TST Use different diffusion model for testing

So far, tests are using hf-internal-testing/tiny-stable-diffusion-torch
for testing diffusion models. However, this model has some issues:

- still uses pickle (.bin) instead of safetensors
- there is a FutureWarning because of the config

Now, using hf-internal-testing/tiny-sd-pipe instead which doesn't have
those issues.

On top of that, there is often a Hub error when loading the diffusion
model, although it's unclear why this particular model is so prone to
that. Maybe the new model eschews this error, but even if it doesn't, it
is better to make the switch.

tests/test_helpers.py

@@ -179,7 +179,7 @@ def test_scaling_set_to_zero(self, tokenizer):
         assert torch.allclose(logits_base_model, logits_lora_model)
 
     def test_diffusers_pipeline(self):
-        model_id = "hf-internal-testing/tiny-stable-diffusion-torch"
+        model_id = "hf-internal-testing/tiny-sd-pipe"
         pipeline = StableDiffusionPipeline.from_pretrained(model_id)
 
         text_encoder_kwargs = {

tests/test_stablediffusion.py

@@ -36,7 +36,7 @@
 from .testing_utils import temp_seed
 
 
-PEFT_DIFFUSERS_SD_MODELS_TO_TEST = ["hf-internal-testing/tiny-stable-diffusion-torch"]
+PEFT_DIFFUSERS_SD_MODELS_TO_TEST = ["hf-internal-testing/tiny-sd-pipe"]
 CONFIG_TESTING_KWARGS = (
     {
         "text_encoder": {

tests/test_tuners_utils.py

@@ -331,7 +331,7 @@ def test_maybe_include_all_linear_layers_internals(self, initial_target_modules,
             assert new_config.target_modules == expected_target_modules
 
     def test_maybe_include_all_linear_layers_diffusion(self):
-        model_id = "hf-internal-testing/tiny-stable-diffusion-torch"
+        model_id = "hf-internal-testing/tiny-sd-pipe"
         model = StableDiffusionPipeline.from_pretrained(model_id)
         config = LoraConfig(base_model_name_or_path=model_id, target_modules="all-linear")