Modular: support for importing functions from any file (#35692)

* fix function imports

* improve comment

* Update modeling_switch_function.py

* make checks more robust

* improvement

* rename

* final test update

examples/modular-transformers/modeling_add_function.py

@@ -0,0 +1,66 @@
+#                🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨
+#           This file was automatically generated from examples/modular-transformers/modular_add_function.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_add_function.py file directly. One of our CI enforces this.
+#                🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨
+# Note that zamba does not have the `apply_rotary_pos_emb` function!
+from typing import Optional, Tuple
+
+import torch
+from torch import nn
+
+
+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
+
+
+class TestAttention(nn.Module):
+    """
+    Multi-headed attention from 'Attention Is All You Need' paper. Modified to use sliding window attention: Longformer
+    and "Generating Long Sequences with Sparse Transformers".
+
+    Adapted from transformers.models.mistral.modeling_mistral.MistralAttention:
+    The input dimension here is attention_hidden_size = 2 * hidden_size, and head_dim = attention_hidden_size // num_heads.
+    The extra factor of 2 comes from the input being the concatenation of original_hidden_states with the output of the previous (mamba) layer
+    (see fig. 2 in https://arxiv.org/pdf/2405.16712).
+    Additionally, replaced
+    attn_weights = torch.matmul(query_states, key_states.transpose(2, 3)) / math.sqrt(self.head_dim) with
+    attn_weights = torch.matmul(query_states, key_states.transpose(2, 3)) / math.sqrt(self.head_dim/2)
+    """
+
+    def __init__(self):
+        pass
+
+    def forward(self) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
+        _ = apply_rotary_pos_emb(1, 1, 1, 1)

examples/modular-transformers/modeling_dummy.py

@@ -45,13 +45,8 @@ class DummyRMSNorm(nn.Module):
 
 
 class DummyRotaryEmbedding(nn.Module):
-    def __init__(
-        self,
-        config: DummyConfig,
-        device=None,
-    ):
+    def __init__(self, config: DummyConfig, device=None):
         super().__init__()
-        self.rope_kwargs = {}
         # BC: "rope_type" was originally "type"
         if hasattr(config, "rope_scaling") and config.rope_scaling is not None:
             self.rope_type = config.rope_scaling.get("rope_type", config.rope_scaling.get("type"))
@@ -63,7 +58,7 @@ class DummyRotaryEmbedding(nn.Module):
         self.config = config
         self.rope_init_fn = ROPE_INIT_FUNCTIONS[self.rope_type]
 
-        inv_freq, self.attention_scaling = self.rope_init_fn(self.config, device, **self.rope_kwargs)
+        inv_freq, self.attention_scaling = self.rope_init_fn(self.config, device)
         self.register_buffer("inv_freq", inv_freq, persistent=False)
         self.original_inv_freq = self.inv_freq
 
@@ -75,13 +70,14 @@ class DummyRotaryEmbedding(nn.Module):
         """
         seq_len = torch.max(position_ids) + 1
         if seq_len > self.max_seq_len_cached:  # growth
-            inv_freq, self.attention_scaling = self.rope_init_fn(
-                self.config, device, seq_len=seq_len, **self.rope_kwargs
-            )
+            inv_freq, self.attention_scaling = self.rope_init_fn(self.config, device, seq_len=seq_len)
             self.register_buffer("inv_freq", inv_freq, persistent=False)  # TODO joao: may break with compilation
             self.max_seq_len_cached = seq_len
 
         if seq_len < self.original_max_seq_len and self.max_seq_len_cached > self.original_max_seq_len:  # reset
+            # This .to() is needed if the model has been moved to a device after being initialized (because
+            # the buffer is automatically moved, but not the original copy)
+            self.original_inv_freq = self.original_inv_freq.to(device)
             self.register_buffer("inv_freq", self.original_inv_freq, persistent=False)
             self.max_seq_len_cached = self.original_max_seq_len
 
