Add BLIP-2 (#21441)

* First draft

* More improvements

* More improvements

* Improve conversion script

* Convert all weights

* Make forward pass work

* Make logits match

* More improvements

* More improvements

* More improvements

* Use get_input_embeddings

* Improve some more

* Improve model tests

* Improve model tests

* More improvements

* Fix processor

* Update files

* Update prepare_inputs_for_generation

* More improvements

* Fix copies

* More fixes

* Make fixup

* More improvements

* Add support for seq2seq language model

* More improvements

* Fix test

* More improvements

* Improve conversion script

* Remove some todo's

* Fix README's

* Improve conversion script

* Fix generation

* Fix style and remove Blip2Model

* Fix model outputs

* More improvements

* Set eos_token_id in config

* Fix quality

* Small improvements

* Add processor tests

* More improvements

* Apply suggestions

* Apply suggestions

* Add integration test

* Update image URL

* Add integration test

* Fix model_type

* Update style

* Improve docs

* Add doc tests

* Fix copies

* Remove tests which are passing

* Improve some more

* Add tests for seq2seq language models

* Minor fix

* Convert more checkpoints

* finalize CI

* Fix blip and blip2 processors

* add `accelerate` support for `blip2`

* clean up

* make style

* Update conversion script

* Update conversion script some more

* Update organization

* revert toc file

* add blip-2 to toc file

* Some more improvements

* Fix docstring

* Improve docs

---------

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
Co-authored-by: younesbelkada <younesbelkada@gmail.com>

utils/check_repo.py

@@ -53,6 +53,7 @@ PRIVATE_MODELS = [
 # Being in this list is an exception and should **not** be the rule.
 IGNORE_NON_TESTED = PRIVATE_MODELS.copy() + [
     # models to ignore for not tested
+    "Blip2QFormerModel",  # Building part of bigger (tested) model.
     "DetaEncoder",  # Building part of bigger (tested) model.
     "DetaDecoder",  # Building part of bigger (tested) model.
     "GraphormerEncoder",  # Building part of bigger (tested) model.
@@ -171,6 +172,9 @@ TEST_FILES_WITH_NO_COMMON_TESTS = [
 # should **not** be the rule.
 IGNORE_NON_AUTO_CONFIGURED = PRIVATE_MODELS.copy() + [
     # models to ignore for model xxx mapping
+    "Blip2ForConditionalGeneration",
+    "Blip2QFormerModel",
+    "Blip2VisionModel",
     "GitVisionModel",
     "GraphormerModel",
     "GraphormerForGraphClassification",

utils/documentation_tests.txt

@@ -35,6 +35,7 @@ src/transformers/models/blenderbot/configuration_blenderbot.py
 src/transformers/models/blenderbot/modeling_blenderbot.py
 src/transformers/models/blenderbot_small/configuration_blenderbot_small.py
 src/transformers/models/blenderbot_small/modeling_blenderbot_small.py
+src/transformers/models/blip_2/modeling_blip_2.py
 src/transformers/models/blip/modeling_blip.py
 src/transformers/models/bloom/configuration_bloom.py
 src/transformers/models/camembert/configuration_camembert.py