masked_language_modeling.md
View File

@@ -231,7 +231,7 @@ pip install transformers datasets evaluate
```py
>>> training_args = TrainingArguments(
... output_dir="my_awesome_eli5_mlm_model",
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... learning_rate=2e-5,
... num_train_epochs=3,
... weight_decay=0.01,

2
docs/source/ja/tasks/multiple_choice.md
View File

@@ -266,7 +266,7 @@ tokenized_swag = swag.map(preprocess_function, batched=True)
```py
>>> training_args = TrainingArguments(
... output_dir="my_awesome_swag_model",
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... save_strategy="epoch",
... load_best_model_at_end=True,
... learning_rate=5e-5,

2
docs/source/ja/tasks/question_answering.md
View File

@@ -220,7 +220,7 @@ pip install transformers datasets evaluate
```py
>>> training_args = TrainingArguments(
... output_dir="my_awesome_qa_model",
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... learning_rate=2e-5,
... per_device_train_batch_size=16,
... per_device_eval_batch_size=16,

2
docs/source/ja/tasks/semantic_segmentation.md
View File

@@ -323,7 +323,7 @@ pip install -q datasets transformers evaluate
... per_device_train_batch_size=2,
... per_device_eval_batch_size=2,
... save_total_limit=3,
... evaluation_strategy="steps",
... eval_strategy="steps",
... save_strategy="steps",
... save_steps=20,
... eval_steps=20,

2
docs/source/ja/tasks/sequence_classification.md
View File

@@ -324,7 +324,7 @@ pip install -q datasets transformers evaluate
... per_device_train_batch_size=2,
... per_device_eval_batch_size=2,
... save_total_limit=3,
... evaluation_strategy="steps",
... eval_strategy="steps",
... save_strategy="steps",
... save_steps=20,
... eval_steps=20,

2
docs/source/ja/tasks/summarization.md
View File

@@ -204,7 +204,7 @@ pip install transformers datasets evaluate rouge_score
```py
>>> training_args = Seq2SeqTrainingArguments(
... output_dir="my_awesome_billsum_model",
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... learning_rate=2e-5,
... per_device_train_batch_size=16,
... per_device_eval_batch_size=16,

2
docs/source/ja/tasks/text-to-speech.md
View File

@@ -477,7 +477,7 @@ SpeechT5 では、モデルのデコーダ部分への入力が 2 分の 1 に
... max_steps=4000,
... gradient_checkpointing=True,
... fp16=True,
... evaluation_strategy="steps",
... eval_strategy="steps",
... per_device_eval_batch_size=2,
... save_steps=1000,
... eval_steps=1000,

2
docs/source/ja/tasks/token_classification.md
View File

@@ -288,7 +288,7 @@ pip install transformers datasets evaluate seqeval
... per_device_eval_batch_size=16,
... num_train_epochs=2,
... weight_decay=0.01,
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... save_strategy="epoch",
... load_best_model_at_end=True,
... push_to_hub=True,

2
docs/source/ja/tasks/translation.md
View File

@@ -208,7 +208,7 @@ pip install transformers datasets evaluate sacrebleu
```py
>>> training_args = Seq2SeqTrainingArguments(
... output_dir="my_awesome_opus_books_model",
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... learning_rate=2e-5,
... per_device_train_batch_size=16,
... per_device_eval_batch_size=16,

2
docs/source/ja/tasks/video_classification.md
View File

@@ -360,7 +360,7 @@ You should probably TRAIN this model on a down-stream task to be able to use it
>>> args = TrainingArguments(
... new_model_name,
... remove_unused_columns=False,
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... save_strategy="epoch",
... learning_rate=5e-5,
... per_device_train_batch_size=batch_size,

4
docs/source/ja/training.md
View File

@@ -135,12 +135,12 @@ BERTモデルの事前学習済みのヘッドは破棄され、ランダムに
... return metric.compute(predictions=predictions, references=labels)
```
評価メトリクスをファインチューニング中に監視したい場合、トレーニング引数で `evaluation_strategy` パラメータを指定して、各エポックの終了時に評価メトリクスを報告します：
評価メトリクスをファインチューニング中に監視したい場合、トレーニング引数で `eval_strategy` パラメータを指定して、各エポックの終了時に評価メトリクスを報告します：
```python
>>> from transformers import TrainingArguments, Trainer
>>> training_args = TrainingArguments(output_dir="test_trainer", evaluation_strategy="epoch")
>>> training_args = TrainingArguments(output_dir="test_trainer", eval_strategy="epoch")
```
### Trainer

