Commit Graph

121 Commits

Author SHA1 Message Date
OlivierDehaene f59fb8b630
feat(router): add ngrok integration (#453) 2023-06-16 16:25:11 +02:00
OlivierDehaene 5ce89059f8
feat(server): pre-allocate past key values for flash causal LM (#412) 2023-06-12 18:30:29 +02:00
OlivierDehaene e496c9ba5b
feat(server): optimize dist ops (#434) 2023-06-09 11:55:29 +02:00
Nicolas Patry abd58ff82c
feat(server): Rework model loading (#344)
# What does this PR do?

Reworked the loading logic. Idea is to use cleaner loading code:

- Remove need for `no_init_weights`
- Remove all weird `bnb_linear` and `load_weights` and
`post_load_weights`.

New code layout:

- New class `Weights` in charge of handling loading the weights from
multiple files into appropiate tensors (potentially sharded)
- TP layers now are "shells", they contain the code to know what kind of
sharding we need + eventual `all_reduce`. They do not inherit from
linear, but they contain some kind of Linear instead
- the contained linear can be either FastLinear, BnbLinear or GPTq
Linear next.
- All modeling code is explictly made for sharding, process group is
just no-ops for non sharded code (removes a lot of test cases)

![Screenshot from 2023-05-19
23-19-59](https://github.com/huggingface/text-generation-inference/assets/204321/9a802654-74a3-488c-87a8-073743a6143f)

---------

Co-authored-by: Ubuntu <ubuntu@ip-172-31-41-161.taildb5d.ts.net>
Co-authored-by: Ubuntu <ubuntu@ip-172-31-41-161.ec2.internal>
Co-authored-by: OlivierDehaene <olivier@huggingface.co>
Co-authored-by: OlivierDehaene <23298448+OlivierDehaene@users.noreply.github.com>
2023-06-08 14:51:52 +02:00
OlivierDehaene 6abec14a7e
feat(server): batch tokenization for flash causal lm (#411) 2023-06-05 16:09:41 +02:00
OlivierDehaene 895c5f1562
feat(server): only compute prefill logprobs when asked (#406)
Close #288
2023-06-02 17:12:30 +02:00
OlivierDehaene e7248fe90e v0.8.2 2023-06-01 19:49:13 +02:00
OlivierDehaene 95d3546976
feat(server): load santacoder/starcoder models with safetensors (#393)
Fix #366
2023-06-01 12:10:35 +02:00
OlivierDehaene c0928e6f26
feat(server): remove trust_remote_code requirement for falcon models (#396) 2023-06-01 12:07:41 +02:00
OlivierDehaene d69a0633be
fix(server): fix has_position_ids (#395)
Fix #389
2023-06-01 11:41:35 +02:00
OlivierDehaene 337afb2842
fix(server): fix bnb quantization for CausalLM models (#385) 2023-05-31 11:48:28 +02:00
OlivierDehaene 87dc034b59
feat(server): add retry on download (#384) 2023-05-31 10:57:53 +02:00
OlivierDehaene b8b950b37c
feat(server): support RefinedWeb models (#379) 2023-05-30 18:25:19 +02:00
OlivierDehaene bf7f1d5434 fix(server): fix quantization 2023-05-30 13:56:03 +02:00
CL-Shang 5fde8d9991
Fix issue when load AutoModelForSeq2SeqLM model (#370) 2023-05-26 12:31:47 +02:00
OlivierDehaene 62f91f78ac
feat(server): support vectorized warpers in flash causal lm (#317)
Co-authored-by: Joel Lamy-Poirier <joel.lamy-poirier@servicenow.com>
2023-05-26 12:30:27 +02:00
OlivierDehaene 218c9adaa5
feat: decrease IPC proto size (#367)
Closes #307 #308
2023-05-24 19:19:57 +02:00
OlivierDehaene e3e487dc71
feat(server): support trust_remote_code (#363) 2023-05-23 20:40:39 +02:00
OlivierDehaene e9669a4085
feat(server): do not use device_map auto on single GPU (#362) 2023-05-23 19:12:12 +02:00
OlivierDehaene cfaa858070
feat(server): support fp16 for t5 (#360)
Fixes #349
2023-05-23 18:16:48 +02:00
OlivierDehaene 4f4c9c1665
fix(server): t5 cannot run in f16 (#356)
Fix #349
2023-05-23 12:15:54 +02:00
OlivierDehaene 91d9beec90
fix(server): fix init for flash causal lm (#352)
Fixes #347
2023-05-22 15:05:32 +02:00
OlivierDehaene e649bf9a55
feat(server): Support BLOOMChat-176B (#348) (#351)
@njhill, 
temporary workaround to be able to run our CI as secrets are not
available to runners run by external contributors. I will ask around to
see if there is a better way.

