* docs(swin): Update Swin model card to standard format
* docs(swin): Refine link to Microsoft organization for Swin models
Apply suggestion from @stevhliu in PR #37628.
This change updates the link pointing to the official Microsoft Swin Transformer checkpoints on the Hugging Face Hub.
The link now directs users specifically to the Microsoft organization page, filtered for Swin models, providing a clearer and more canonical reference compared to the previous general search link.
Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>
* docs(swin): Clarify padding description and link to backbone docs
Apply suggestion from @stevhliu in PR #37628.
This change introduces two improvements to the Swin model card:
1. Refines the wording describing how Swin handles input padding for better clarity.
2. Adds an internal documentation link to the general "backbones" page when discussing Swin's capability as a backbone model.
These updates enhance readability and improve navigation within the Transformers documentation.
Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>
* docs(swin): Change Swin paper link to huggingface.co/papers as suggested
Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>
---------
Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>
* update model card.
* Apply suggestions from code review
Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>
* update quantization example.
* update example.
* update
---------
Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>
* assign the correct data layout for xpu
Signed-off-by: jiqing-feng <jiqing.feng@intel.com>
* check torch version before using torchao xpu
Signed-off-by: jiqing-feng <jiqing.feng@intel.com>
* fix the log
Signed-off-by: jiqing-feng <jiqing.feng@intel.com>
* fix zero point type
Signed-off-by: jiqing-feng <jiqing.feng@intel.com>
* fix check torch version
Signed-off-by: jiqing-feng <jiqing.feng@intel.com>
---------
Signed-off-by: jiqing-feng <jiqing.feng@intel.com>
* _get_padding_size module
* do not patchify images when processing multi image
* modify llava onevision image processor fast
* tensor to list of tensors
* backward compat
* reuse pad_to_square in llave & some clarification
* add to doc
* fix: consider no image cases (text only or video)
* add integration test
* style & repo_consistency
* accept custom device_mesh
* fix device_map
* assert that num_heads % tp_size == 0
* todo.
* ReplicateParallel
* handle tied weights
* handle dtensor in save_pretrained with safe_serialization
* tp test works
* doesnt work
* fix shard_and_distribute_module's rank should be local_rank
* tp=4 is correct
* dp+tp is broken
* todo allreduce with dtensors on another dim is annoying
* workaround to sync dp grads when using dtensors
* loading a checkpoint works
* wandb and compare losses with different tp/dp
* cleaning
* cleaning
* .
* .
* logs
* CP2 DP2 no mask works after commenting attn_mask and is_causal from scaled_dot_product_attention
* DP=2 TP=2 now works even with tied embeddings
* model.parameters() and model.module.parameters() are empty..
* reformat sanity_check_tensor_sync
* set atol=1e-4 for CP to pass
* try populate _parameters from named_modules
* refactors
TP2 DP2 works
CP2 DP2 works
* is_causal=True and pack sequences, no attn mask, and preshuffle dataset
* fix packing
* CP=4 doesn't work
* fix labels and position_ids for CP
* DP CP works with transformers 🥳🥳🥳
* refactor
* add example cp
* fixup
* revert sdpa changes
* example cleared
* add CP, DP to the mesh init
* nit
* clean
* use `ALL_PARALLEL_STYLES`
* style
* FSDP works
* log on 1 rank
* .
* fix?
* FSDP1 also has .parameters() bug
* reported gradnorm when using FSDP1 is wrong, but loss is correct so it's okay
* .
* style and fixup
* move stuff around
* fix tests
* style
* let's make it a check
* add missing licences
* warning should be an info
* tp plan should not be NONE
* test all
* god damn it
* test all
---------
Co-authored-by: nouamanetazi <nouamane98@gmail.com>
* add seq_idx and fa kwargs
* update tests
* docs and grad ckpt support
* fmt
* better names
* test_raise_missing_padding_free_kwarg_errs
* + seq_idx in doc strings
* padding free training docs
* add link to pr plots
* raise err on attn_mask with padding free
* rm raising missing padding free err test
* BambaFlashAttentionKwargs
* run modular util for modular_granitemoehybrid.py
* accept custom device_mesh
* fix device_map
* assert that num_heads % tp_size == 0
* todo.