2
docs/source/ko/model_memory_anatomy.md
View File

@@ -132,7 +132,7 @@ Tue Jan 11 08:58:05 2022
```py
default_args = {
"output_dir": "tmp",
"evaluation_strategy": "steps",
"eval_strategy": "steps",
"num_train_epochs": 1,
"log_level": "error",
"report_to": "none",

2
docs/source/ko/tasks/asr.md
View File

@@ -274,7 +274,7 @@ MInDS-14 데이터 세트의 샘플링 레이트는 8000kHz이므로([데이터
... gradient_checkpointing=True,
... fp16=True,
... group_by_length=True,
... evaluation_strategy="steps",
... eval_strategy="steps",
... per_device_eval_batch_size=8,
... save_steps=1000,
... eval_steps=1000,

2
docs/source/ko/tasks/audio_classification.md
View File

@@ -221,7 +221,7 @@ MinDS-14 데이터 세트의 샘플링 속도는 8000khz이므로(이 정보는
```py
>>> training_args = TrainingArguments(
... output_dir="my_awesome_mind_model",
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... save_strategy="epoch",
... learning_rate=3e-5,
... per_device_train_batch_size=32,

2
docs/source/ko/tasks/document_question_answering.md
View File

@@ -385,7 +385,7 @@ end_index 18
... num_train_epochs=20,
... save_steps=200,
... logging_steps=50,
... evaluation_strategy="steps",
... eval_strategy="steps",
... learning_rate=5e-5,
... save_total_limit=2,
... remove_unused_columns=False,

2
docs/source/ko/tasks/image_captioning.md
View File

@@ -201,7 +201,7 @@ training_args = TrainingArguments(
per_device_eval_batch_size=32,
gradient_accumulation_steps=2,
save_total_limit=3,
evaluation_strategy="steps",
eval_strategy="steps",
eval_steps=50,
save_strategy="steps",
save_steps=50,

2
docs/source/ko/tasks/image_classification.md
View File

@@ -301,7 +301,7 @@ food["test"].set_transform(preprocess_val)
>>> training_args = TrainingArguments(
... output_dir="my_awesome_food_model",
... remove_unused_columns=False,
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... save_strategy="epoch",
... learning_rate=5e-5,
... per_device_train_batch_size=16,

2
docs/source/ko/tasks/language_modeling.md
View File

@@ -233,7 +233,7 @@ pip install transformers datasets evaluate
```py
>>> training_args = TrainingArguments(
... output_dir="my_awesome_eli5_clm-model",
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... learning_rate=2e-5,
... weight_decay=0.01,
... push_to_hub=True,

2
docs/source/ko/tasks/masked_language_modeling.md
View File

@@ -236,7 +236,7 @@ Hugging Face 계정에 로그인하여 모델을 업로드하고 커뮤니티와
```py
>>> training_args = TrainingArguments(
... output_dir="my_awesome_eli5_mlm_model",
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... learning_rate=2e-5,
... num_train_epochs=3,
... weight_decay=0.01,

2
docs/source/ko/tasks/multiple_choice.md
View File

@@ -265,7 +265,7 @@ tokenized_swag = swag.map(preprocess_function, batched=True)
```py
>>> training_args = TrainingArguments(
... output_dir="my_awesome_swag_model",
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... save_strategy="epoch",
... load_best_model_at_end=True,
... learning_rate=5e-5,

2
docs/source/ko/tasks/question_answering.md
View File

@@ -215,7 +215,7 @@ pip install transformers datasets evaluate
```py
>>> training_args = TrainingArguments(
... output_dir="my_awesome_qa_model",
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... learning_rate=2e-5,
... per_device_train_batch_size=16,
... per_device_eval_batch_size=16,

2
docs/source/ko/tasks/semantic_segmentation.md
View File

@@ -317,7 +317,7 @@ pip install -q datasets transformers evaluate
... per_device_train_batch_size=2,
... per_device_eval_batch_size=2,
... save_total_limit=3,
... evaluation_strategy="steps",
... eval_strategy="steps",
... save_strategy="steps",
... save_steps=20,
... eval_steps=20,

2
docs/source/ko/tasks/sequence_classification.md
View File

@@ -185,7 +185,7 @@ tokenized_imdb = imdb.map(preprocess_function, batched=True)
... per_device_eval_batch_size=16,
... num_train_epochs=2,
... weight_decay=0.01,
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... save_strategy="epoch",
... load_best_model_at_end=True,
... push_to_hub=True,

