SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

VLMs: enable generation tests (#33533)

Browse Source 
* add tests

* fix whisper

* update

* nit

* add qwen2-vl

* more updates!

* better this way

* fix this one

* fix more tests

* fix final tests, hope so

* fix led

* Update tests/generation/test_utils.py

Co-authored-by: Joao Gante <joaofranciscocardosogante@gmail.com>

* pr comments

* not pass pixels and extra for low-mem tests, very flaky because of visio tower

---------

Co-authored-by: Joao Gante <joaofranciscocardosogante@gmail.com>
This commit is contained in:
Raushan Turganbay
2024-09-19 12:04:24 +02:00
committed by GitHub
parent e40bb4845e
commit d7975a5874
22 changed files with 500 additions and 207 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
tests/models/llava/test_modeling_llava.py
View File

@@ -36,6 +36,7 @@ from transformers.testing_utils import (
torch_device,
)
from ...generation.test_utils import GenerationTesterMixin
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor
@@ -80,7 +81,7 @@ class LlavaVisionText2TextModelTester:
"initializer_range": 0.02,
"num_labels": 3,
"num_choices": 4,
"pad_token_id": 0,
"pad_token_id": 1,
},
is_training=True,
vision_config={
@@ -106,7 +107,7 @@ class LlavaVisionText2TextModelTester:
self.vision_feature_layer = vision_feature_layer
self.text_config = text_config
self.vision_config = vision_config
self.seq_length = seq_length
self.pad_token_id = text_config["pad_token_id"]
self.num_hidden_layers = text_config["num_hidden_layers"]
self.vocab_size = text_config["vocab_size"]
@@ -118,6 +119,8 @@ class LlavaVisionText2TextModelTester:
self.num_channels = 3
self.image_size = 336
self.encoder_seq_length = 231
self.num_image_tokens = 224
self.seq_length = seq_length + self.num_image_tokens
def get_config(self):
return LlavaConfig(
@@ -128,6 +131,7 @@ class LlavaVisionText2TextModelTester:
projector_hidden_act=self.projector_hidden_act,
vision_feature_select_strategy=self.vision_feature_select_strategy,
vision_feature_layer=self.vision_feature_layer,
image_seq_length=self.num_image_tokens,
)
def prepare_config_and_inputs(self):
@@ -148,8 +152,8 @@ class LlavaVisionText2TextModelTester:
config, pixel_values = config_and_inputs
input_ids = ids_tensor([self.batch_size, self.seq_length], config.text_config.vocab_size - 1) + 1
attention_mask = input_ids.ne(1).to(torch_device)
# we are giving 3 images let's make sure we pass in 3 image tokens
input_ids[:, 1] = config.image_token_index
input_ids[input_ids == config.image_token_index] = self.pad_token_id
input_ids[:, : self.num_image_tokens] = config.image_token_index
inputs_dict = {
"pixel_values": pixel_values,
"input_ids": input_ids,
@@ -172,12 +176,13 @@ class LlavaVisionText2TextModelTester:
@require_torch
class LlavaForConditionalGenerationModelTest(ModelTesterMixin, unittest.TestCase):
class LlavaForConditionalGenerationModelTest(ModelTesterMixin, GenerationTesterMixin, unittest.TestCase):
"""
Model tester for `LlavaForConditionalGeneration`.
"""
all_model_classes = (LlavaForConditionalGeneration,) if is_torch_available() else ()
all_generative_model_classes = (LlavaForConditionalGeneration,) if is_torch_available() else ()
pipeline_model_mapping = {"image-to-text": LlavaForConditionalGeneration} if is_torch_available() else {}
test_pruning = False
test_head_masking = False