Merge branch 'main' into sam-rle-bug

src/transformers/configuration_utils.py

@@ -994,8 +994,11 @@ def dict_torch_dtype_to_str(self, d: Dict[str, Any]) -> None:
         converts torch.dtype to a string of just the type. For example, `torch.float32` get converted into *"float32"*
         string, which can then be stored in the json format.
         """
-        if d.get("torch_dtype", None) is not None and not isinstance(d["torch_dtype"], str):
-            d["torch_dtype"] = str(d["torch_dtype"]).split(".")[1]
+        if d.get("torch_dtype", None) is not None:
+            if isinstance(d["torch_dtype"], dict):
+                d["torch_dtype"] = {k: str(v).split(".")[-1] for k, v in d["torch_dtype"].items()}
+            elif not isinstance(d["torch_dtype"], str):
+                d["torch_dtype"] = str(d["torch_dtype"]).split(".")[1]
         for value in d.values():
             if isinstance(value, dict):
                 self.dict_torch_dtype_to_str(value)

src/transformers/modeling_utils.py

@@ -1312,11 +1312,10 @@ def __init__(self, config: PretrainedConfig, *inputs, **kwargs):
                 "`PretrainedConfig`. To create a model from a pretrained model use "
                 f"`model = {self.__class__.__name__}.from_pretrained(PRETRAINED_MODEL_NAME)`"
             )
-        # Save config and origin of the pretrained weights if given in model
         if not getattr(config, "_attn_implementation_autoset", False):
-            config = self._autoset_attn_implementation(
-                config, torch_dtype=torch.get_default_dtype(), check_device_map=False
-            )
+            # config usually has a `torch_dtype` but we need the next line for the `no_super_init` tests
+            dtype = config.torch_dtype if hasattr(config, "torch_dtype") else torch.get_default_dtype()
+            config = self._autoset_attn_implementation(config, torch_dtype=dtype, check_device_map=False)
         self.config = config
 
         # for initialization of the loss
@@ -1411,7 +1410,10 @@ def _from_config(cls, config, **kwargs):
         # when we init a model from within another model (e.g. VLMs) and dispatch on FA2
         # a warning is raised that dtype should be fp16. Since we never pass dtype from within
         # modeling code, we can try to infer it here same way as done in `from_pretrained`
-        torch_dtype = kwargs.pop("torch_dtype", torch.get_default_dtype())
+        torch_dtype = kwargs.pop("torch_dtype", config.torch_dtype)
+        if isinstance(torch_dtype, str):
+            torch_dtype = getattr(torch, torch_dtype)
+
         use_flash_attention_2 = kwargs.pop("use_flash_attention_2", False)
 
         # override default dtype if needed
@@ -4020,11 +4022,37 @@ def from_pretrained(
                             )
                     elif hasattr(torch, torch_dtype):
                         torch_dtype = getattr(torch, torch_dtype)
-                    else:
-                        raise ValueError(
-                            f'`torch_dtype` can be one of: `torch.dtype`, `"auto"` or a string of a valid `torch.dtype`, but received {torch_dtype}'
-                        )
+                        for sub_config_key in config.sub_configs.keys():
+                            sub_config = getattr(config, sub_config_key)
+                            sub_config.torch_dtype = torch_dtype
+                elif isinstance(torch_dtype, torch.dtype):
+                    pass
+                elif isinstance(torch_dtype, dict):
+                    for key, curr_dtype in torch_dtype.items():
+                        if hasattr(config, key):
+                            value = getattr(config, key)
+                            value.torch_dtype = curr_dtype
+                    # main torch dtype for modules that aren't part of any sub-config
+                    torch_dtype = torch_dtype.get("")
+                    config.torch_dtype = torch_dtype
+                    if isinstance(torch_dtype, str) and hasattr(torch, torch_dtype):
+                        torch_dtype = getattr(torch, torch_dtype)
+                    elif torch_dtype is None:
+                        torch_dtype = torch.float32
+                else:
+                    raise ValueError(
+                        f"`torch_dtype` can be one of: `torch.dtype`, `'auto'`, a string of a valid `torch.dtype` or a `dict` with valid `torch_dtype` "
+                        f"for each sub-config in composite configs, but received {torch_dtype}"
+                    )
+
                 dtype_orig = cls._set_default_torch_dtype(torch_dtype)
+            else:
+                # set fp32 as the default dtype for BC
+                default_dtype = str(torch.get_default_dtype()).split(".")[-1]
+                config.torch_dtype = default_dtype
+                for key in config.sub_configs.keys():
+                    value = getattr(config, key)
+                    value.torch_dtype = default_dtype
 
