Skip to content

[chatgpt] unnify datasets #3218

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
Fazziekey merged 1 commit into from
Mar 23, 2023
Merged

[chatgpt] unnify datasets #3218

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
11 changes: 8 additions & 3 deletions applications/ChatGPT/chatgpt/dataset/sft_dataset.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -54,7 +54,8 @@ class SFTDataset(Dataset):

def __init__(self, dataset, tokenizer: Callable, max_length: int=512) -> None:
super().__init__()
self.prompts = []
# self.prompts = []
self.input_ids = []

for data in tqdm(dataset, disable=not is_rank_0()):
prompt = data['prompt'] + data['completion'] + "<|endoftext|>"
Expand All @@ -64,14 +65,18 @@ def __init__(self, dataset, tokenizer: Callable, max_length: int=512) -> None:
truncation=True,
return_tensors="pt")

self.prompts.append(prompt_token)
# self.prompts.append(prompt_token)s
self.input_ids.append(prompt_token)
self.labels = copy.deepcopy(self.input_ids)

def __len__(self):
length = len(self.prompts)
return length

def __getitem__(self, idx):
return self.prompts[idx]
# dict(input_ids=self.input_ids[i], labels=self.labels[i])
return dict(input_ids=self.input_ids[i], labels=self.labels[i])
# return dict(self.prompts[idx], self.prompts[idx])


def _tokenize_fn(strings: Sequence[str], tokenizer: transformers.PreTrainedTokenizer) -> Dict:
Expand Down
6 changes: 4 additions & 2 deletions applications/ChatGPT/chatgpt/trainer/sft.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -63,11 +63,13 @@ def fit(self, logger, use_lora, log_interval=10):
for batch_id, batch in enumerate(self.train_dataloader):
prompt_ids = batch["input_ids"]
p_mask = batch["attention_mask"]
labels = batch["labels"]
prompt_ids = prompt_ids.squeeze(1).cuda()
p_mask = p_mask.squeeze(1).cuda()
prompt_logits = self.model(prompt_ids, attention_mask=p_mask)
# prompt_logits = self.model(prompt_ids, attention_mask=p_mask, labels=labels)
loss, prompt_logits = self.model(prompt_ids, attention_mask=p_mask, labels=labels)

loss = self.loss_fn(prompt_logits, prompt_ids)
# loss = self.loss_fn(prompt_logits, labels)
self.strategy.backward(loss, self.model, self.optimizer)
self.strategy.optimizer_step(self.optimizer)
self.optimizer.zero_grad()
Expand Down