[modular] Follow global indexing and attribute setting, and their dependencies (huggingface#39180)

* export global indexing statements

* add example

* style

* examples

Author: Cyrilvallez

examples/modular-transformers/modeling_global_indexing.py

@@ -0,0 +1,169 @@
+#                🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨
+#           This file was automatically generated from examples/modular-transformers/modular_global_indexing.py.
+#               Do NOT edit this file manually as any edits will be overwritten by the generation of
+#             the file from the modular. If any change should be done, please apply the change to the
+#                          modular_global_indexing.py file directly. One of our CI enforces this.
+#                🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨
+from typing import Callable, Optional
+
+import torch
+from torch import nn
+
+from transformers.modeling_utils import AttentionInterface
+
+from ...cache_utils import Cache
+from ...processing_utils import Unpack
+from ...utils import TransformersKwargs
+from .configuration_global_indexing import GlobalIndexingConfig
+
+
+def rotate_half(x):
+    """Rotates half the hidden dims of the input."""
+    x1 = x[..., : x.shape[-1] // 2]
+    x2 = x[..., x.shape[-1] // 2 :]
+    return torch.cat((-x2, x1), dim=-1)
+
+
+def apply_rotary_pos_emb(q, k, cos, sin, position_ids=None, unsqueeze_dim=1):
+    """Applies Rotary Position Embedding to the query and key tensors.
+
+    Args:
+        q (`torch.Tensor`): The query tensor.
+        k (`torch.Tensor`): The key tensor.
+        cos (`torch.Tensor`): The cosine part of the rotary embedding.
+        sin (`torch.Tensor`): The sine part of the rotary embedding.
+        position_ids (`torch.Tensor`, *optional*):
+            Deprecated and unused.
+        unsqueeze_dim (`int`, *optional*, defaults to 1):
+            The 'unsqueeze_dim' argument specifies the dimension along which to unsqueeze cos[position_ids] and
+            sin[position_ids] so that they can be properly broadcasted to the dimensions of q and k. For example, note
+            that cos[position_ids] and sin[position_ids] have the shape [batch_size, seq_len, head_dim]. Then, if q and
+            k have the shape [batch_size, heads, seq_len, head_dim], then setting unsqueeze_dim=1 makes
+            cos[position_ids] and sin[position_ids] broadcastable to the shapes of q and k. Similarly, if q and k have
+            the shape [batch_size, seq_len, heads, head_dim], then set unsqueeze_dim=2.
+    Returns:
+        `tuple(torch.Tensor)` comprising of the query and key tensors rotated using the Rotary Position Embedding.
+    """
+    cos = cos.unsqueeze(unsqueeze_dim)
+    sin = sin.unsqueeze(unsqueeze_dim)
+    q_embed = (q * cos) + (rotate_half(q) * sin)
+    k_embed = (k * cos) + (rotate_half(k) * sin)
+    return q_embed, k_embed
+
+
+def repeat_kv(hidden_states: torch.Tensor, n_rep: int) -> torch.Tensor:
+    """
+    This is the equivalent of torch.repeat_interleave(x, dim=1, repeats=n_rep). The hidden states go from (batch,
+    num_key_value_heads, seqlen, head_dim) to (batch, num_attention_heads, seqlen, head_dim)
+    """
+    batch, num_key_value_heads, slen, head_dim = hidden_states.shape
+    if n_rep == 1:
+        return hidden_states
+    hidden_states = hidden_states[:, :, None, :, :].expand(batch, num_key_value_heads, n_rep, slen, head_dim)
+    return hidden_states.reshape(batch, num_key_value_heads * n_rep, slen, head_dim)
+
+
+def eager_attention_forward(
+    module: nn.Module,
+    query: torch.Tensor,
+    key: torch.Tensor,
+    value: torch.Tensor,
+    attention_mask: Optional[torch.Tensor],
+    scaling: float,
+    dropout: float = 0.0,
+    **kwargs: Unpack[TransformersKwargs],
+):
+    key_states = repeat_kv(key, module.num_key_value_groups)
+    value_states = repeat_kv(value, module.num_key_value_groups)
+
+    attn_weights = torch.matmul(query, key_states.transpose(2, 3)) * scaling
+    if attention_mask is not None:
+        causal_mask = attention_mask[:, :, :, : key_states.shape[-2]]
