Skip to content
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.

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

Add DBRX Model #29921

Merged
merged 137 commits into from
Apr 18, 2024
Merged

Add DBRX Model #29921

merged 137 commits into from
Apr 18, 2024

Conversation

abhi-mosaic
Copy link
Contributor

What does this PR do?

Add support for DbrxConfig, DbrxModel, and DbrxForCausalLM.

https://huggingface.co/databricks/dbrx-base
https://huggingface.co/databricks/dbrx-instruct

Before submitting

  • This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
  • Did you read the contributor guideline,
    Pull Request section?
  • Was this discussed/approved via a Github issue or the forum? Please add a link
    to it if that's the case.
  • Did you make sure to update the documentation with your changes? Here are the
    documentation guidelines, and
    here are tips on formatting docstrings.
  • Did you write any new necessary tests?

Who can review?

Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.

@giyaseddin
Copy link

Duplicate work here too #29910

Copy link
Collaborator

@ArthurZucker ArthurZucker left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Great work already 🔥
Would recommend you to use transfomers-cli add-new-model-like and use mixtral template, because that will fill all the necessary configuration_auto and etc for you.
Also make sure to run make fixup for code formatting, and make fix-copies to make is overwritten with the copied from

src/transformers/models/dbrx/modeling_dbrx.py Show resolved Hide resolved
src/transformers/models/dbrx/modeling_dbrx.py Show resolved Hide resolved
src/transformers/models/dbrx/modeling_dbrx.py Show resolved Hide resolved
src/transformers/models/dbrx/modeling_dbrx.py Show resolved Hide resolved
src/transformers/models/dbrx/modeling_dbrx.py Outdated Show resolved Hide resolved
tests/models/dbrx/test_modeling_dbrx.py Show resolved Hide resolved


@require_torch
class DbrxModelIntegrationTest(unittest.TestCase):
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

we need generation tests as with a tiny model from both lib for the logits! Otherwise it's too big for our CI

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Out of curiosity, what is the biggest size model your CI can handle?

I was looking at the tests for falcon and it looks like they test on a 7B and 40B "tiny" model.

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We can handle 7B in float16 so ~16GB. Tiny models are way smaller.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Hi @ArthurZucker, whats the TODO here? Do we still need to build a small checkpoint or is it ok to leave this test as is with the @slow decorator?

src/transformers/models/dbrx/modeling_dbrx.py Outdated Show resolved Hide resolved
@eitanturok eitanturok mentioned this pull request Mar 28, 2024
5 tasks
@Rocketknight1
Copy link
Member

Hey all! I'll be handling reviews for this model from here - it looks like you have enough to work with for now, but if you get stuck at any point, or you'd like help, please let me know! We're quite excited to get DBRX ported, so don't be shy about reaching out. You can ping me here, or I'm also Matt in the collaboration channel on Slack

Copy link
Member

@Rocketknight1 Rocketknight1 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Hey - DbrxBlock shouldn't be imported, because we generally don't make those internal layers available in the top-level namespace! This is also why the doctest checker is complaining - those errors should clear up if we remove these imports.

src/transformers/__init__.py Outdated Show resolved Hide resolved
src/transformers/__init__.py Outdated Show resolved Hide resolved
src/transformers/models/dbrx/__init__.py Outdated Show resolved Hide resolved
src/transformers/models/dbrx/__init__.py Outdated Show resolved Hide resolved
Copy link
Collaborator

@amyeroberts amyeroberts left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Beautiful 🤩

@amyeroberts
Copy link
Collaborator

The failing tests are quite weird and possible connection errors. I'm rerunning on CI which should hopefully resolve 🤞

@eitanturok
Copy link
Contributor

Addressed your comments! Let's do the final push and get this in!

