github/ColossalAI
1
0
Fork 0
mirror of https://github.com/hpcaitech/ColossalAI.git synced 2025-09-12 20:54:35 +00:00
Code Issues Packages Projects Releases Wiki Activity

[Test/CI] remove test cases to reduce CI duration (#5753)

Browse Source 
* [test] smaller gpt2 test case

* [test] reduce test cases: tests/test_zero/test_gemini/test_zeroddp_state_dict.py

* [test] reduce test cases: tests/test_zero/test_gemini/test_grad_accum.py

* [test] reduce test cases tests/test_zero/test_gemini/test_optim.py

* Revert "[test] smaller gpt2 test case"

Some tests might depend on the size of model (num of chunks)

This reverts commit df705a5210.

* [test] reduce test cases: tests/test_checkpoint_io/test_gemini_checkpoint_io.py

* [CI] smaller test model for two mwo the two modifid cases

* [CI] hardcode gpt model for tests/test_zero/test_gemini/test_search.py since we need a fixed answer there
This commit is contained in:
botbw
2024-06-05 11:29:04 +08:00
committed by GitHub
parent 79f7a7b211
commit 80c3c8789b
6 changed files with 40 additions and 76 deletions
Download Patch File Download Diff File Expand all files Collapse all files

49
tests/kit/model_zoo/transformers/gpt.py
View File

@@ -18,23 +18,8 @@ def data_gen():
# tokenized_input = tokenizer(input, return_tensors='pt')
# input_ids = tokenized_input['input_ids']
# attention_mask = tokenized_input['attention_mask']
# input_ids = torch.tensor([[15496, 11, 616, 3290, 318, 13779, 318, 13779]], dtype=torch.int64)
# attention_mask = torch.tensor([[1, 1, 1, 1, 1, 1, 1, 1]], dtype=torch.int64)
input_ids = torch.tensor(
[
[15496, 11, 616, 3290, 318, 13779, 318, 13779, 15496, 11, 616, 3290, 318, 13779, 318, 13779],
[15496, 11, 616, 3290, 318, 13779, 318, 13779, 15496, 11, 616, 3290, 318, 13779, 318, 13779],
],
dtype=torch.int64,
)
attention_mask = torch.tensor(
[
[1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1],
[1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1],
],
dtype=torch.int64,
)
input_ids = torch.tensor([[22, 11, 616, 4, 5, 13, 318, 345]], dtype=torch.int64)
attention_mask = torch.tensor([[1, 1, 1, 1, 1, 1, 1, 1]], dtype=torch.int64)
return dict(input_ids=input_ids, attention_mask=attention_mask)
@@ -50,9 +35,9 @@ def data_gen_for_question_answering():
# question answering data gen
# `labels` is the type not the token id for token classification, 0 or 1
data = data_gen()
start_positions = torch.tensor([[0], [0]], dtype=torch.int64)
start_positions = torch.tensor([0], dtype=torch.int64)
data["start_positions"] = start_positions
end_positions = torch.tensor([[1], [1]], dtype=torch.int64)
end_positions = torch.tensor([1], dtype=torch.int64)
data["end_positions"] = end_positions
return data
@@ -61,20 +46,14 @@ def data_gen_for_token_classification():
# token classification data gen
# `labels` is the type not the token id for token classification, 0 or 1
data = data_gen()
data["labels"] = torch.tensor(
[
[0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1],
[0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1],
],
dtype=torch.int64,
)
data["labels"] = torch.tensor([[0, 0, 0, 0, 0, 0, 0, 1]], dtype=torch.int64)
return data
def data_gen_for_sequence_classification():
# sequence classification data gen
data = data_gen()
data["labels"] = torch.tensor([[1], [1]], dtype=torch.int64)
data["labels"] = torch.tensor([1], dtype=torch.int64)
return data
@@ -82,18 +61,12 @@ def date_gen_for_double_heads():
num_choices = 2
batch_size = 2
input_ids = torch.tensor(
[
[15496, 11, 616, 3290, 318, 13779, 318, 13779, 15496, 11, 616, 3290, 318, 13779, 318, 13779],
[15496, 11, 616, 3290, 318, 13779, 318, 13779, 15496, 11, 616, 3290, 318, 13779, 318, 13779],
],
[[46, 11, 616, 432, 318, 19, 318, 555], [777, 11, 235, 333, 318, 231, 468, 136]],
dtype=torch.int64,
)
attention_mask = torch.tensor(
[[1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]],
dtype=torch.int64,
)
attention_mask = torch.tensor([[1, 1, 1, 1, 1, 1, 1, 1], [1, 1, 1, 1, 1, 1, 1, 1]], dtype=torch.int64)
mc_labels = torch.zeros(input_ids.shape[0], dtype=torch.int64)
mc_token_ids = torch.arange(0, num_choices, dtype=torch.int64)
mc_token_ids = mc_token_ids.expand((batch_size, num_choices))
multiple_choice_inputs_ids = input_ids.unsqueeze(1).expand(-1, num_choices, -1).contiguous()
@@ -122,14 +95,14 @@ config = transformers.GPT2Config(
n_layer=2,
n_head=4,
n_embd=128,
vocab_size=50258,
vocab_size=1024,
attn_pdrop=0,
embd_pdrop=0,
resid_pdrop=0,
summary_first_dropout=0,
hidden_dropout=0,
problem_type="single_label_classification",
pad_token_id=50256,
pad_token_id=1022,
tie_word_embeddings=True,
)