+        attn_weights = attn_weights + causal_mask
+
+    attn_weights = nn.functional.softmax(attn_weights, dim=-1, dtype=torch.float32).to(query.dtype)
+    attn_weights = nn.functional.dropout(attn_weights, p=dropout, training=module.training)
+    attn_output = torch.matmul(attn_weights, value_states)
+    attn_output = attn_output.transpose(1, 2).contiguous()
+
+    return attn_output, attn_weights
+
+
+def custom_flex(x, **kwargs):
+    """Dummy function."""
+    return x
+
+
+ALL_ATTENTION_FUNCTIONS = AttentionInterface()
+# This indexing statement and associated function should be exported correctly!
+ALL_ATTENTION_FUNCTIONS["flex_attention"] = custom_flex
+
+
+class GlobalIndexingAttention(nn.Module):
+    """Multi-headed attention from 'Attention Is All You Need' paper"""
+
+    def __init__(self, config: GlobalIndexingConfig, layer_idx: int):
+        super().__init__()
+        self.config = config
+        self.layer_idx = layer_idx
+        self.head_dim = getattr(config, "head_dim", config.hidden_size // config.num_attention_heads)
+        self.num_key_value_groups = config.num_attention_heads // config.num_key_value_heads
+        self.scaling = self.head_dim**-0.5
+        self.attention_dropout = config.attention_dropout
+        self.is_causal = True
+
+        self.q_proj = nn.Linear(
+            config.hidden_size, config.num_attention_heads * self.head_dim, bias=config.attention_bias
+        )
+        self.k_proj = nn.Linear(
+            config.hidden_size, config.num_key_value_heads * self.head_dim, bias=config.attention_bias
+        )
+        self.v_proj = nn.Linear(
+            config.hidden_size, config.num_key_value_heads * self.head_dim, bias=config.attention_bias
+        )
+        self.o_proj = nn.Linear(
+            config.num_attention_heads * self.head_dim, config.hidden_size, bias=config.attention_bias
+        )
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        position_embeddings: tuple[torch.Tensor, torch.Tensor],
+        attention_mask: Optional[torch.Tensor],
+        past_key_value: Optional[Cache] = None,
+        cache_position: Optional[torch.LongTensor] = None,
+        **kwargs: Unpack[TransformersKwargs],
+    ) -> tuple[torch.Tensor, torch.Tensor]:
+        input_shape = hidden_states.shape[:-1]
+        hidden_shape = (*input_shape, -1, self.head_dim)
+
+        query_states = self.q_proj(hidden_states).view(hidden_shape).transpose(1, 2)
+        key_states = self.k_proj(hidden_states).view(hidden_shape).transpose(1, 2)
+        value_states = self.v_proj(hidden_states).view(hidden_shape).transpose(1, 2)
+
+        cos, sin = position_embeddings
+        query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin)
+
+        if past_key_value is not None:
+            # sin and cos are specific to RoPE models; cache_position needed for the static cache
+            cache_kwargs = {"sin": sin, "cos": cos, "cache_position": cache_position}
+            key_states, value_states = past_key_value.update(key_states, value_states, self.layer_idx, cache_kwargs)
+
+        attention_interface: Callable = eager_attention_forward
+        if self.config._attn_implementation != "eager":
+            attention_interface = ALL_ATTENTION_FUNCTIONS[self.config._attn_implementation]
+
+        attn_output, attn_weights = attention_interface(
+            self,
+            query_states,
+            key_states,
+            value_states,
+            attention_mask,
+            dropout=0.0 if not self.training else self.attention_dropout,
+            scaling=self.scaling,
+            **kwargs,
+        )
+
+        attn_output = attn_output.reshape(*input_shape, -1).contiguous()
+        attn_output = self.o_proj(attn_output)
+        return attn_output, attn_weights

examples/modular-transformers/modeling_multimodal2.py

@@ -289,7 +289,6 @@ def __init__(self, config):
         self.layers = nn.ModuleList([Multimodal2VisionEncoderLayer(config) for _ in range(config.num_hidden_layers)])
         self.gradient_checkpointing = False
 
