(Part 2) feat: allow for tp_size attr for tplizing the model (#37054)

* feat: custom tp_size, new transformers tp interface

Signed-off-by: Mehant Kammakomati <mehant.kammakomati2@ibm.com>

* fix: review cmt - error when tp_plan not set for tp_size

Signed-off-by: Mehant Kammakomati <mehant.kammakomati2@ibm.com>

* fix: nit in docs

Signed-off-by: Mehant Kammakomati <mehant.kammakomati2@ibm.com>

---------

Signed-off-by: Mehant Kammakomati <mehant.kammakomati2@ibm.com>
Co-authored-by: Marc Sun <57196510+SunMarc@users.noreply.github.com>
Co-authored-by: Matej Sirovatka <54212263+S1ro1@users.noreply.github.com>

docs/source/ar/trainer.md

@@ -674,29 +674,7 @@ use_cpu: false
 ```
 
 </hfoption>
-<hfoption id="Tensor Parallelism with PyTorch 2">
 
-```yml
-compute_environment: LOCAL_MACHINE
-tp_config:
-  tp_size: 4
-distributed_type: TP
-downcast_bf16: 'no'
-machine_rank: 0
-main_training_function: main
-mixed_precision: 'no'
-num_machines: 1
-num_processes: 4
-rdzv_backend: static
-same_network: true
-tpu_env: []
-tpu_use_cluster: false
-tpu_use_sudo: false
-use_cpu: false
-
-```
-
-</hfoption>
 </hfoptions>
 يُعد أمر  [`accelerate_launch`](https://huggingface.co/docs/accelerate/package_reference/cli#accelerate-launch) هو الطريقة المُوصى بها لتشغيل نص البرمجى للتدريب على نظام موزع باستخدام Accelerate و [`Trainer`] مع المعلمات المحددة في `config_file.yaml`. يتم حفظ هذا الملف في مجلد ذاكرة التخزين المؤقت لـ Accelerate ويتم تحميله تلقائيًا عند تشغيل `accelerate_launch`.
 

docs/source/en/trainer.md

@@ -341,29 +341,9 @@ use_cpu: false
 ```
 
 </hfoption>
-<hfoption id="Tensor parallelism with PyTorch 2">
-
-```yaml
-compute_environment: LOCAL_MACHINE
-tp_config:
-  tp_size: 4
-distributed_type: TP
-downcast_bf16: 'no'
-machine_rank: 0
-main_training_function: main
-mixed_precision: 'no'
-num_machines: 1
-num_processes: 4
-rdzv_backend: static
-same_network: true
-tpu_env: []
-tpu_use_cluster: false
-tpu_use_sudo: false
-use_cpu: false
-```
-
 </hfoptions>
 
+
 Run [accelerate_launch](https://hf.co/docs/accelerate/package_reference/cli#accelerate-launch) to start training with the configurations set in `config_file.yaml`. This file is saved to the Accelerate cache folder and automatically loaded when you run `accelerate_launch`.
 
 The example below launches the [run_glue.py](../../../examples/pytorch/text-classification/run_glue) script with the FSDP configuration shown earlier. Parameters from the `config_file.yaml` file can also be directly set in the command line.

docs/source/es/trainer.md

@@ -363,29 +363,6 @@ use_cpu: false
 
 </hfoption>
 
-<hfoption id="Tensor Parallelism with PyTorch 2">
-
-```yml
-compute_environment: LOCAL_MACHINE
-tp_config:
-  tp_size: 4
-distributed_type: TP
-downcast_bf16: 'no'
-machine_rank: 0
-main_training_function: main
-mixed_precision: 'no'
-num_machines: 1
-num_processes: 4
-rdzv_backend: static
-same_network: true
-tpu_env: []
-tpu_use_cluster: false
-tpu_use_sudo: false
-use_cpu: false
-
-```
-
-</hfoption>
 </hfoptions>
 
 El comando [`accelerate_launch`](https://huggingface.co/docs/accelerate/package_reference/cli#accelerate-launch) es la forma recomendada de lanzar tu script de entrenamiento en un sistema distribuido con Accelerate y [`Trainer`] con los parámetros especificados en `config_file.yaml`. Este archivo se guarda en la carpeta de caché de Accelerate y se carga automáticamente cuando ejecutas `accelerate_launch`.

docs/source/ko/trainer.md

@@ -549,29 +549,7 @@ use_cpu: false
 ```
 
 </hfoption>
-<hfoption id="Tensor Parallelism with PyTorch 2">
 
-```yml
-compute_environment: LOCAL_MACHINE
-tp_config:
-  tp_size: 4
-distributed_type: TP
-downcast_bf16: 'no'
-machine_rank: 0
-main_training_function: main
-mixed_precision: 'no'
-num_machines: 1
-num_processes: 4
-rdzv_backend: static
-same_network: true
-tpu_env: []
-tpu_use_cluster: false
-tpu_use_sudo: false
-use_cpu: false
-
-```
-
-</hfoption>
 </hfoptions>
 
 [`accelerate_launch`](https://huggingface.co/docs/accelerate/package_reference/cli#accelerate-launch) 명령은 Accelerate와 [`Trainer`]를 사용하여 분산 시스템에서 훈련 스크립트를 실행하는 권장 방법이며, `config_file.yaml`에 지정된 매개변수를 사용합니다. 이 파일은 Accelerate 캐시 폴더에 저장되며 `accelerate_launch`를 실행할 때 자동으로 로드됩니다.