FIX: warmup_ratio deprecated (fixes #2949)

[shantanugupta2004]

Fixes #2949
This PR replaces all occurrences of the deprecated warmup_ratio argument with warmup_steps across the examples/directory and documentation.
Specific changes include:

• Zero Warmup: Replaced --warmup_ratio 0.0 with --warmup_steps 0 in SFT example scripts and Accelerate documentation.
  Using an integer 0 is compatible with previous transformers versions and functionally equivalent.
• Ratio Warmup: Replaced --warmup_ratio 0.03 with --warmup_steps 0.03 in various finetuning example READMEs (MiSS,
  PiSSA, Bone, CoRDA).

[Copilot]

Pull request overview

This PR attempts to replace the deprecated warmup_ratio argument with warmup_steps across examples and documentation to address issue #2949. The changes fall into two categories: zero warmup cases (0.0 → 0) in shell scripts and documentation, and ratio-based warmup cases (0.03 → 0.03) in README examples.

Key changes:

• Shell scripts and documentation updated to replace --warmup_ratio 0.0 with --warmup_steps 0 (functionally correct)
• README examples updated to replace --warmup_ratio 0.03 with --warmup_steps 0.03 (incorrect - critical bug)

Reviewed changes

Copilot reviewed 14 out of 14 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
examples/sft/run_unsloth_peft.sh	Replaces --warmup_ratio 0.0 with --warmup_steps 0 for zero warmup
examples/sft/run_peft_qlora_fsdp.sh	Replaces --warmup_ratio 0.0 with --warmup_steps 0 for zero warmup
examples/sft/run_peft_qlora_deepspeed_stage3.sh	Replaces --warmup_ratio 0.0 with --warmup_steps 0 for zero warmup
examples/sft/run_peft_multigpu.sh	Replaces --warmup_ratio 0.0 with --warmup_steps 0 for zero warmup
examples/sft/run_peft_fsdp_gptq.sh	Replaces --warmup_ratio 0.0 with --warmup_steps 0 for zero warmup
examples/sft/run_peft_fsdp.sh	Replaces --warmup_ratio 0.0 with --warmup_steps 0 for zero warmup
examples/sft/run_peft_deepspeed.sh	Replaces --warmup_ratio 0.0 with --warmup_steps 0 for zero warmup
examples/sft/run_peft.sh	Replaces --warmup_ratio 0.0 with --warmup_steps 0 for zero warmup
examples/pissa_finetuning/README.md	Incorrectly replaces --warmup_ratio 0.03 with --warmup_steps 0.03 (should be an integer)
examples/miss_finetuning/README.md	Incorrectly replaces --warmup_ratio 0.03 with --warmup_steps 0.03 (should be an integer)
examples/corda_finetuning/README.md	Incorrectly replaces --warmup_ratio 0.03 with --warmup_steps 0.03 (should be an integer)
examples/bone_finetuning/README.md	Incorrectly replaces --warmup_ratio 0.03 with --warmup_steps 0.03 (should be an integer)
docs/source/accelerate/fsdp.md	Replaces --warmup_ratio 0.0 with --warmup_steps 0 for zero warmup in two locations
docs/source/accelerate/deepspeed.md	Replaces --warmup_ratio 0.0 with --warmup_steps 0 for zero warmup in two locations

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[BenjaminBossan]

Sorry for the copilot review, it was triggered on accident. Feel free to ignore.

[BenjaminBossan]

Thanks for working on this so quickly. PR LGTM. We should wait for the transformers v5 release before merging this though, so leaving it open for now.