2
docs/source/ko/tasks/summarization.md
View File

@@ -211,7 +211,7 @@ Hugging Face 계정에 로그인하면 모델을 업로드하고 커뮤니티에
```py
>>> training_args = Seq2SeqTrainingArguments(
... output_dir="my_awesome_billsum_model",
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... learning_rate=2e-5,
... per_device_train_batch_size=16,
... per_device_eval_batch_size=16,

2
docs/source/ko/tasks/token_classification.md
View File

@@ -288,7 +288,7 @@ Hugging Face 계정에 로그인하여 모델을 업로드하고 커뮤니티에
... per_device_eval_batch_size=16,
... num_train_epochs=2,
... weight_decay=0.01,
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... save_strategy="epoch",
... load_best_model_at_end=True,
... push_to_hub=True,

2
docs/source/ko/tasks/translation.md
View File

@@ -209,7 +209,7 @@ pip install transformers datasets evaluate sacrebleu
```py
>>> training_args = Seq2SeqTrainingArguments(
... output_dir="my_awesome_opus_books_model",
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... learning_rate=2e-5,
... per_device_train_batch_size=16,
... per_device_eval_batch_size=16,

2
docs/source/ko/tasks/video_classification.md
View File

@@ -358,7 +358,7 @@ You should probably TRAIN this model on a down-stream task to be able to use it
>>> args = TrainingArguments(
... new_model_name,
... remove_unused_columns=False,
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... save_strategy="epoch",
... learning_rate=5e-5,
... per_device_train_batch_size=batch_size,

4
docs/source/ko/training.md
View File

@@ -129,12 +129,12 @@ rendered properly in your Markdown viewer.
... return metric.compute(predictions=predictions, references=labels)
```
미세 튜닝 중에 평가 지표를 모니터링하려면 훈련 인수에 `evaluation_strategy` 파라미터를 지정하여 각 에폭이 끝날 때 평가 지표를 확인할 수 있습니다:
미세 튜닝 중에 평가 지표를 모니터링하려면 훈련 인수에 `eval_strategy` 파라미터를 지정하여 각 에폭이 끝날 때 평가 지표를 확인할 수 있습니다:
```py
>>> from transformers import TrainingArguments, Trainer
>>> training_args = TrainingArguments(output_dir="test_trainer", evaluation_strategy="epoch")
>>> training_args = TrainingArguments(output_dir="test_trainer", eval_strategy="epoch")
```
### 훈련 하기[[trainer]]

2
docs/source/pt/tasks/token_classification.md
View File

@@ -180,7 +180,7 @@ Nesse ponto, restam apenas três passos:
```py
>>> training_args = TrainingArguments(
... output_dir="./results",
... evaluation_strategy="epoch",
... eval_strategy="epoch",
... learning_rate=2e-5,
... per_device_train_batch_size=16,
... per_device_eval_batch_size=16,

4
docs/source/pt/training.md
View File

@@ -146,13 +146,13 @@ todos os modelos de 🤗 Transformers retornam logits).
... return metric.compute(predictions=predictions, references=labels)
```
Se quiser controlar as suas métricas de avaliação durante o fine-tuning, especifique o parâmetro `evaluation_strategy`
Se quiser controlar as suas métricas de avaliação durante o fine-tuning, especifique o parâmetro `eval_strategy`
nos seus argumentos de treinamento para que o modelo considere a métrica de avaliação ao final de cada época:
```py
>>> from transformers import TrainingArguments
>>> training_args = TrainingArguments(output_dir="test_trainer", evaluation_strategy="epoch")
>>> training_args = TrainingArguments(output_dir="test_trainer", eval_strategy="epoch")
```
### Trainer

2
docs/source/zh/tasks/asr.md
View File

@@ -288,7 +288,7 @@ Wav2Vec2 分词器仅训练了大写字符，因此您需要确保文本与分
... gradient_checkpointing=True,
... fp16=True,
... group_by_length=True,
... evaluation_strategy="steps",
... eval_strategy="steps",
... per_device_eval_batch_size=8,
... save_steps=1000,
... eval_steps=1000,

4
docs/source/zh/training.md
View File

@@ -125,12 +125,12 @@ rendered properly in your Markdown viewer.
... return metric.compute(predictions=predictions, references=labels)
```
如果您希望在微调过程中监视评估指标，请在您的训练参数中指定 `evaluation_strategy` 参数，以在每个`epoch`结束时展示评估指标：
如果您希望在微调过程中监视评估指标，请在您的训练参数中指定 `eval_strategy` 参数，以在每个`epoch`结束时展示评估指标：
```py
>>> from transformers import TrainingArguments, Trainer
>>> training_args = TrainingArguments(output_dir="test_trainer", evaluation_strategy="epoch")
>>> training_args = TrainingArguments(output_dir="test_trainer", eval_strategy="epoch")
```
### 训练器