-    @can_return_tuple
     def forward(
         self,
         inputs_embeds,
@@ -455,7 +454,6 @@ def __init__(self, config):
         self.encoder = Multimodal2VisionEncoder(config)
         self.post_layernorm = nn.LayerNorm(embed_dim, eps=config.layer_norm_eps)
 
-    @can_return_tuple
     @auto_docstring
     def forward(
         self,

examples/modular-transformers/modeling_my_new_model2.py

@@ -12,13 +12,13 @@
 from ...activations import ACT2FN
 from ...cache_utils import Cache, DynamicCache
 from ...masking_utils import create_causal_mask
-from ...modeling_flash_attention_utils import FlashAttentionKwargs
 from ...modeling_layers import GradientCheckpointingLayer
 from ...modeling_outputs import BaseModelOutputWithPast, SequenceClassifierOutputWithPast
 from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update
 from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel
 from ...processing_utils import Unpack
-from ...utils import auto_docstring, can_return_tuple, logging
+from ...utils import TransformersKwargs, auto_docstring, can_return_tuple, logging
+from ...utils.generic import check_model_inputs
 from .configuration_my_new_model2 import MyNewModel2Config
 
 
@@ -149,7 +149,7 @@ def eager_attention_forward(
     attention_mask: Optional[torch.Tensor],
     scaling: float,
     dropout: float = 0.0,
-    **kwargs,
+    **kwargs: Unpack[TransformersKwargs],
 ):
     key_states = repeat_kv(key, module.num_key_value_groups)
     value_states = repeat_kv(value, module.num_key_value_groups)
@@ -200,8 +200,8 @@ def forward(
         attention_mask: Optional[torch.Tensor],
         past_key_value: Optional[Cache] = None,
         cache_position: Optional[torch.LongTensor] = None,
-        **kwargs: Unpack[FlashAttentionKwargs],
-    ) -> tuple[torch.Tensor, Optional[torch.Tensor], Optional[tuple[torch.Tensor]]]:
+        **kwargs: Unpack[TransformersKwargs],
+    ) -> tuple[torch.Tensor, torch.Tensor]:
         input_shape = hidden_states.shape[:-1]
         hidden_shape = (*input_shape, -1, self.head_dim)
 
@@ -254,22 +254,19 @@ def forward(
         attention_mask: Optional[torch.Tensor] = None,
         position_ids: Optional[torch.LongTensor] = None,
         past_key_value: Optional[Cache] = None,
-        output_attentions: Optional[bool] = False,
         use_cache: Optional[bool] = False,
         cache_position: Optional[torch.LongTensor] = None,
         position_embeddings: Optional[tuple[torch.Tensor, torch.Tensor]] = None,  # necessary, but kept here for BC
-        **kwargs: Unpack[FlashAttentionKwargs],
-    ) -> tuple[torch.FloatTensor, Optional[tuple[torch.FloatTensor, torch.FloatTensor]]]:
+        **kwargs: Unpack[TransformersKwargs],
+    ) -> tuple[torch.Tensor]:
         residual = hidden_states
         hidden_states = self.input_layernorm(hidden_states)
-
         # Self Attention
-        hidden_states, self_attn_weights = self.self_attn(
+        hidden_states, _ = self.self_attn(
             hidden_states=hidden_states,
             attention_mask=attention_mask,
             position_ids=position_ids,
             past_key_value=past_key_value,
-            output_attentions=output_attentions,
             use_cache=use_cache,
             cache_position=cache_position,
             position_embeddings=position_embeddings,
@@ -282,12 +279,7 @@ def forward(
         hidden_states = self.post_attention_layernorm(hidden_states)
         hidden_states = self.mlp(hidden_states)
         hidden_states = residual + hidden_states
-
-        outputs = (hidden_states,)
-        if output_attentions:
-            outputs += (self_attn_weights,)
-
-        return outputs
+        return hidden_states
 