             # Check if `_keep_in_fp32_modules` is not None
             use_keep_in_fp32_modules = (cls._keep_in_fp32_modules is not None) and (

src/transformers/models/chameleon/modeling_chameleon.py

@@ -967,62 +967,6 @@ def forward(self, pixel_values: torch.LongTensor):
         return last_hidden_state
 
 
-CHAMELEON_VQ_START_DOCSTRING = r"""
-    This model inherits from [`PreTrainedModel`]. Check the superclass documentation for the generic methods the
-    library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads
-    etc.)
-
-    This model is also a PyTorch [torch.nn.Module](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) subclass.
-    Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage
-    and behavior.
-
-    Parameters:
-        config ([`ChameleonVQVAEConfig`]):
-            Model configuration class with all the parameters of the model. Initializing with a config file does not
-            load the weights associated with the model, only the configuration. Check out the
-            [`~PreTrainedModel.from_pretrained`] method to load the model weights.
-"""
-
-
-@add_start_docstrings(
-    """The VQ-VAE model used in Chameleon for encoding/decoding images into discrete tokens.
-    This model follows the "Make-a-scene: Scene-based text-to-image generation with human priors" paper from
-    [ Oran Gafni, Adam Polyak, Oron Ashual, Shelly Sheynin, Devi Parikh, and Yaniv Taigman](https://arxiv.org/abs/2203.13131).
-    """,
-    CHAMELEON_VQ_START_DOCSTRING,
-)
-class ChameleonVQVAE(PreTrainedModel):
-    config_class = ChameleonVQVAEConfig
-    _no_split_modules = ["ChameleonVQVAEVectorQuantizer"]
-
-    def _init_weights(self, module):
-        std = self.config.initializer_range
-        if isinstance(module, nn.Embedding):
-            module.weight.data.normal_(mean=0.0, std=std)
-        elif isinstance(module, nn.GroupNorm):
-            module.bias.data.zero_()
-            module.weight.data.fill_(1.0)
-        elif isinstance(module, (nn.Linear, nn.Conv2d)):
-            module.weight.data.normal_(mean=0.0, std=std)
-            if module.bias is not None:
-                module.bias.data.zero_()
-
-    def __init__(self, config: ChameleonVQVAEConfig):
-        super().__init__(config)
-
-        self.encoder = ChameleonVQVAEEncoder(config)
-        self.quantize = ChameleonVQVAEVectorQuantizer(config)
-        self.quant_conv = torch.nn.Conv2d(config.latent_channels, config.embed_dim, 1)
-        self.post_quant_conv = torch.nn.Conv2d(config.embed_dim, config.latent_channels, 1)
-        self.eval()  # Chameleon's VQ model is frozen
-
-    def encode(self, pixel_values: torch.LongTensor):
-        hidden_states = self.encoder(pixel_values)
-        hidden_states = self.quant_conv(hidden_states)
-        quant, emb_loss, indices = self.quantize(hidden_states)
-        return quant, emb_loss, indices
-
-
 class ChameleonImageVocabularyMapping:
     """
     A class for mapping discrete image tokens from VQGAN to BPE tokens.
@@ -1118,6 +1062,62 @@ def _init_weights(self, module):
                 module.weight.data[module.padding_idx].zero_()
 