Co-authored-by: Nick Hill <nickhill@us.ibm.com>
2023-05-22 13:36:00 +02:00
OlivierDehaene 5a58226130
fix(server): fix decode token (#334)
Fixes #333

---------

Co-authored-by: Nicolas Patry <patry.nicolas@protonmail.com>
2023-05-16 23:23:27 +02:00
OlivierDehaene e71471bec9
feat: add snapshot testing (#282) 2023-05-15 23:36:30 +02:00
Nicolas Patry f58f0a0364
Single place for TP layers + Dropout Layer Norm + FastLinear (#329)
# What does this PR do?

<!--
Congratulations! You've made it this far! You're not quite done yet
though.

Once merged, your PR is going to appear in the release notes with the
title you set, so make sure it's a great title that fully reflects the
extent of your awesome contribution.

Then, please replace this with a description of the change and which
issue is fixed (if applicable). Please also include relevant motivation
and context. List any dependencies (if any) that are required for this
change.

Once you're done, someone will review your PR shortly (see the section
"Who can review?" below to tag some potential reviewers). They may
suggest changes to make the code even better. If no one reviewed your PR
after a week has passed, don't hesitate to post a new comment
@-mentioning the same persons---sometimes notifications get lost.
-->

<!-- Remove if not applicable -->

Fixes # (issue)


## 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](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#start-contributing-pull-requests),
      Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the
[forum](https://discuss.huggingface.co/)? 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](https://github.com/huggingface/transformers/tree/main/docs),
and
[here are tips on formatting
docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [ ] 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.

<!-- Your PR will be replied to more quickly if you can figure out the
right person to tag with @


@OlivierDehaene OR @Narsil

 -->
2023-05-15 17:30:47 +02:00
Nicolas Patry d7a97aa0b6
Removing dead variables. (#327)
# What does this PR do?

<!--
Congratulations! You've made it this far! You're not quite done yet
though.

Once merged, your PR is going to appear in the release notes with the
title you set, so make sure it's a great title that fully reflects the
extent of your awesome contribution.

Then, please replace this with a description of the change and which
issue is fixed (if applicable). Please also include relevant motivation
and context. List any dependencies (if any) that are required for this
change.

Once you're done, someone will review your PR shortly (see the section
"Who can review?" below to tag some potential reviewers). They may
suggest changes to make the code even better. If no one reviewed your PR
after a week has passed, don't hesitate to post a new comment
@-mentioning the same persons---sometimes notifications get lost.
-->

<!-- Remove if not applicable -->

Fixes # (issue)


## 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](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#start-contributing-pull-requests),
      Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the
[forum](https://discuss.huggingface.co/)? 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](https://github.com/huggingface/transformers/tree/main/docs),
and
[here are tips on formatting
docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [ ] 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.

<!-- Your PR will be replied to more quickly if you can figure out the
right person to tag with @


@OlivierDehaene OR @Narsil

 -->
2023-05-15 15:14:17 +02:00
Nicolas Patry 91e674bb85
Lifting check_unitialized. (#325)
# What does this PR do?

Lifting check_unitialized.

<!--
Congratulations! You've made it this far! You're not quite done yet
though.

Once merged, your PR is going to appear in the release notes with the
title you set, so make sure it's a great title that fully reflects the
extent of your awesome contribution.

Then, please replace this with a description of the change and which
issue is fixed (if applicable). Please also include relevant motivation
and context. List any dependencies (if any) that are required for this
change.

Once you're done, someone will review your PR shortly (see the section
"Who can review?" below to tag some potential reviewers). They may
suggest changes to make the code even better. If no one reviewed your PR
after a week has passed, don't hesitate to post a new comment
@-mentioning the same persons---sometimes notifications get lost.
-->

<!-- Remove if not applicable -->

Fixes # (issue)


## 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](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#start-contributing-pull-requests),
      Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the
[forum](https://discuss.huggingface.co/)? 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](https://github.com/huggingface/transformers/tree/main/docs),
and
[here are tips on formatting
docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [ ] 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.

<!-- Your PR will be replied to more quickly if you can figure out the
right person to tag with @


@OlivierDehaene OR @Narsil

 -->
2023-05-15 11:32:25 +02:00
Nicolas Patry 73d84c6ee5
Hotfixes for santacoder/bigcode. (#294)
# What does this PR do?