@@ -356,6 +352,7 @@ class DummyPreTrainedModel(PreTrainedModel):
     _skip_keys_device_placement = ["past_key_values"]
     _supports_flash_attn_2 = True
     _supports_sdpa = True
+    _supports_flex_attn = True
     _supports_cache_class = True
     _supports_quantized_cache = True
     _supports_static_cache = True

examples/modular-transformers/modeling_multimodal1.py

@@ -45,13 +45,8 @@ class Multimodal1TextRMSNorm(nn.Module):
 
 
 class Multimodal1TextRotaryEmbedding(nn.Module):
-    def __init__(
-        self,
-        config: Multimodal1TextConfig,
-        device=None,
-    ):
+    def __init__(self, config: Multimodal1TextConfig, device=None):
         super().__init__()
-        self.rope_kwargs = {}
         # BC: "rope_type" was originally "type"
         if hasattr(config, "rope_scaling") and config.rope_scaling is not None:
             self.rope_type = config.rope_scaling.get("rope_type", config.rope_scaling.get("type"))
@@ -63,7 +58,7 @@ class Multimodal1TextRotaryEmbedding(nn.Module):
         self.config = config
         self.rope_init_fn = ROPE_INIT_FUNCTIONS[self.rope_type]
 
-        inv_freq, self.attention_scaling = self.rope_init_fn(self.config, device, **self.rope_kwargs)
+        inv_freq, self.attention_scaling = self.rope_init_fn(self.config, device)
         self.register_buffer("inv_freq", inv_freq, persistent=False)
         self.original_inv_freq = self.inv_freq
 
@@ -75,13 +70,14 @@ class Multimodal1TextRotaryEmbedding(nn.Module):
         """
         seq_len = torch.max(position_ids) + 1
         if seq_len > self.max_seq_len_cached:  # growth
-            inv_freq, self.attention_scaling = self.rope_init_fn(
-                self.config, device, seq_len=seq_len, **self.rope_kwargs
-            )
+            inv_freq, self.attention_scaling = self.rope_init_fn(self.config, device, seq_len=seq_len)
             self.register_buffer("inv_freq", inv_freq, persistent=False)  # TODO joao: may break with compilation
             self.max_seq_len_cached = seq_len
 
         if seq_len < self.original_max_seq_len and self.max_seq_len_cached > self.original_max_seq_len:  # reset
+            # This .to() is needed if the model has been moved to a device after being initialized (because
+            # the buffer is automatically moved, but not the original copy)
+            self.original_inv_freq = self.original_inv_freq.to(device)
             self.register_buffer("inv_freq", self.original_inv_freq, persistent=False)
             self.max_seq_len_cached = self.original_max_seq_len
 
@@ -356,6 +352,7 @@ class Multimodal1TextPreTrainedModel(PreTrainedModel):
     _skip_keys_device_placement = ["past_key_values"]
     _supports_flash_attn_2 = True
     _supports_sdpa = True
+    _supports_flex_attn = True
     _supports_cache_class = True
     _supports_quantized_cache = True
     _supports_static_cache = True

examples/modular-transformers/modeling_my_new_model2.py

@@ -61,13 +61,8 @@ class MyNewModel2MLP(nn.Module):
 
 
 class MyNewModel2RotaryEmbedding(nn.Module):
-    def __init__(
-        self,
-        config: MyNewModel2Config,
-        device=None,
-    ):
+    def __init__(self, config: MyNewModel2Config, device=None):
         super().__init__()
-        self.rope_kwargs = {}
         # BC: "rope_type" was originally "type"
         if hasattr(config, "rope_scaling") and config.rope_scaling is not None:
             self.rope_type = config.rope_scaling.get("rope_type", config.rope_scaling.get("type"))
@@ -79,7 +74,7 @@ class MyNewModel2RotaryEmbedding(nn.Module):
         self.config = config
         self.rope_init_fn = ROPE_INIT_FUNCTIONS[self.rope_type]
 
-        inv_freq, self.attention_scaling = self.rope_init_fn(self.config, device, **self.rope_kwargs)
+        inv_freq, self.attention_scaling = self.rope_init_fn(self.config, device)
         self.register_buffer("inv_freq", inv_freq, persistent=False)
         self.original_inv_freq = self.inv_freq
 
@@ -91,13 +86,14 @@ class MyNewModel2RotaryEmbedding(nn.Module):
         """
         seq_len = torch.max(position_ids) + 1
         if seq_len > self.max_seq_len_cached:  # growth
-            inv_freq, self.attention_scaling = self.rope_init_fn(
-                self.config, device, seq_len=seq_len, **self.rope_kwargs
-            )
+            inv_freq, self.attention_scaling = self.rope_init_fn(self.config, device, seq_len=seq_len)
             self.register_buffer("inv_freq", inv_freq, persistent=False)  # TODO joao: may break with compilation
             self.max_seq_len_cached = seq_len
 
         if seq_len < self.original_max_seq_len and self.max_seq_len_cached > self.original_max_seq_len:  # reset
+            # This .to() is needed if the model has been moved to a device after being initialized (because
+            # the buffer is automatically moved, but not the original copy)
+            self.original_inv_freq = self.original_inv_freq.to(device)
             self.register_buffer("inv_freq", self.original_inv_freq, persistent=False)
             self.max_seq_len_cached = self.original_max_seq_len
 