README.md Outdated
@@ -341,6 +341,8 @@ Current number of checkpoints: ![](https://img.shields.io/endpoint?url=https://h
1. **[CTRL](https://huggingface.co/docs/transformers/model_doc/ctrl)** (from Salesforce) released with the paper [CTRL: A Conditional Transformer Language Model for Controllable Generation](https://arxiv.org/abs/1909.05858) by Nitish Shirish Keskar*, Bryan McCann*, Lav R. Varshney, Caiming Xiong and Richard Socher.
1. **[CvT](https://huggingface.co/docs/transformers/model_doc/cvt)** (from Microsoft) released with the paper [CvT: Introducing Convolutions to Vision Transformers](https://arxiv.org/abs/2103.15808) by Haiping Wu, Bin Xiao, Noel Codella, Mengchen Liu, Xiyang Dai, Lu Yuan, Lei Zhang.
1. **[Data2Vec](https://huggingface.co/docs/transformers/model_doc/data2vec)** (from Facebook) released with the paper [Data2Vec: A General Framework for Self-supervised Learning in Speech, Vision and Language](https://arxiv.org/abs/2202.03555) by Alexei Baevski, Wei-Ning Hsu, Qiantong Xu, Arun Babu, Jiatao Gu, Michael Auli.

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

quality checks are currently failing on the models_in_readme check. I suspect the extra newline might be to blame here

Suggested change

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yep, that was the error. Pushing now.

Copy link
Collaborator

@amyeroberts amyeroberts left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Only thing left is resolving the conflicts with main for docs/source/en/perf_infer_gpu_one.md and docs/source/en/tasks/language_modeling.md.

As the tests are looking good -- quality checks now pass -- I'd recommend resolving the conflicts before the CI finished. The test suite is unfortunately pretty slow and we can be pretty confident the tests should all pass, as they were two commits ago and the changes have been minimal

@eitanturok
Copy link
Contributor

Only thing left is resolving the conflicts with main for docs/source/en/perf_infer_gpu_one.md and docs/source/en/tasks/language_modeling.md.

As the tests are looking good -- quality checks now pass -- I'd recommend resolving the conflicts before the CI finished. The test suite is unfortunately pretty slow and we can be pretty confident the tests should all pass, as they were two commits ago and the changes have been minimal

Yes, I was just hoping on the train. I'm taking care of this now.

@eitanturok
Copy link
Contributor

Just merged the conflicts. Now we just let the tests run and hope for the best!

@LysandreJik
Copy link
Member

Awesome! Thanks all, merging!

@LysandreJik LysandreJik merged commit 005b957 into huggingface:main Apr 18, 2024
22 checks passed
@eitanturok
Copy link
Contributor

As a follow up, when can we expect to see DBRX updated in the transformers documentation?

@Rocketknight1
Copy link
Member

Rocketknight1 commented Apr 18, 2024

@eitanturok the documentation is automatically generated from the docstrings / .md files in this PR, so you should see it once the release goes out!

@eitanturok
Copy link
Contributor

Understood. And the release goes out later today, correct?

@Rocketknight1
Copy link
Member

Yes! The release has just gone out here. The frontend docs will be updated as soon as the doc builder finishes, which will happen later today. In the meantime, you can preview the docs on the main doc branch here.

zucchini-nlp pushed a commit to zucchini-nlp/transformers that referenced this pull request Apr 18, 2024
* wip

* fix __init__.py

* add docs

* Apply suggestions from code review

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* address comments 1

* work on make fixup

* pass configs down

* add sdpa attention

* remove DbrxBlock

* add to configuration_auto

* docstring now passes formatting test

* fix style

* update READMEs

* add dbrx to modeling_auto

* make fix-copies generated this

* add DBRX_PRETRAINED_CONFIG_ARCHIVE_MAP

* config docstring passes formatting test

* rename moe_loss_weight to router_aux_loss_coef

* add to flash-attn documentation

* fix model-path in tests

* Explicitly make `"suli"` the default `ffn_act_fn`

Co-authored-by: Wing Lian <wing.lian@gmail.com>

* default to using router_aux_loss_coef over ffn_config[moe_loss_weight]