 
+CHAMELEON_VQ_START_DOCSTRING = r"""
+    This model inherits from [`PreTrainedModel`]. Check the superclass documentation for the generic methods the
+    library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads
+    etc.)
+
+    This model is also a PyTorch [torch.nn.Module](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) subclass.
+    Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage
+    and behavior.
+
+    Parameters:
+        config ([`ChameleonVQVAEConfig`]):
+            Model configuration class with all the parameters of the model. Initializing with a config file does not
+            load the weights associated with the model, only the configuration. Check out the
+            [`~PreTrainedModel.from_pretrained`] method to load the model weights.
+"""
+
+
+@add_start_docstrings(
+    """The VQ-VAE model used in Chameleon for encoding/decoding images into discrete tokens.
+    This model follows the "Make-a-scene: Scene-based text-to-image generation with human priors" paper from
+    [ Oran Gafni, Adam Polyak, Oron Ashual, Shelly Sheynin, Devi Parikh, and Yaniv Taigman](https://arxiv.org/abs/2203.13131).
+    """,
+    CHAMELEON_VQ_START_DOCSTRING,
+)
+class ChameleonVQVAE(ChameleonPreTrainedModel):
+    config_class = ChameleonVQVAEConfig
+    _no_split_modules = ["ChameleonVQVAEVectorQuantizer"]
+
+    def _init_weights(self, module):
+        std = self.config.initializer_range
+        if isinstance(module, nn.Embedding):
+            module.weight.data.normal_(mean=0.0, std=std)
+        elif isinstance(module, nn.GroupNorm):
+            module.bias.data.zero_()
+            module.weight.data.fill_(1.0)
+        elif isinstance(module, (nn.Linear, nn.Conv2d)):
+            module.weight.data.normal_(mean=0.0, std=std)
+            if module.bias is not None:
+                module.bias.data.zero_()
+
+    def __init__(self, config: ChameleonVQVAEConfig):
+        super().__init__(config)
+
+        self.encoder = ChameleonVQVAEEncoder(config)
+        self.quantize = ChameleonVQVAEVectorQuantizer(config)
+        self.quant_conv = torch.nn.Conv2d(config.latent_channels, config.embed_dim, 1)
+        self.post_quant_conv = torch.nn.Conv2d(config.embed_dim, config.latent_channels, 1)
+        self.eval()  # Chameleon's VQ model is frozen
+
+    def encode(self, pixel_values: torch.LongTensor):
+        hidden_states = self.encoder(pixel_values)
+        hidden_states = self.quant_conv(hidden_states)
+        quant, emb_loss, indices = self.quantize(hidden_states)
+        return quant, emb_loss, indices
+
+
 CHAMELEON_INPUTS_DOCSTRING = r"""
     Args:
         input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
@@ -1211,7 +1211,7 @@ def __init__(self, config: ChameleonConfig):
             [decoder_layer(config, layer_idx) for layer_idx in range(config.num_hidden_layers)]
         )
         self.norm = ChameleonRMSNorm(config.hidden_size, eps=config.rms_norm_eps)
-        self.vqmodel = ChameleonVQVAE(config.vq_config)
+        self.vqmodel = ChameleonVQVAE._from_config(config.vq_config)
         self.gradient_checkpointing = False
 
         # Initialize weights and apply final processing

src/transformers/models/phi/modeling_phi.py

@@ -727,7 +727,7 @@ def __init__(self, config):
         super().__init__(config)
         self.model = PhiModel(config)
         self.vocab_size = config.vocab_size
-        self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)
+        self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=True)
 
