Zach Mueller
1681a6d452
🚨 Fully revert atomic checkpointing 🚨 ( #29370 )
...
Fully revert atomic checkpointing
2024-03-04 06:17:42 -05:00
Nick DeGroot
8ef9862864
Fix OneFormer post_process_instance_segmentation for panoptic tasks ( #29304 )
...
* 🐛 Fix oneformer instance post processing when using panoptic task type
* ✅ Add unit test for oneformer instance post processing panoptic bug
---------
Co-authored-by: Nick DeGroot <1966472+nickthegroot@users.noreply.github.com>
2024-03-04 11:04:49 +00:00
Sean (Seok-Won) Yi
81220cba61
Fix: Fixed the previous tracking URI setting logic to prevent clashes with original MLflow code. ( #29096 )
...
* Changed logic for setting the tracking URI.
The previous code was calling the `mlflow.set_tracking_uri` function
regardless of whether or not the environment variable
`MLFLOW_TRACKING_URI` is even set. This led to clashes with the original
MLflow implementation and therefore the logic was changed to only
calling the function when the environment variable is explicitly set.
* Check if tracking URI has already been set.
The previous code did not consider the possibility that the tracking URI
may already be set elsewhere and was therefore (erroneously) overriding
previously set tracking URIs using the environment variable.
* Removed redundant parentheses.
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>
* Fix docstring to reflect library convention properly.
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>
* Fix docstring to reflect library convention properly.
"Unset by default" is the correct expression rather than "Default to `None`."
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>
---------
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>
2024-03-04 10:53:58 +00:00
NielsRogge
5e4b69dc12
Convert SlimSAM checkpoints ( #28379 )
...
* First commit
* Improve conversion script
* Convert more checkpoints
* Update src/transformers/models/sam/convert_sam_original_to_hf_format.py
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
* Rename file
* More updates
* Update docstring
* Update script
---------
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
2024-03-04 11:51:16 +01:00
Traun Leyden
c38a12270a
Workaround for #27758 to avoid ZeroDivisionError ( #28756 )
2024-03-04 10:23:40 +01:00
Y4hL
704b3f74f9
Add mlx support to BatchEncoding.convert_to_tensors ( #29406 )
...
* Add mlx support
* Fix import order and use def instead of lambda
* Another fix for ruff format :)
* Add detecting mlx from repr, add is_mlx_array
2024-03-04 10:19:13 +01:00
Siming Dai
39ef3fb248
[Mixtral] Fixes attention masking in the loss ( #29363 )
...
Fix mixtral load balancing loss
Co-authored-by: dingkunbo <dingkunbo@baidu.com>
2024-03-04 09:08:56 +01:00
Poedator
38953a75c1
update path to hub files in the error message ( #29369 )
...
update path to hub files
need to add `tree/` to path to files at HF hub.
see example path:
`https://huggingface.co/meta-llama/Llama-2-7b-hf/tree/main `
2024-03-04 08:26:01 +01:00
Fanli Lin
aade711d1e
[tests] enable automatic speech recognition pipeline tests on XPU ( #29308 )
...
* use require_torch_gpu
* enable on XPU
2024-03-04 08:24:38 +01:00
David Valente
831bc25d8f
Correct zero division error in inverse sqrt scheduler ( #28982 )
...
* Correct zero division error in inverse sqrt scheduler
* default timescale to 10_000
2024-03-01 17:04:40 +00:00
Zach Mueller
1a7c117df9
Fix deprecated arg issue ( #29372 )
...
* Fix deprecated arg issue
* Trainer check too
* Check for dict or dataclass
* Simplify, make config always AcceleratorConfig
* Upstream to Trainer
2024-03-01 12:00:29 -05:00
Marc Sun
cec773345a
Fix llama + gemma accelete tests ( #29380 )
2024-03-01 10:32:36 -05:00
Jingya HUANG
15f8296a9b
Support subfolder with AutoProcessor ( #29169 )
...
enable subfolder
2024-03-01 10:29:21 +00:00
amyeroberts
f1b1379f37
[YOLOS] Fix - return padded annotations ( #29300 )
...