* fix _flash_attn_uses_top_left_mask and is_causal

* fix tests path

* don't use token type IDs

* follow Llama and remove token_type_ids from test

* init ConfigTester differently so tests pass

* remove multiple choice test

* remove question + answer test

* remove sequence classification test

* remove token classification test

* copy Llama tests and remove token_type_ids from test inputs

* do not test pruning or headmasking; style code

* add _tied_weights_keys parameter to pass test

* add type hints

* fix type check

* update config tester

* remove masked_lm test

* remove encoder tests

* initialize DbrxModelTester with correct params

* style

* torch_dtype does not rely on torch

* run make fixup, fix-copies

* use https://huggingface.co/v2ray/dbrx-base-fixed/blob/main/modeling_dbrx.py

* add copyright info

* fix imports and DbrxRotaryEmbedding

* update DbrxModel docstring

* use copies

* change model path in docstring

* use config in DbrxFFN

* fix flashattention2, sdpaattention

* input config to DbrXAttention, DbrxNormAttentionNorm

* more fixes

* fix

* fix again!

* add informative comment

* fix ruff?

* remove print statement + style

* change doc-test

* fix doc-test

* fix docstring

* delete commented out text

* make defaults match dbrx-instruct

* replace `router_aux_loss_coef` with `moe_loss_weight`

* is_decoder=True

* remove is_decoder from configtester

* implement sdpa properly

* make is_decoder pass tests

* start on the GenerationTesterMixin tests

* add dbrx to sdpa documentation

* skip weight typing test

* style

* initialize smaller model

Co-authored-by: Matt <Rocketknight1@users.noreply.github.com>

* Add DBRX to toctree

* skip test_new_cache_format

* make config defaults smaller again

* add pad_token_id

* remove pad_token_id from config

* Remove all references to DBRX_PRETRAINED_CONFIG_ARCHIVE_MAP

* Update src/transformers/models/dbrx/__init__.py

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* Update src/transformers/models/dbrx/modeling_dbrx.py

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* Update docs/source/en/model_doc/dbrx.md

Co-authored-by: Matt <Rocketknight1@users.noreply.github.com>

* Update src/transformers/models/dbrx/configuration_dbrx.py

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* Update docs/source/en/model_doc/dbrx.md

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* fix typo

* Apply suggestions from code review

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* update docs, fix configuration_auto.py

* address pr comments

* remove is_decoder flag

* slice

* fix requires grad

* remove grad

* disconnect differently

* remove grad

* enable grads

* patch

* detach expert

* nissan al ghaib

* Update modeling_dbrx.py

* Update src/transformers/models/dbrx/modeling_dbrx.py

Co-authored-by: Matt <Rocketknight1@users.noreply.github.com>

* replace "Gemma" with "Dbrx"

* remove # type: ignore

* don't hardcode vocab_size

* remove ToDo

* Re-add removed idefics2 line

* Update test to use tiny-random!

* Remove TODO

* Remove one more case of loading the entire dbrx-instruct in the tests

* Update src/transformers/models/dbrx/modeling_dbrx.py

Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* address some comments

* small model

* add dbrx to tokenization_auto

* More docstrings with add_start_docstrings

* Dbrx for now

* add PipelineTesterMixin

* Update src/transformers/models/dbrx/configuration_dbrx.py

Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* remove flash-attn2 import error

* fix docstring

Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* add useage example

* put on one line

Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* fix ffn_act_fn

Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* change "dbrx" to "DBRX" for display purposes.

* fix __init__.py?