Hotfixes:

- Uses `model_type`=`gpt_bigcode` for more general usage.
- Hotfixes linked lm_head vs wte_embedding (safetensors file do not
contain the key, correctly when the file is sharded, where as pytorch
copies the tensor)


<!--
Congratulations! You've made it this far! You're not quite done yet
though.

Once merged, your PR is going to appear in the release notes with the
title you set, so make sure it's a great title that fully reflects the
extent of your awesome contribution.

Then, please replace this with a description of the change and which
issue is fixed (if applicable). Please also include relevant motivation
and context. List any dependencies (if any) that are required for this
change.

Once you're done, someone will review your PR shortly (see the section
"Who can review?" below to tag some potential reviewers). They may
suggest changes to make the code even better. If no one reviewed your PR
after a week has passed, don't hesitate to post a new comment
@-mentioning the same persons---sometimes notifications get lost.
-->

<!-- Remove if not applicable -->

Fixes # (issue)


## 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](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#start-contributing-pull-requests),
      Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the
[forum](https://discuss.huggingface.co/)? 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](https://github.com/huggingface/transformers/tree/main/docs),
and
[here are tips on formatting
docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [ ] 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.

<!-- Your PR will be replied to more quickly if you can figure out the
right person to tag with @


@OlivierDehaene OR @Narsil

 -->

---------

Co-authored-by: Ubuntu <ubuntu@ip-172-31-41-161.ec2.internal>
Co-authored-by: OlivierDehaene <olivier@huggingface.co>
2023-05-15 10:35:20 +02:00
OlivierDehaene 8a8f43410d
chore(docker): use nvidia base image (#318) 2023-05-12 17:32:40 +02:00
Nicolas Patry 76a48cd365
feat(server): GPTQ quantization (step1) (#277)
Changes only the type from `bool` to `Option<Enum>` pretty much
everywhere.
- Use `Optional[str]` in Python (easier to manage than importing type
everywhere). Except for the cli to get proper validation
- Updated all models to handle gracefully new values. (Error out if
unknown value, or gptq since not implemented).

<!--
Congratulations! You've made it this far! You're not quite done yet
though.

Once merged, your PR is going to appear in the release notes with the
title you set, so make sure it's a great title that fully reflects the
extent of your awesome contribution.

Then, please replace this with a description of the change and which
issue is fixed (if applicable). Please also include relevant motivation
and context. List any dependencies (if any) that are required for this
change.

Once you're done, someone will review your PR shortly (see the section
"Who can review?" below to tag some potential reviewers). They may
suggest changes to make the code even better. If no one reviewed your PR
after a week has passed, don't hesitate to post a new comment
@-mentioning the same persons---sometimes notifications get lost.
-->

<!-- Remove if not applicable -->

Fixes # (issue)


## 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](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#start-contributing-pull-requests),
      Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the
[forum](https://discuss.huggingface.co/)? 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](https://github.com/huggingface/transformers/tree/main/docs),
and
[here are tips on formatting
docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [ ] 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.

<!-- Your PR will be replied to more quickly if you can figure out the
right person to tag with @


@OlivierDehaene OR @Narsil

 -->
2023-05-12 14:46:41 +02:00
OlivierDehaene 745f596c88
feat(server): use float16 (#304) 2023-05-10 15:51:10 +02:00
OlivierDehaene 68e9d6ab33
feat(server): shard token decode (#303) 2023-05-10 15:48:21 +02:00
OlivierDehaene ad66f6ef9a
feat(server): optim flash causal lm decode_token (#285) 2023-05-09 18:26:19 +02:00
OlivierDehaene 85aa7e2e7b
feat(server): support hf endpoint weight layout (#266) 2023-05-03 11:36:24 +02:00
OlivierDehaene 4096000e34
fix(server): fix typo in tokenizers decode (#269)
closes #268
2023-05-03 10:10:34 +02:00
Ehsan M. Kermani f092ba9b22
feat(server): add watermarking tests (#248) 2023-04-27 19:16:35 +02:00
Nick Hill 34bca0b8d3
fix(server): Small tidy of code from recent changes (#251)
remaining_decode_tokens was calculated twice in Seq2SeqLMBatch.filter()
2023-04-27 09:57:28 +02:00
Nick Hill b4cf832c40
fix(server): fix reshaping of bloom past_key_values in concatenate() (#252)
Introduced in #214 