* Fix yolos processing
* Add back slow marker - protects for pycocotools in slow
* Slow decorator goes above copied from header
2024-03-01 09:42:13 +00:00
Sanchit Gandhi
0a0a279e99
🚨 🚨 [Whisper Tok] Update integration test ( #29368 )
...
* [Whisper Tok] Update integration test
* make style
2024-03-01 09:22:31 +00:00
Arthur
e7b9837065
[Llama + AWQ] fix prepare_inputs_for_generation 🫠 ( #29381 )
...
* use the generation config 🫠
* fixup
2024-03-01 08:59:26 +01:00
Younes Belkada
50db7ca4e8
FIX [quantization / ESM] Fix ESM 8bit / 4bit with bitsandbytes ( #29329 )
...
* fix ESM 8bit
* Apply suggestions from code review
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
* fixup
---------
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
2024-03-01 03:01:53 +01:00
Leon Engländer
2858d6c634
Fix Base Model Name of LlamaForQuestionAnswering ( #29258 )
...
* LlamaForQuestionAnswering self.transformer->self.model
* fix "Copied from" string
* Llama QA model: set base_model_prefix = "transformer"
2024-03-01 02:58:19 +01:00
Song Fuchang
5ee0868a4b
Expose offload_buffers parameter of accelerate to PreTrainedModel.from_pretrained method ( #28755 )
...
Expose offload_buffers parameter to from_pretrained method
2024-03-01 02:12:51 +01:00
Lucain
0ad770c373
Fix @require_read_token in tests ( #29367 )
2024-02-29 11:25:16 +01:00
NielsRogge
bb4f816ad4
Patch YOLOS and others ( #29353 )
...
Fix issue
2024-02-29 11:09:50 +01:00
Yih-Dar
44fe1a1cc4
Avoid using uncessary get_values(MODEL_MAPPING) ( #29362 )
...
* more fixes
* more fixes
---------
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
2024-02-29 17:19:17 +08:00
Younes Belkada
b647acdb53
FIX [CI] require_read_token in the llama FA2 test ( #29361 )
...
Update test_modeling_llama.py
2024-02-29 04:49:01 +01:00
Younes Belkada
8d8ac9c2df
FIX [CI]: Fix failing tests for peft integration ( #29330 )
...
fix failing tests for peft integration
2024-02-29 03:56:16 +01:00
Younes Belkada
1aee9afd1c
FIX [CI / starcoder2] Change starcoder2 path to correct one for slow tests ( #29359 )
...
change starcoder2 path to correct one
2024-02-29 03:52:13 +01:00
Michael
2209b7afa0
[i18n-zh] Sync source/zh/index.md ( #29331 )
...
* [i18n-zh] Sync source/zh/index.md
* apply review comments
2024-02-28 09:41:18 -08:00
fxmarty
49204c1d37
Better SDPA unmasking implementation ( #29318 )
...
* better unmask imple
* comment
* typo
* bug report pytorch
* cleanup
* fix import
* add back example
* retrigger ci
* come on
2024-02-28 16:36:47 +01:00
Marc Sun
f54d82cace
[CI] Quantization workflow ( #29046 )
...
* [CI] Quantization workflow
* build dockerfile
* fix dockerfile
* update self-cheduled.yml
* test build dockerfile on push
* fix torch install
* udapte to python 3.10
* update aqlm version
* uncomment build dockerfile
* tests if the scheduler works
* fix docker
* do not trigger on psuh again
* add additional runs
* test again
* all good
* style
* Update .github/workflows/self-scheduled.yml
Co-authored-by: Younes Belkada <49240599+younesbelkada@users.noreply.github.com>
* test build dockerfile with torch 2.2.0
* fix extra
* clean
* revert changes
* Revert "revert changes"
This reverts commit 4cb52b8822
2024-02-28 10:09:25 -05:00
jiqing-feng
554e7ada89
check if position_ids exists before using it ( #29306 )
...