 
 @auto_docstring
@@ -304,6 +296,10 @@ class MyNewModel2PreTrainedModel(PreTrainedModel):
     _supports_quantized_cache = True
     _supports_static_cache = True
     _supports_attention_backend = True
+    _can_record_outputs = {
+        "hidden_states": MyNewModel2DecoderLayer,
+        "attentions": MyNewModel2Attention,
+    }
 
     def _init_weights(self, module):
         std = self.config.initializer_range
@@ -343,7 +339,7 @@ def get_input_embeddings(self):
     def set_input_embeddings(self, value):
         self.embed_tokens = value
 
-    @can_return_tuple
+    @check_model_inputs
     @auto_docstring
     def forward(
         self,
@@ -353,26 +349,12 @@ def forward(
         past_key_values: Optional[Cache] = None,
         inputs_embeds: Optional[torch.FloatTensor] = None,
         use_cache: Optional[bool] = None,
-        output_attentions: Optional[bool] = None,
-        output_hidden_states: Optional[bool] = None,
         cache_position: Optional[torch.LongTensor] = None,
-        **kwargs: Unpack[FlashAttentionKwargs],
+        **kwargs: Unpack[TransformersKwargs],
     ) -> BaseModelOutputWithPast:
-        output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
-        output_hidden_states = (
-            output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
-        )
-        use_cache = use_cache if use_cache is not None else self.config.use_cache
-
         if (input_ids is None) ^ (inputs_embeds is not None):
             raise ValueError("You must specify exactly one of input_ids or inputs_embeds")
 
-        if self.gradient_checkpointing and self.training and use_cache:
-            logger.warning_once(
-                "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`."
-            )
-            use_cache = False
-
         if inputs_embeds is None:
             inputs_embeds = self.embed_tokens(input_ids)
 
@@ -394,6 +376,7 @@ def forward(
             attention_mask=attention_mask,
             cache_position=cache_position,
             past_key_values=past_key_values,
+            position_ids=position_ids,
         )
 
         # embed positions
@@ -408,42 +391,21 @@ def forward(
         normalizer = torch.tensor(self.config.hidden_size**0.5, dtype=hidden_states.dtype)
         hidden_states = hidden_states * normalizer
 
-        # decoder layers
-        all_hidden_states = () if output_hidden_states else None
-        all_self_attns = () if output_attentions else None
-
         for decoder_layer in self.layers[: self.config.num_hidden_layers]:
-            if output_hidden_states:
-                all_hidden_states += (hidden_states,)
-
-            layer_outputs = decoder_layer(
+            hidden_states = decoder_layer(
                 hidden_states,
                 attention_mask=causal_mask,
                 position_ids=position_ids,
                 past_key_value=past_key_values,
-                output_attentions=output_attentions,
                 use_cache=use_cache,
                 cache_position=cache_position,
                 position_embeddings=position_embeddings,
                 **kwargs,
             )
-
-            hidden_states = layer_outputs[0]
-
-            if output_attentions:
-                all_self_attns += (layer_outputs[1],)
-
         hidden_states = self.norm(hidden_states)
-
-        # add hidden states from the last decoder layer
-        if output_hidden_states:
-            all_hidden_states += (hidden_states,)
-
         return BaseModelOutputWithPast(
             last_hidden_state=hidden_states,
             past_key_values=past_key_values if use_cache else None,
-            hidden_states=all_hidden_states,
-            attentions=all_self_attns,
         )
 
 
@@ -488,8 +450,7 @@ def forward(
         inputs_embeds: Optional[torch.FloatTensor] = None,
         labels: Optional[torch.LongTensor] = None,
         use_cache: Optional[bool] = None,
-        output_attentions: Optional[bool] = None,
-        output_hidden_states: Optional[bool] = None,
+        **kwargs: Unpack[TransformersKwargs],
     ) -> SequenceClassifierOutputWithPast:
         r"""
         labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
@@ -505,8 +466,7 @@ def forward(
             past_key_values=past_key_values,
             inputs_embeds=inputs_embeds,
             use_cache=use_cache,
-            output_attentions=output_attentions,
-            output_hidden_states=output_hidden_states,
+            **kwargs,
         )
         hidden_states = transformer_outputs.last_hidden_state
         logits = self.score(hidden_states)