Fixes #249
2023-04-27 09:51:27 +02:00
OlivierDehaene ebc74d5666
feat(router): use number of tokens in batch as input for dynamic batching (#226)
Co-authored-by: Nick Hill <nickhill@us.ibm.com>
2023-04-24 17:59:00 +02:00
Nick Hill 4a7dd4085a
feat(server): reduce memory requirement (#214) 2023-04-24 14:15:42 +02:00
OlivierDehaene 4b460e72fb
fix(server): fix flash batch filtering (#220) 2023-04-21 20:26:01 +02:00
OlivierDehaene 1ffea36ec2
fix(server): fix flash causal (#219) 2023-04-21 19:49:08 +02:00
OlivierDehaene 86bca365df
fix(server): fix flash causal (#218) 2023-04-21 19:42:16 +02:00
OlivierDehaene afc5b999d0
fix(server): cleanup new flash past_key_values logic (#217) 2023-04-21 16:19:04 +02:00
OlivierDehaene db4cb5e4ed
fix(server): fix past key values logic (#216)
@njhill fyi
2023-04-21 15:59:18 +02:00
OlivierDehaene 343437c7b5
feat(router): add device and dtype info (#215) 2023-04-21 15:36:29 +02:00
Nick Hill ac8c0f6fe4
feat(server): flash attention past key value optimizations (#213) 2023-04-21 14:57:18 +02:00
OlivierDehaene 709d8936f6
feat(router): drop requests when client closes the channel (#202) 2023-04-20 11:07:40 +02:00
OlivierDehaene b6ee0ec7b0
feat(router): add git sha to info route (#208) 2023-04-19 21:36:59 +02:00
OlivierDehaene a88c54bb4c
feat(server): check cuda capability when importing flash models (#201)
close #198
2023-04-19 12:52:37 +02:00
OlivierDehaene e14ae3b5e9
feat(server): support quantization for flash models (#200)
closes #197
2023-04-19 12:51:11 +02:00
OlivierDehaene 7a1ba58557
fix(docker): fix docker image dependencies (#187) 2023-04-17 00:26:47 +02:00
OlivierDehaene 880a76eed5
feat(server): support sharded santacoder (#167) 2023-04-12 17:18:08 +02:00
OlivierDehaene 5fa8ae041c
feat(server): optimize decode for sane tokenizers (#170) 2023-04-12 12:03:10 +02:00
OlivierDehaene f26dfd0dc1
feat(server): support OPT models (#55)
OPT models do not all have a `tokenizer.json` file on the hub at the
moment. Can't merge for now.
2023-04-11 19:16:41 +02:00
OlivierDehaene 299217c95c
feat(server): add flash attention llama (#144) 2023-04-11 16:38:22 +02:00
OlivierDehaene 9987960062
feat(router): make router input validation optional (#164) 2023-04-09 20:22:27 +02:00
OlivierDehaene 3f2542bb6a
fix(server): fix escape characters in stop sequence (#155) 2023-04-05 19:37:41 +02:00
OlivierDehaene c0aeb32583
feat(server): flash santacoder (#153) 2023-04-03 19:06:42 +02:00
OlivierDehaene 08b7e4a282
fix(server): fix flash neox rotary embeddings (#150) 2023-03-30 16:12:23 +02:00
OlivierDehaene c9bdaa8b73
feat(server): reduce mlp and attn in one op for flash neox (#145) 2023-03-28 16:51:41 +02:00
Nick Hill 462530c2b0
fix(server): Avoid using try/except to determine kind of AutoModel (#142) 2023-03-27 09:23:22 +02:00
OlivierDehaene 678b2f3900
feat(server): cleanup flash neox loading (#139) 2023-03-26 16:37:21 +02:00
OlivierDehaene d6a93fe992
fix(server): fix flash-neox scores warping (#137) 2023-03-24 18:21:41 +01:00
OlivierDehaene 05e9a796cc
feat(server): flash neoX (#133) 2023-03-24 14:02:14 +01:00
OlivierDehaene b49dbf2d88
fix(server): use server tokenizer as gt (#128) 2023-03-16 12:12:26 +01:00
OlivierDehaene 8ad60b752f
fix(server): add position ids to neox (#126) 2023-03-15 13:12:49 +01:00
OlivierDehaene 941cd42e0c
fix(server): fix index out of range for watermarking (#110) 2023-03-08 18:29:08 +01:00
OlivierDehaene b1485e18c5
fix(server): fix galactica batch (#106)
closes #105
2023-03-07 20:05:21 +01:00
OlivierDehaene 3fef90d50f
feat(clients): Python client (#103) 2023-03-07 18:52:22 +01:00