Co-authored-by: Joao Gante <joao@huggingface.co>
2024-02-28 14:56:25 +00:00
Daniel Han
d3a4b47544
RoPE loses precision for Llama / Gemma + Gemma logits.float() ( #29285 )
...
* Update modeling_llama.py
Llama - Force float32 since bfloat16 loses precision on long contexts
* Update modeling_llama.py
* Update modeling_gemma.py
Fix RoPE and logits.float()
* @torch.no_grad()
* @torch.no_grad()
* Cos, Sin to float32
* cos, sin to float32
* Update src/transformers/models/gemma/modeling_gemma.py
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
* Update src/transformers/models/llama/modeling_llama.py
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
* Resolve PR conflicts
* Fix RoPE for llama
* Revert "Fix RoPE for llama"
This reverts commit b860a22dab
2024-02-28 15:16:53 +01:00
Joao Gante
7628b3a0f4
Idefics: generate fix ( #29320 )
2024-02-28 11:34:54 +00:00
Leonardo Emili
2ce56d35f6
Disable Mixtral output_router_logits during inference ( #29249 )
...
* Set output_router_logits=False in prepare_inputs_for_generation for mixtral
* Add output_router_logits=False to prepare_inputs_for_generation for mixtral
* Fix style
2024-02-28 11:16:15 +01:00
Arthur
8a8a0a4ae0
[Llama ROPE] Fix torch export but also slow downs in forward ( #29198 )
...
* remove control flow
* update gptneox
* update ....
* nits
* Actually let's just break. Otherwise we are silently failing which imo is not optimal
* version BC
* fix tests
* fix eager causal
* nit
* add a test
* style
* nits
* nits
* more nits for the test
* update and fix
* make sure cuda graphs are not skipped
* read token is needed for meta llama
* update!
* fiixup
* compile test should be slow
* fix thet fix copies
* stle 🫠
2024-02-28 10:45:53 +01:00
Arthur
7c87f3577e
[T5 and Llama Tokenizer] remove warning ( #29346 )
...
* remove warning
* add co-author
* update
---------
Co-authored-by: hiaoxui <hiaoxui@users.noreply.github.com>
2024-02-28 10:41:58 +01:00
Arthur
a52888524d
[require_read_token] fix typo ( #29345 )
...
fix wrapper
2024-02-28 10:13:57 +01:00
fxmarty
e715c78c66
Remove numpy usage from owlvit ( #29326 )
...
* remove numpy usage from owlvit
* fix init owlv2
* style
2024-02-28 09:38:44 +01:00
Younes Belkada
ad00c482c7
FIX [Gemma / CI] Make sure our runners have access to the model ( #29242 )
...
* pu hf token in gemma tests
* update suggestion
* add to flax
* revert
* fix
* fixup
* forward contrib credits from discussion
---------
Co-authored-by: ArthurZucker <ArthurZucker@users.noreply.github.com>
2024-02-28 06:25:23 +01:00
Jared Van Bortel
bd5b986306
simplify get_class_in_module and fix for paths containing a dot ( #29262 )
2024-02-28 03:10:36 +01:00
RaymondLi0
63caa370e6
Starcoder2 model - bis ( #29215 )
...
* Copy model
* changes
* misc
* fixes
* add embed and residual dropout (#30 )
* misc
* remove rms norm and gated MLP
* remove copied mentions where its not a copy anymore
* remove unused _shape
* copied from mistral instead
* fix copies
* fix copies
* add not doctested
* fix
* fix copyright
* Update docs/source/en/model_doc/starcoder2.md
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
* Update src/transformers/models/starcoder2/configuration_starcoder2.py
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
* Update src/transformers/models/starcoder2/configuration_starcoder2.py
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
* fix doc
* revert some changes
* add fa2 tests
* fix styling nit
* fix
* push dummy docs
---------
Co-authored-by: Joel Lamy-Poirier <joel.lamy-poirier@servicenow.com>
Co-authored-by: younesbelkada <younesbelkada@gmail.com>
Co-authored-by: Younes Belkada <49240599+younesbelkada@users.noreply.github.com>
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
2024-02-28 01:24:34 +01:00
Michael
83ab0115d1
[i18n-zh] Translate fsdp.md into Chinese ( #29305 )
...
