mod gpt
This commit is contained in:
parent
746ce6bb8a
commit
ffe1bf5eab
7
gpt.py
7
gpt.py
|
|
@ -39,11 +39,8 @@ def main():
|
||||||
tokenized_dataset = dataset.map(tokenize_function, batched=True)
|
tokenized_dataset = dataset.map(tokenize_function, batched=True)
|
||||||
|
|
||||||
# Model i data collator
|
# Model i data collator
|
||||||
model = AutoModelForCausalLM.from_pretrained(
|
model = AutoModelForCausalLM.from_pretrained(MODEL_NAME)
|
||||||
MODEL_NAME,
|
model.resize_token_embeddings(len(tokenizer), mean_resizing=False)
|
||||||
mean_resizing=False
|
|
||||||
)
|
|
||||||
model.resize_token_embeddings(len(tokenizer))
|
|
||||||
|
|
||||||
data_collator = DataCollatorForLanguageModeling(
|
data_collator = DataCollatorForLanguageModeling(
|
||||||
tokenizer=tokenizer,
|
tokenizer=tokenizer,
|
||||||
|
|
|
||||||
Loading…
Reference in New Issue