* fix __init__.py

* fix README

* return the aux_loss

* remove extra spaces

* fix configuration_auto.py

* fix format in tokenization_auto

* remove new line

* add more useage examples

---------

Co-authored-by: Abhi Venigalla <abhi.venigalla@databricks.com>
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
Co-authored-by: Eitan Turok <eitan.turok@databricks.com>
Co-authored-by: Eitan Turok <150733043+eitanturok@users.noreply.github.com>
Co-authored-by: Wing Lian <wing.lian@gmail.com>
Co-authored-by: Eitan Turok <eitanturok@gmail.com>
Co-authored-by: Matt <Rocketknight1@users.noreply.github.com>
Co-authored-by: Matt <rocketknight1@gmail.com>
Co-authored-by: Your Name <you@example.com>
Co-authored-by: Mihir Patel <mihir.v.patel7@gmail.com>
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>
ArthurZucker added a commit that referenced this pull request Apr 22, 2024
* wip

* fix __init__.py

* add docs

* Apply suggestions from code review

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* address comments 1

* work on make fixup

* pass configs down

* add sdpa attention

* remove DbrxBlock

* add to configuration_auto

* docstring now passes formatting test

* fix style

* update READMEs

* add dbrx to modeling_auto

* make fix-copies generated this

* add DBRX_PRETRAINED_CONFIG_ARCHIVE_MAP

* config docstring passes formatting test

* rename moe_loss_weight to router_aux_loss_coef

* add to flash-attn documentation

* fix model-path in tests

* Explicitly make `"suli"` the default `ffn_act_fn`

Co-authored-by: Wing Lian <wing.lian@gmail.com>

* default to using router_aux_loss_coef over ffn_config[moe_loss_weight]

* fix _flash_attn_uses_top_left_mask and is_causal

* fix tests path

* don't use token type IDs

* follow Llama and remove token_type_ids from test

* init ConfigTester differently so tests pass

* remove multiple choice test

* remove question + answer test

* remove sequence classification test

* remove token classification test

* copy Llama tests and remove token_type_ids from test inputs

* do not test pruning or headmasking; style code

* add _tied_weights_keys parameter to pass test

* add type hints

* fix type check

* update config tester

* remove masked_lm test

* remove encoder tests

* initialize DbrxModelTester with correct params

* style

* torch_dtype does not rely on torch

* run make fixup, fix-copies

* use https://huggingface.co/v2ray/dbrx-base-fixed/blob/main/modeling_dbrx.py

* add copyright info

* fix imports and DbrxRotaryEmbedding

* update DbrxModel docstring

* use copies

* change model path in docstring

* use config in DbrxFFN

* fix flashattention2, sdpaattention

* input config to DbrXAttention, DbrxNormAttentionNorm

* more fixes

* fix

* fix again!

* add informative comment

* fix ruff?

* remove print statement + style

* change doc-test

* fix doc-test

* fix docstring

* delete commented out text

* make defaults match dbrx-instruct

* replace `router_aux_loss_coef` with `moe_loss_weight`

* is_decoder=True

* remove is_decoder from configtester

* implement sdpa properly

* make is_decoder pass tests

* start on the GenerationTesterMixin tests

* add dbrx to sdpa documentation

* skip weight typing test

* style

* initialize smaller model

Co-authored-by: Matt <Rocketknight1@users.noreply.github.com>

* Add DBRX to toctree

* skip test_new_cache_format

* make config defaults smaller again

* add pad_token_id

* remove pad_token_id from config

* Remove all references to DBRX_PRETRAINED_CONFIG_ARCHIVE_MAP

* Update src/transformers/models/dbrx/__init__.py

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* Update src/transformers/models/dbrx/modeling_dbrx.py

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* Update docs/source/en/model_doc/dbrx.md

Co-authored-by: Matt <Rocketknight1@users.noreply.github.com>

* Update src/transformers/models/dbrx/configuration_dbrx.py

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* Update docs/source/en/model_doc/dbrx.md

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* fix typo

* Apply suggestions from code review

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* update docs, fix configuration_auto.py

* address pr comments

* remove is_decoder flag

* slice

* fix requires grad

* remove grad

* disconnect differently

* remove grad

* enable grads

* patch

* detach expert

* nissan al ghaib

* Update modeling_dbrx.py

* Update src/transformers/models/dbrx/modeling_dbrx.py

Co-authored-by: Matt <Rocketknight1@users.noreply.github.com>

* replace "Gemma" with "Dbrx"

* remove # type: ignore

* don't hardcode vocab_size

* remove ToDo

* Re-add removed idefics2 line

* Update test to use tiny-random!