* ReplicateParallel
* handle tied weights
* handle dtensor in save_pretrained with safe_serialization
* tp test works
* doesnt work
* fix shard_and_distribute_module's rank should be local_rank
* tp=4 is correct
* dp+tp is broken
* todo allreduce with dtensors on another dim is annoying
* workaround to sync dp grads when using dtensors
* loading a checkpoint works
* wandb and compare losses with different tp/dp
* cleaning
* cleaning
* .
* .
* logs
* CP2 DP2 no mask works after commenting attn_mask and is_causal from scaled_dot_product_attention
* DP=2 TP=2 now works even with tied embeddings
* model.parameters() and model.module.parameters() are empty..
* reformat sanity_check_tensor_sync
* set atol=1e-4 for CP to pass
* try populate _parameters from named_modules
* refactors
TP2 DP2 works
CP2 DP2 works
* is_causal=True and pack sequences, no attn mask, and preshuffle dataset
* fix packing
* CP=4 doesn't work
* fix labels and position_ids for CP
* DP CP works with transformers 🥳🥳🥳
* refactor
* add example cp
* fixup
* revert sdpa changes
* example cleared
* add CP, DP to the mesh init
* nit
* clean
* use `ALL_PARALLEL_STYLES`
* style
* FSDP works
* log on 1 rank
* .
* fix?
* FSDP1 also has .parameters() bug
* reported gradnorm when using FSDP1 is wrong, but loss is correct so it's okay
* .
* style and fixup
* move stuff around
* fix tests
* style
* let's make it a check
* warning should be an info
---------
Co-authored-by: Arthur Zucker <arthur.zucker@gmail.com>
When preparing the causal attention mask at this point the mask comes
in as a float tensor with min value as a masked value.
It is not correct to convert it to bool and treat it as a bool mask as
this inverts the mask.
`torch.nn.functional.scaled_dot_product_attention` expects that a masked value is `False`.
I suspect that the `sdpa` implementation variant may not have been
thoroughly tested and that is why this error was not caught earlier.
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
Add Llama4TextModel to AutoModel mapping
using Llama4TextConfig on AutoModel.from_config raises a ValueError when it is expected to instantiate a Llama4TextModel
bnb quant tests: remove obsolete trust_remote_code test
The MPT model is now natively integrated in Transformers and no longer requires trust_remote_code=True. This removes the failing test_get_keys_to_not_convert_trust_remote_code and related usage, which depended on remote code and caused CI issues due to missing dependencies (e.g., triton_pre_mlir).
* fix sliding attn
* make style
* Update tests/test_modeling_common.py
Co-authored-by: Joao Gante <joaofranciscocardosogante@gmail.com>
* no a second throught, should default to `True` fo BC
---------
Co-authored-by: Joao Gante <joaofranciscocardosogante@gmail.com>
* use device agnostic APIs in tests
Signed-off-by: Matrix Yao <matrix.yao@intel.com>
* more
Signed-off-by: Matrix Yao <matrix.yao@intel.com>
* fix style
Signed-off-by: Matrix Yao <matrix.yao@intel.com>
* add reset_peak_memory_stats API
Signed-off-by: YAO Matrix <matrix.yao@intel.com>
* update
---------
Signed-off-by: Matrix Yao <matrix.yao@intel.com>
Signed-off-by: YAO Matrix <matrix.yao@intel.com>
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
* Update modular_qwen2_5_omni.py
fix the error when loading quantized model by AuotAWQ.
* Update modeling_qwen2_5_omni.py
sync code to modular_qwen2_5_omni.py