         # Initialize weights and apply final processing
         self.post_init()

src/transformers/models/phi/modular_phi.py

@@ -284,7 +284,9 @@ def forward(
 
 
 class PhiForCausalLM(LlamaForCausalLM):
-    pass
+    def __init__(self, config):
+        super().__init__(config)
+        self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=True)
 
 
 class PhiForSequenceClassification(LlamaForSequenceClassification):

tests/fsdp/test_fsdp.py

@@ -32,7 +32,6 @@
     require_accelerate,
     require_fsdp,
     require_torch_accelerator,
-    require_torch_gpu,
     require_torch_multi_accelerator,
     slow,
     torch_device,
@@ -288,7 +287,7 @@ def test_training_and_can_resume_normally(self, state_dict_type):
 
     @require_torch_multi_accelerator
     @slow
-    @require_torch_gpu
+    @require_torch_accelerator
     @require_fsdp
     def test_fsdp_cpu_offloading(self):
         try:

tests/generation/test_utils.py

@@ -33,6 +33,7 @@
     require_flash_attn,
     require_optimum_quanto,
     require_torch,
+    require_torch_accelerator,
     require_torch_gpu,
     require_torch_multi_accelerator,
     require_torch_multi_gpu,
@@ -2042,16 +2043,10 @@ def test_generate_with_quant_cache(self):
             with self.assertRaises(ValueError):
                 model.generate(**generation_kwargs, **inputs_dict)
 
-    @parameterized.expand(
-        [
-            ("forward_only", False),  # TODO (@joao): a few models failing. After fixed, this should not be "@slow"
-            ("end_to_end", True),  # TODO (@joao): end-to-end compilation is broken with torch 2.5+, explore and fix
-        ]
-    )
     @pytest.mark.generate
-    @require_torch_gpu
+    @require_torch_accelerator
     @slow
-    def test_generate_compile(self, _, end_to_end):
+    def test_generate_compile_model_forward(self):
         """
         Tests that `.generate` is compatible with torch.compile without graph breaks, keeping the same results. Tests
         end-to-end compilation and forward pass compilation only.
@@ -2061,14 +2056,7 @@ def test_generate_compile(self, _, end_to_end):
             if not model_class._supports_static_cache:
                 self.skipTest("This model doesn't support static cache")
 
-            # TODO (joao) -- fix and enable me :)
-            if end_to_end and any(model_name in model_class.__name__.lower() for model_name in ["whisper"]):
-                self.skipTest("whisper model end-to-end generate compile not yet supported")
-
             config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common()
-            # TODO (joao) -- fix and enable me :)
-            if end_to_end and config.is_encoder_decoder:
-                self.skipTest("Encoder-decoder model end-to-end generate compile not yet supported")
 
             model = model_class(config).to(torch_device)
             model.eval()  # otherwise `self.training` is `True` -- this flag is used at attn mask creation time
@@ -2084,10 +2072,8 @@ def test_generate_compile(self, _, end_to_end):
                 "max_new_tokens": 10,
                 "return_dict_in_generate": True,
                 "output_scores": True,
+                "cache_implementation": "static",
             }
-            # end-to-end works best with dynamic cache, forward compilation works best with static cache
-            if not end_to_end:
-                generation_kwargs["cache_implementation"] = "static"
 
             # get eager + dynamic cache results for future comparison
             dynamic_outputs = []
@@ -2098,10 +2084,8 @@ def test_generate_compile(self, _, end_to_end):
             generation_config = copy.deepcopy(model.generation_config)
             generation_config.update(**generation_kwargs)
             torch.compiler.reset()
-            if end_to_end:
-                model.generate = torch.compile(model.generate, fullgraph=True, mode="reduce-overhead")
-            else:
-                model.forward = torch.compile(model.forward, fullgraph=True, mode="reduce-overhead")
+
+            model.forward = torch.compile(model.forward, fullgraph=True, mode="reduce-overhead")
 
             compiled_outputs = []
             for model_inputs in input_ids_sets:
@@ -3808,10 +3792,12 @@ def test_assisted_decoding_in_different_gpu(self):
         self.assertTrue(input_length <= out.shape[-1] <= input_length + 20)
 
