Merge of PR #1626

PiperOrigin-RevId: 258838381

Author: voluntadpear

docs/walkthrough.md

@@ -210,6 +210,7 @@ There are a number of translation data-sets in T2T:
 * English-Czech: `--problem=translate_encs_wmt32k`
 * English-Chinese: `--problem=translate_enzh_wmt32k`
 * English-Vietnamese: `--problem=translate_envi_iwslt32k`
+* English-Spanish: `--problem=translate_enes_wmt32k`
 
 You can get translations in the other direction by appending `_rev` to
 the problem name, e.g., for German-English use

tensor2tensor/data_generators/all_problems.py

@@ -78,6 +78,7 @@
     "tensor2tensor.data_generators.transduction_problems",
     "tensor2tensor.data_generators.translate_encs",
     "tensor2tensor.data_generators.translate_ende",
+    "tensor2tensor.data_generators.translate_enes",
     "tensor2tensor.data_generators.translate_enet",
     "tensor2tensor.data_generators.translate_enfr",
     "tensor2tensor.data_generators.translate_enid",

tensor2tensor/data_generators/translate_enes.py

@@ -53,6 +53,7 @@
     ],
 ]
 
+
 @registry.register_problem
 class TranslateEnesWmt32k(translate.TranslateProblem):
   """En-es translation trained on WMT corpus."""
@@ -70,6 +71,7 @@ def source_data_files(self, dataset_split):
   def vocab_data_files(self):
     return _ENES_TRAIN_DATASETS
 
+
 @registry.register_problem
 class TranslateEnesWmtClean32k(TranslateEnesWmt32k):
   """En-es translation trained on WMT with further cleaning."""
@@ -82,6 +84,7 @@ def use_vocab_from_other_problem(self):
   def datatypes_to_clean(self):
     return ["txt"]
 
+
 @registry.register_problem
 class TranslateEnesWmt32kPacked(TranslateEnesWmt32k):
 
@@ -93,6 +96,7 @@ def packed_length(self):
   def use_vocab_from_other_problem(self):
     return TranslateEnesWmt32k()
 
+
 @registry.register_problem
 class TranslateEnesWmt8k(TranslateEnesWmt32k):
   """Problem spec for WMT En-Es translation."""
@@ -101,6 +105,7 @@ class TranslateEnesWmt8k(TranslateEnesWmt32k):
   def approx_vocab_size(self):
     return 2**13  # 8192
 
+
 @registry.register_problem
 class TranslateEnesWmt8kPacked(TranslateEnesWmt8k):
 
@@ -112,10 +117,11 @@ def packed_length(self):
   def use_vocab_from_other_problem(self):
     return TranslateEnesWmt8k()
 
+
 @registry.register_problem
 class TranslateEnesWmtCharacters(TranslateEnesWmt8k):
   """Problem spec for WMT En-Es translation."""
 
   @property
   def vocab_type(self):
-    return text_problems.VocabType.CHARACTER
+    return text_problems.VocabType.CHARACTER