* Remove TODO

* Remove one more case of loading the entire dbrx-instruct in the tests

* Update src/transformers/models/dbrx/modeling_dbrx.py

Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* address some comments

* small model

* add dbrx to tokenization_auto

* More docstrings with add_start_docstrings

* Dbrx for now

* add PipelineTesterMixin

* Update src/transformers/models/dbrx/configuration_dbrx.py

Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* remove flash-attn2 import error

* fix docstring

Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* add useage example

* put on one line

Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* fix ffn_act_fn

Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* change "dbrx" to "DBRX" for display purposes.

* fix __init__.py?

* fix __init__.py

* fix README

* return the aux_loss

* remove extra spaces

* fix configuration_auto.py

* fix format in tokenization_auto

* remove new line

* add more useage examples

---------

Co-authored-by: Abhi Venigalla <abhi.venigalla@databricks.com>
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
Co-authored-by: Eitan Turok <eitan.turok@databricks.com>
Co-authored-by: Eitan Turok <150733043+eitanturok@users.noreply.github.com>
Co-authored-by: Wing Lian <wing.lian@gmail.com>
Co-authored-by: Eitan Turok <eitanturok@gmail.com>
Co-authored-by: Matt <Rocketknight1@users.noreply.github.com>
Co-authored-by: Matt <rocketknight1@gmail.com>
Co-authored-by: Your Name <you@example.com>
Co-authored-by: Mihir Patel <mihir.v.patel7@gmail.com>
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>
ydshieh pushed a commit that referenced this pull request Apr 23, 2024
* wip

* fix __init__.py

* add docs

* Apply suggestions from code review

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* address comments 1

* work on make fixup

* pass configs down

* add sdpa attention

* remove DbrxBlock

* add to configuration_auto

* docstring now passes formatting test

* fix style

* update READMEs

* add dbrx to modeling_auto

* make fix-copies generated this

* add DBRX_PRETRAINED_CONFIG_ARCHIVE_MAP

* config docstring passes formatting test

* rename moe_loss_weight to router_aux_loss_coef

* add to flash-attn documentation

* fix model-path in tests

* Explicitly make `"suli"` the default `ffn_act_fn`

Co-authored-by: Wing Lian <wing.lian@gmail.com>

* default to using router_aux_loss_coef over ffn_config[moe_loss_weight]

* fix _flash_attn_uses_top_left_mask and is_causal

* fix tests path

* don't use token type IDs

* follow Llama and remove token_type_ids from test

* init ConfigTester differently so tests pass

* remove multiple choice test

* remove question + answer test

* remove sequence classification test

* remove token classification test

* copy Llama tests and remove token_type_ids from test inputs

* do not test pruning or headmasking; style code

* add _tied_weights_keys parameter to pass test

* add type hints

* fix type check

* update config tester

* remove masked_lm test

* remove encoder tests

* initialize DbrxModelTester with correct params

* style

* torch_dtype does not rely on torch

* run make fixup, fix-copies

* use https://huggingface.co/v2ray/dbrx-base-fixed/blob/main/modeling_dbrx.py

* add copyright info

* fix imports and DbrxRotaryEmbedding

* update DbrxModel docstring

* use copies

* change model path in docstring

* use config in DbrxFFN

* fix flashattention2, sdpaattention

* input config to DbrXAttention, DbrxNormAttentionNorm

* more fixes

* fix

* fix again!

* add informative comment

* fix ruff?