@@ -356,6 +352,7 @@ class MyNewModel2PreTrainedModel(PreTrainedModel):
     _skip_keys_device_placement = ["past_key_values"]
     _supports_flash_attn_2 = True
     _supports_sdpa = True
+    _supports_flex_attn = True
     _supports_cache_class = True
     _supports_quantized_cache = True
     _supports_static_cache = True

examples/modular-transformers/modeling_new_task_model.py

@@ -107,7 +107,6 @@ class NewTaskModelPreTrainedModel(PreTrainedModel):
     _supports_cache_class = True
     _supports_quantized_cache = True
     _supports_static_cache = True
-    _supports_cache_class = True
     _supports_flash_attn_2 = True
     _supports_sdpa = True
 

examples/modular-transformers/modeling_super.py

@@ -45,13 +45,8 @@ class SuperRMSNorm(nn.Module):
 
 
 class SuperRotaryEmbedding(nn.Module):
-    def __init__(
-        self,
-        config: SuperConfig,
-        device=None,
-    ):
+    def __init__(self, config: SuperConfig, device=None):
         super().__init__()
-        self.rope_kwargs = {}
         # BC: "rope_type" was originally "type"
         if hasattr(config, "rope_scaling") and config.rope_scaling is not None:
             self.rope_type = config.rope_scaling.get("rope_type", config.rope_scaling.get("type"))
@@ -63,7 +58,7 @@ class SuperRotaryEmbedding(nn.Module):
         self.config = config
         self.rope_init_fn = ROPE_INIT_FUNCTIONS[self.rope_type]
 
-        inv_freq, self.attention_scaling = self.rope_init_fn(self.config, device, **self.rope_kwargs)
+        inv_freq, self.attention_scaling = self.rope_init_fn(self.config, device)
         self.register_buffer("inv_freq", inv_freq, persistent=False)
         self.original_inv_freq = self.inv_freq
 
@@ -75,13 +70,14 @@ class SuperRotaryEmbedding(nn.Module):
         """
         seq_len = torch.max(position_ids) + 1
         if seq_len > self.max_seq_len_cached:  # growth
-            inv_freq, self.attention_scaling = self.rope_init_fn(
-                self.config, device, seq_len=seq_len, **self.rope_kwargs
-            )
+            inv_freq, self.attention_scaling = self.rope_init_fn(self.config, device, seq_len=seq_len)
             self.register_buffer("inv_freq", inv_freq, persistent=False)  # TODO joao: may break with compilation
             self.max_seq_len_cached = seq_len
 
         if seq_len < self.original_max_seq_len and self.max_seq_len_cached > self.original_max_seq_len:  # reset
+            # This .to() is needed if the model has been moved to a device after being initialized (because
+            # the buffer is automatically moved, but not the original copy)
+            self.original_inv_freq = self.original_inv_freq.to(device)
             self.register_buffer("inv_freq", self.original_inv_freq, persistent=False)
             self.max_seq_len_cached = self.original_max_seq_len
 
@@ -356,6 +352,7 @@ class SuperPreTrainedModel(PreTrainedModel):
     _skip_keys_device_placement = ["past_key_values"]
     _supports_flash_attn_2 = True
     _supports_sdpa = True
+    _supports_flex_attn = True
     _supports_cache_class = True
     _supports_quantized_cache = True
     _supports_static_cache = True

examples/modular-transformers/modeling_switch_function.py

@@ -0,0 +1,170 @@
+#                🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨
+#           This file was automatically generated from examples/modular-transformers/modular_switch_function.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_switch_function.py file directly. One of our CI enforces this.
+#                🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨
+# Note that llama and cohere have different definitions for rotate_half
+from typing import Callable, Optional, Tuple
+
+import torch
+from torch import nn
+
+from ...cache_utils import Cache
+from ...modeling_flash_attention_utils import FlashAttentionKwargs
+from ...modeling_utils import ALL_ATTENTION_FUNCTIONS
+from ...processing_utils import Unpack
+from ...utils import logging
+from .configuration_switch_function import SwitchFunctionConfig
+
+
+logger = logging.get_logger(__name__)
+
+
+def rotate_half(x):
+    # Split and rotate. Note that this function is different from e.g. Llama.
+    x1 = x[..., ::2]
+    x2 = x[..., 1::2]
+    rot_x = torch.stack([-x2, x1], dim=-1).flatten(-2)
+    return rot_x
+
+
+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,
+):
+    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
+
+
+class SwitchFunctionAttention(nn.Module):
+    """Multi-headed attention from 'Attention Is All You Need' paper"""
+
+    def __init__(self, config: SwitchFunctionConfig, 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[FlashAttentionKwargs],
+    ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[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":
+            if self.config._attn_implementation == "sdpa" and kwargs.get("output_attentions", False):
+                logger.warning_once(
+                    "`torch.nn.functional.scaled_dot_product_attention` does not support `output_attentions=True`. Falling back to "
+                    'eager attention. This warning can be removed using the argument `attn_implementation="eager"` when loading the model.'
+                )
+            else:
+                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/modular_add_function.py

