def setup(self):
# called on every GPU
self.dataset = load_dataset("wmt14", "de-en", "val")
self.tokenizer = FSMTTokenizer.from_pretrained("facebook/wmt19-de-en")
self.vocab_size = self.tokenizer.vocab_size
val_len = len(self.dataset["validation"])
n_val = int(val_len*self.val_fraction)
n_train = val_len - n_val
for key in ["validation", "test"]:
self.dataset[key] = self.dataset[key].map(self.tokenize)
self.dataset["train"], self.dataset["validation"] = random_split(self.dataset["validation"], [n_train, n_val])
def get_tokenizer(self, mname):
if mname not in self.tokenizers_cache:
self.tokenizers_cache[mname] = FSMTTokenizer.from_pretrained(mname)
return self.tokenizers_cache[mname]
# This script creates a super tiny model that is useful inside tests, when we just want to test that
# the machinery works, without needing to the check the quality of the outcomes.
#
# This version creates a tiny model through reduction of a normal pre-trained model, but keeping the
# full vocab, merges file, and thus also resulting in a larger model due to a large vocab size.
# This gives ~3MB in total for all files.
#
# If you want a 50 times smaller than this see `fsmt-make-super-tiny-model.py`, which is slightly more complicated
#
#
# It will be used then as "stas/tiny-wmt19-en-de"
# Build
from transformers import FSMTTokenizer, FSMTConfig, FSMTForConditionalGeneration
mname = "facebook/wmt19-en-de"
tokenizer = FSMTTokenizer.from_pretrained(mname)
# get the correct vocab sizes, etc. from the master model
config = FSMTConfig.from_pretrained(mname)
config.update(dict(
d_model=4,
encoder_layers=1, decoder_layers=1,
encoder_ffn_dim=4, decoder_ffn_dim=4,
encoder_attention_heads=1, decoder_attention_heads=1))
tiny_model = FSMTForConditionalGeneration(config)
print(f"num of params {tiny_model.num_parameters()}")
# Test
batch = tokenizer(["Making tiny model"], return_tensors="pt")
outputs = tiny_model(**batch)
def get_tokenizer(self, mname):
return FSMTTokenizer.from_pretrained(mname)
def prepare_data(self):
# called only on 1 GPU
load_dataset("wmt14", "de-en", "val")
FSMTTokenizer.from_pretrained("facebook/wmt19-de-en")