     @slow
-    @require_torch_gpu
+    @require_torch_accelerator
     def test_assisted_decoding_model_in_gpu_assistant_in_cpu(self):
         # PT-only test: TF doesn't support assisted decoding yet.
-        model = AutoModelForCausalLM.from_pretrained("hf-internal-testing/tiny-random-MistralForCausalLM").to("cuda")
+        model = AutoModelForCausalLM.from_pretrained("hf-internal-testing/tiny-random-MistralForCausalLM").to(
+            torch_device
+        )
         assistant = AutoModelForCausalLM.from_pretrained("hf-internal-testing/tiny-random-MistralForCausalLM").to(
             "cpu"
         )

tests/models/blip_2/test_modeling_blip_2.py

@@ -27,6 +27,7 @@
 from transformers import CONFIG_MAPPING, Blip2Config, Blip2QFormerConfig, Blip2VisionConfig
 from transformers.testing_utils import (
     require_torch,
+    require_torch_accelerator,
     require_torch_fp16,
     require_torch_gpu,
     require_torch_multi_accelerator,
@@ -1565,7 +1566,7 @@ def test_forward_signature(self):
             self.assertListEqual(arg_names[: len(expected_arg_names)], expected_arg_names)
 
     @slow
-    @require_torch_gpu
+    @require_torch_accelerator
     def test_model_from_pretrained(self):
         model_name = "Salesforce/blip2-itm-vit-g"
         model = Blip2TextModelWithProjection.from_pretrained(model_name)
@@ -2191,7 +2192,7 @@ def test_expansion_in_processing(self):
 
         self.assertTrue(generated_text_expanded == generated_text)
 
-    @require_torch_gpu
+    @require_torch_accelerator
     def test_inference_itm(self):
         model_name = "Salesforce/blip2-itm-vit-g"
         processor = Blip2Processor.from_pretrained(model_name)
@@ -2210,7 +2211,7 @@ def test_inference_itm(self):
         self.assertTrue(torch.allclose(torch.nn.Softmax()(out_itm[0].cpu()), expected_scores, rtol=1e-3, atol=1e-3))
         self.assertTrue(torch.allclose(out[0].cpu(), torch.Tensor([[0.4406]]), rtol=1e-3, atol=1e-3))
 
-    @require_torch_gpu
+    @require_torch_accelerator
     @require_torch_fp16
     def test_inference_itm_fp16(self):
         model_name = "Salesforce/blip2-itm-vit-g"
@@ -2232,7 +2233,7 @@ def test_inference_itm_fp16(self):
         )
         self.assertTrue(torch.allclose(out[0].cpu().float(), torch.Tensor([[0.4406]]), rtol=1e-3, atol=1e-3))
 
-    @require_torch_gpu
+    @require_torch_accelerator
     @require_torch_fp16
     def test_inference_vision_with_projection_fp16(self):
         model_name = "Salesforce/blip2-itm-vit-g"
@@ -2256,7 +2257,7 @@ def test_inference_vision_with_projection_fp16(self):
         ]
         self.assertTrue(np.allclose(out.image_embeds[0][0][:6].tolist(), expected_image_embeds, atol=1e-3))
 
-    @require_torch_gpu
+    @require_torch_accelerator
     @require_torch_fp16
     def test_inference_text_with_projection_fp16(self):
         model_name = "Salesforce/blip2-itm-vit-g"

tests/models/chameleon/test_modeling_chameleon.py

@@ -333,7 +333,7 @@ def test_batching_equivalence(self):
 
     # TODO (joao, raushan): fix me -- the problem is in `cache_position[0] == 0`, i.e. dynamic control flow
     @unittest.skip("Chameleon is not compatible with end-to-end generation compilation")
-    def test_generate_compile_fullgraph(self):
+    def test_generate_compile_model_forward(self):
         pass