* remove print statement + style

* change doc-test

* fix doc-test

* fix docstring

* delete commented out text

* make defaults match dbrx-instruct

* replace `router_aux_loss_coef` with `moe_loss_weight`

* is_decoder=True

* remove is_decoder from configtester

* implement sdpa properly

* make is_decoder pass tests

* start on the GenerationTesterMixin tests

* add dbrx to sdpa documentation

* skip weight typing test

* style

* initialize smaller model

Co-authored-by: Matt <Rocketknight1@users.noreply.github.com>

* Add DBRX to toctree

* skip test_new_cache_format

* make config defaults smaller again

* add pad_token_id

* remove pad_token_id from config

* Remove all references to DBRX_PRETRAINED_CONFIG_ARCHIVE_MAP

* Update src/transformers/models/dbrx/__init__.py

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* Update src/transformers/models/dbrx/modeling_dbrx.py

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* Update docs/source/en/model_doc/dbrx.md

Co-authored-by: Matt <Rocketknight1@users.noreply.github.com>

* Update src/transformers/models/dbrx/configuration_dbrx.py

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* Update docs/source/en/model_doc/dbrx.md

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* fix typo

* Apply suggestions from code review

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* update docs, fix configuration_auto.py

* address pr comments

* remove is_decoder flag

* slice

* fix requires grad

* remove grad

* disconnect differently

* remove grad

* enable grads

* patch

* detach expert

* nissan al ghaib

* Update modeling_dbrx.py

* Update src/transformers/models/dbrx/modeling_dbrx.py

Co-authored-by: Matt <Rocketknight1@users.noreply.github.com>

* replace "Gemma" with "Dbrx"

* remove # type: ignore

* don't hardcode vocab_size

* remove ToDo

* Re-add removed idefics2 line

* Update test to use tiny-random!

* Remove TODO

* Remove one more case of loading the entire dbrx-instruct in the tests

* Update src/transformers/models/dbrx/modeling_dbrx.py

Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* address some comments

* small model

* add dbrx to tokenization_auto

* More docstrings with add_start_docstrings

* Dbrx for now

* add PipelineTesterMixin

* Update src/transformers/models/dbrx/configuration_dbrx.py

Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* remove flash-attn2 import error

* fix docstring

Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* add useage example

* put on one line

Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* fix ffn_act_fn

Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* change "dbrx" to "DBRX" for display purposes.

* fix __init__.py?

* fix __init__.py

* fix README

* return the aux_loss

* remove extra spaces

* fix configuration_auto.py

* fix format in tokenization_auto

* remove new line

* add more useage examples

---------

Co-authored-by: Abhi Venigalla <abhi.venigalla@databricks.com>
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
Co-authored-by: Eitan Turok <eitan.turok@databricks.com>
Co-authored-by: Eitan Turok <150733043+eitanturok@users.noreply.github.com>
Co-authored-by: Wing Lian <wing.lian@gmail.com>
Co-authored-by: Eitan Turok <eitanturok@gmail.com>
Co-authored-by: Matt <Rocketknight1@users.noreply.github.com>
Co-authored-by: Matt <rocketknight1@gmail.com>
Co-authored-by: Your Name <you@example.com>
Co-authored-by: Mihir Patel <mihir.v.patel7@gmail.com>
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>
itazap pushed a commit that referenced this pull request May 14, 2024
* wip

* fix __init__.py

* add docs

* Apply suggestions from code review

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* address comments 1

* work on make fixup

* pass configs down

* add sdpa attention

* remove DbrxBlock

* add to configuration_auto

* docstring now passes formatting test

* fix style

* update READMEs

* add dbrx to modeling_auto

* make fix-copies generated this

* add DBRX_PRETRAINED_CONFIG_ARCHIVE_MAP

* config docstring passes formatting test

* rename moe_loss_weight to router_aux_loss_coef

* add to flash-attn documentation

* fix model-path in tests

* Explicitly make `"suli"` the default `ffn_act_fn`

Co-authored-by: Wing Lian <wing.lian@gmail.com>

* default to using router_aux_loss_coef over ffn_config[moe_loss_weight]