@@ -0,0 +1,15 @@
+# Note that zamba does not have the `apply_rotary_pos_emb` function!
+from transformers.models.llama.modeling_llama import apply_rotary_pos_emb
+from transformers.models.zamba.modeling_zamba import ZambaAttention
+
+
+# When following ZambaAttention dependencies, the function `apply_rotary_pos_emb` is not present
+# by default as it is absent from the class definition (and the file altogether).
+# Note that this syntax should be able to add both `apply_rotary_pos_emb` as imported directly, but
+# `rotate_half` as well as a dependency from the imported function!!
+class TestAttention(ZambaAttention):
+    def __init__(self):
+        pass
+
+    def forward(self):
+        _ = apply_rotary_pos_emb(1, 1, 1, 1)

examples/modular-transformers/modular_switch_function.py

@@ -0,0 +1,10 @@
+# Note that llama and cohere have different definitions for rotate_half
+from transformers.models.cohere.modeling_cohere import rotate_half  # noqa
+from transformers.models.llama.modeling_llama import LlamaAttention
+
+
+# When following LlamaAttention dependencies, we will grab the function `rotate_half` defined
+# in `modeling_llama.py`. But here we imported it explicitly from Cohere, so it should use Cohere's
+# definition instead
+class SwitchFunctionAttention(LlamaAttention):
+    pass

utils/modular_model_converter.py

@@ -776,7 +776,7 @@ class ModelFileMapper(ModuleMapper):
                 else:
                     merged_dependencies.append(class_dep)
             # Sort both list according to the order in their respective file
-            original_dependencies = sorted(original_dependencies, key=lambda x: self.start_lines[x])
+            original_dependencies = sorted(original_dependencies, key=lambda x: self.start_lines.get(x, 1e10))
             merged_dependencies = sorted(merged_dependencies, key=lambda x: self.modular_file_start_lines[x])
 
             # Add all original node first, then merged ones
@@ -801,7 +801,7 @@ class ModelFileMapper(ModuleMapper):
             else:
                 original_dependencies.append(dep)
         # Sort both list according to the order in their respective file
-        original_dependencies = sorted(original_dependencies, key=lambda x: self.start_lines[x])
+        original_dependencies = sorted(original_dependencies, key=lambda x: self.start_lines.get(x, 1e10))
         merged_dependencies = sorted(merged_dependencies, key=lambda x: self.modular_file_start_lines[x])
 
         # Add all original node first, then merged ones
@@ -1321,6 +1321,20 @@ class ModularFileMapper(ModuleMapper):
                             self.added_objects_file_mapping[dep] = file
                             self.functions[dep] = visited_module.global_nodes[dep]
 
+                # Add/overwrite the imported functions to other visited modules as well, in case it is absent/different
+                # in he modeling source file of the inherited class. See `examples/modular-tranformers/modular_switch_function.py`
+                # and `examples/modular-tranformers/modular_add_function.py` for examples
+                recursive_dependencies = visited_module.object_recursive_dependency_mapping.get(object_name, set())
+                node_recursive_dependencies_mapping = {
+                    dep: visited_module.global_nodes[dep] for dep in recursive_dependencies
+                }
+                for filename, module_mapper in self.visited_modules.items():
+                    if filename != file:
+                        module_mapper.global_nodes[object_name] = visited_module.functions[object_name]
+                        if len(recursive_dependencies) > 0:
+                            module_mapper.object_recursive_dependency_mapping[object_name] = recursive_dependencies
+                            module_mapper.global_nodes.update(node_recursive_dependencies_mapping)
+
             # Add assignments and their dependencies
             elif object_name in visited_module.assignments and object_name not in self.assignments:
                 self.assignments[object_name] = visited_module.assignments[object_name]