riaz.somc/transformers
1
0
Fork 0
mirror of https://github.com/huggingface/transformers.git synced 2025-07-15 02:28:24 +06:00
Code Issues Actions 16 Packages Projects Releases Wiki Activity

more details in GPT-2 usage example

Browse Source
This commit is contained in:
Thomas Wolf 2019-02-18 14:37:41 +01:00 committed by GitHub
parent 522733f6cb
commit ada22a1c9e
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 19 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

28
README.md
View File

@ -400,12 +400,15 @@ logging.basicConfig(level=logging.INFO)
# Load pre-trained model tokenizer (vocabulary) # Load pre-trained model tokenizer (vocabulary)
tokenizer = GPT2Tokenizer.from_pretrained('gpt2') tokenizer = GPT2Tokenizer.from_pretrained('gpt2')
# Encode input # Encode some inputs
text = "Who was Jim Henson ? Jim Henson was a puppeteer" text_1 = "Who was Jim Henson ?"
indexed_tokens = tokenizer.encode(text) text_2 = "Jim Henson was a puppeteer"
indexed_tokens_1 = tokenizer.encode(text_1)
indexed_tokens_2 = tokenizer.encode(text_2)
# Convert inputs to PyTorch tensors # Convert inputs to PyTorch tensors
tokens_tensor = torch.tensor([indexed_tokens]) tokens_tensor_1 = torch.tensor([indexed_tokens_1])
tokens_tensor_2 = torch.tensor([indexed_tokens_2])
``` ```
Let's see how to use `GPT2Model` to get hidden states Let's see how to use `GPT2Model` to get hidden states
@ -416,12 +419,16 @@ model = GPT2Model.from_pretrained('gpt2')
model.eval() model.eval()
# If you have a GPU, put everything on cuda # If you have a GPU, put everything on cuda
tokens_tensor = tokens_tensor.to('cuda') tokens_tensor_1 = tokens_tensor_1.to('cuda')
tokens_tensor_2 = tokens_tensor_2.to('cuda')
model.to('cuda') model.to('cuda')
# Predict hidden states features for each layer # Predict hidden states features for each layer
with torch.no_grad(): with torch.no_grad():
hidden_states = model(tokens_tensor) hidden_states_1, past = model(tokens_tensor_1)
# past can be used to reuse precomputed hidden state in a subsequent predictions
# (see beam-search examples in the run_gpt2.py example
hidden_states-2, past = model(tokens_tensor_2, past=past)
``` ```
And how to use `GPT2LMHeadModel` And how to use `GPT2LMHeadModel`
@ -432,15 +439,18 @@ model = GPT2LMHeadModel.from_pretrained('gpt2')
model.eval() model.eval()
# If you have a GPU, put everything on cuda # If you have a GPU, put everything on cuda
tokens_tensor = tokens_tensor.to('cuda') tokens_tensor_1 = tokens_tensor.to('cuda')
model.to('cuda') model.to('cuda')
# Predict all tokens # Predict all tokens
with torch.no_grad(): with torch.no_grad():
predictions = model(tokens_tensor) predictions_1, past = model(tokens_tensor_1)
# past can be used to reuse precomputed hidden state in a subsequent predictions
# (see beam-search examples in the run_gpt2.py example
predictions_2, past = model(tokens_tensor_2, past=past)
# get the predicted last token # get the predicted last token
predicted_index = torch.argmax(predictions[0, -1, :]).item() predicted_index = torch.argmax(predictions_2[0, -1, :]).item()
predicted_token = tokenizer.decode([predicted_index]) predicted_token = tokenizer.decode([predicted_index])
``` ```