* [i18n-zh] Translate fsdp.md into Chinese
Signed-off-by: windsonsea <haifeng.yao@daocloud.io>
* apply suggestions from Fan-Lin
---------
Signed-off-by: windsonsea <haifeng.yao@daocloud.io>
2024-02-27 11:26:57 -08:00
Sadra Barikbin
227cd54aa5
Fix a few typos in GenerationMixin's docstring ( #29277 )
...
Co-authored-by: Joao Gante <joao@huggingface.co>
2024-02-27 18:15:43 +00:00
Raushan Turganbay
ddf7ac4237
Token level timestamps for long-form generation in Whisper ( #29148 )
2024-02-27 18:15:26 +00:00
Marc Sun
8a1faf2803
Add compatibility with skip_memory_metrics for mps device ( #29264 )
...
* Add compatibility with mps device
* fix
* typo and style
2024-02-27 09:58:43 -05:00
Yih-Dar
5c341d4555
Use torch 2.2 for deepspeed CI ( #29246 )
...
update
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
2024-02-27 17:51:37 +08:00
Fanli Lin
63a0c8f1cb
[tests] enable benchmark unit tests on XPU ( #29284 )
...
* add xpu for benchmark
* no auto_map
* use require_torch_gpu
* use gpu
* revert
* revert
* fix style
2024-02-27 09:44:48 +00:00
fxmarty
6d3b643e2a
Fix attn_implementation documentation ( #29295 )
...
fix
2024-02-27 10:43:01 +01:00
Merve Noyan
83e366bfd4
Image Feature Extraction docs ( #28973 )
...
* Image Feature Extraction docs
* Update docs/source/en/tasks/image_feature_extraction.md
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>
* Update image_feature_extraction.md
* Update docs/source/en/tasks/image_feature_extraction.md
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>
* Update docs/source/en/tasks/image_feature_extraction.md
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>
* Address comments
* Update docs/source/en/tasks/image_feature_extraction.md
Co-authored-by: Maria Khalusova <kafooster@gmail.com>
* Update docs/source/en/tasks/image_feature_extraction.md
Co-authored-by: Maria Khalusova <kafooster@gmail.com>
* Update docs/source/en/tasks/image_feature_extraction.md
Co-authored-by: Maria Khalusova <kafooster@gmail.com>
* Update docs/source/en/tasks/image_feature_extraction.md
Co-authored-by: Maria Khalusova <kafooster@gmail.com>
* Update docs/source/en/tasks/image_feature_extraction.md
Co-authored-by: Maria Khalusova <kafooster@gmail.com>
* Update docs/source/en/tasks/image_feature_extraction.md
Co-authored-by: Maria Khalusova <kafooster@gmail.com>
* Update docs/source/en/tasks/image_feature_extraction.md
Co-authored-by: Maria Khalusova <kafooster@gmail.com>
* Update docs/source/en/tasks/image_feature_extraction.md
Co-authored-by: Maria Khalusova <kafooster@gmail.com>
* Update image_feature_extraction.md
* Update image_feature_extraction.md
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>
---------
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>
Co-authored-by: Maria Khalusova <kafooster@gmail.com>
2024-02-27 09:39:58 +00:00
Andrei Panferov
e3fc90ae68
Cleaner Cache dtype and device extraction for CUDA graph generation for quantizers compatibility ( #29079 )
...
* input_layernorm as the beacon of hope
* cleaner dtype extraction
* AQLM + CUDA graph test
* is available check
* shorter text test
2024-02-27 09:32:39 +01:00
regisss
a3f9221a44
Add generate kwargs to VQA pipeline ( #29134 )
2024-02-27 03:03:00 +01:00
FredericOdermatt
871ba71dfa
GenerationConfig validate both constraints and force_words_ids ( #29163 )
...
GenerationConfig validate both options for constrained decoding: constraints and force_words_ids
2024-02-27 01:43:52 +01:00