* fix _flash_attn_uses_top_left_mask and is_causal

* fix tests path

* don't use token type IDs

* follow Llama and remove token_type_ids from test

* init ConfigTester differently so tests pass

* remove multiple choice test

* remove question + answer test

* remove sequence classification test

* remove token classification test

* copy Llama tests and remove token_type_ids from test inputs

* do not test pruning or headmasking; style code

* add _tied_weights_keys parameter to pass test

* add type hints

* fix type check

* update config tester

* remove masked_lm test

* remove encoder tests

* initialize DbrxModelTester with correct params

* style

* torch_dtype does not rely on torch

* run make fixup, fix-copies

* use https://huggingface.co/v2ray/dbrx-base-fixed/blob/main/modeling_dbrx.py

* add copyright info

* fix imports and DbrxRotaryEmbedding

* update DbrxModel docstring

* use copies

* change model path in docstring

* use config in DbrxFFN

* fix flashattention2, sdpaattention

* input config to DbrXAttention, DbrxNormAttentionNorm

* more fixes

* fix

* fix again!

* add informative comment

* fix ruff?

* remove print statement + style

* change doc-test

* fix doc-test

* fix docstring

* delete commented out text

* make defaults match dbrx-instruct

* replace `router_aux_loss_coef` with `moe_loss_weight`

* is_decoder=True

* remove is_decoder from configtester

* implement sdpa properly

* make is_decoder pass tests

* start on the GenerationTesterMixin tests

* add dbrx to sdpa documentation

* skip weight typing test

* style

* initialize smaller model

Co-authored-by: Matt <Rocketknight1@users.noreply.github.com>

* Add DBRX to toctree

* skip test_new_cache_format

* make config defaults smaller again

* add pad_token_id

* remove pad_token_id from config

* Remove all references to DBRX_PRETRAINED_CONFIG_ARCHIVE_MAP

* Update src/transformers/models/dbrx/__init__.py

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* Update src/transformers/models/dbrx/modeling_dbrx.py

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* Update docs/source/en/model_doc/dbrx.md

Co-authored-by: Matt <Rocketknight1@users.noreply.github.com>

* Update src/transformers/models/dbrx/configuration_dbrx.py

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* Update docs/source/en/model_doc/dbrx.md

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* fix typo

* Apply suggestions from code review

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* update docs, fix configuration_auto.py

* address pr comments

* remove is_decoder flag

* slice

* fix requires grad

* remove grad

* disconnect differently

* remove grad

* enable grads

* patch

* detach expert

* nissan al ghaib

* Update modeling_dbrx.py

* Update src/transformers/models/dbrx/modeling_dbrx.py

Co-authored-by: Matt <Rocketknight1@users.noreply.github.com>

* replace "Gemma" with "Dbrx"

* remove # type: ignore

* don't hardcode vocab_size

* remove ToDo

* Re-add removed idefics2 line

* Update test to use tiny-random!

* Remove TODO

* Remove one more case of loading the entire dbrx-instruct in the tests

* Update src/transformers/models/dbrx/modeling_dbrx.py

Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* address some comments

* small model

* add dbrx to tokenization_auto

* More docstrings with add_start_docstrings

* Dbrx for now

* add PipelineTesterMixin

* Update src/transformers/models/dbrx/configuration_dbrx.py

Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* remove flash-attn2 import error

* fix docstring

Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* add useage example

* put on one line

Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* fix ffn_act_fn

Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* change "dbrx" to "DBRX" for display purposes.

* fix __init__.py?

* fix __init__.py

* fix README

* return the aux_loss

* remove extra spaces

* fix configuration_auto.py

* fix format in tokenization_auto

* remove new line

* add more useage examples

---------

Co-authored-by: Abhi Venigalla <abhi.venigalla@databricks.com>
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
Co-authored-by: Eitan Turok <eitan.turok@databricks.com>
Co-authored-by: Eitan Turok <150733043+eitanturok@users.noreply.github.com>
Co-authored-by: Wing Lian <wing.lian@gmail.com>
Co-authored-by: Eitan Turok <eitanturok@gmail.com>
Co-authored-by: Matt <Rocketknight1@users.noreply.github.com>
Co-authored-by: Matt <rocketknight1@gmail.com>
Co-authored-by: Your Name <you@example.com>
Co-authored-by: Mihir Patel <mihir.v.patel7@gmail.com>
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.