updates

src/transformers/models/cohere/configuration_cohere.py

@@ -1,10 +1,11 @@
-# coding=utf-8
+#           🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨
-# Copyright 2024 Cohere team. All rights reserved.
+#               This file was automatically generated from <path_to_modular_file.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_xxx.py file directly. One of our CI enforces this
+#           🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨
+# Copyright 2024 Google Inc. HuggingFace Inc. team. All rights reserved.
 #
-# This code is based on EleutherAI's GPT-NeoX library and the GPT-NeoX
-# and OPT implementations in this library. It has been modified from its
-# original forms to accommodate minor architectural differences compared
-# to GPT-NeoX and OPT used by the Meta AI team that trained the model.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -17,41 +18,32 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
-"""Cohere model configuration"""
+
 
 from ...configuration_utils import PretrainedConfig
-from ...modeling_rope_utils import rope_config_validation
-from ...utils import logging
-
-
-logger = logging.get_logger(__name__)
 
 
 class CohereConfig(PretrainedConfig):
     r"""
     This is the configuration class to store the configuration of a [`CohereModel`]. It is used to instantiate an Cohere
-    model according to the specified arguments, defining the model architecture.
+    model according to the specified arguments, defining the model architecture. Instantiating a configuration with the
-
+    defaults will yield a similar configuration to that of the Cohere-7B.
+    e.g. [google/cohere-7b](https://huggingface.co/google/cohere-7b)
     Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
-    documentation from [`PretrainedConfig`] for more information. Instantiating a configuration
+    documentation from [`PretrainedConfig`] for more information.
-    with the defaults will yield a similar configuration to that of the [CohereForAI/c4ai-command-r-v01](https://huggingface.co/CohereForAI/c4ai-command-r-v01) model.
-
-
     Args:
         vocab_size (`int`, *optional*, defaults to 256000):
             Vocabulary size of the Cohere model. Defines the number of different tokens that can be represented by the
             `inputs_ids` passed when calling [`CohereModel`]
-        hidden_size (`int`, *optional*, defaults to 8192):
+        hidden_size (`int`, *optional*, defaults to 3072):
             Dimension of the hidden representations.
-        intermediate_size (`int`, *optional*, defaults to 22528):
+        intermediate_size (`int`, *optional*, defaults to 24576):
             Dimension of the MLP representations.
-        logit_scale (`float`, *optional*, defaults to 0.0625):
+        num_hidden_layers (`int`, *optional*, defaults to 28):
-            The scaling factor for the output logits.
-        num_hidden_layers (`int`, *optional*, defaults to 40):
             Number of hidden layers in the Transformer decoder.
-        num_attention_heads (`int`, *optional*, defaults to 64):
+        num_attention_heads (`int`, *optional*, defaults to 16):
             Number of attention heads for each attention layer in the Transformer decoder.
-        num_key_value_heads (`int`, *optional*):
+        num_key_value_heads (`int`, *optional*, defaults to 16):
             This is the number of key_value heads that should be used to implement Grouped Query Attention. If
             `num_key_value_heads=num_attention_heads`, the model will use Multi Head Attention (MHA), if
             `num_key_value_heads=1` the model will use Multi Query Attention (MQA) otherwise GQA is used. When
@@ -59,83 +51,48 @@ class CohereConfig(PretrainedConfig):
             by meanpooling all the original heads within that group. For more details checkout [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to
             `num_attention_heads`.
-        hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
+        head_dim (`int`, *optional*, defaults to 256):
-            The non-linear activation function (function or string) in the decoder.
+            The attention head dimension.
+        hidden_act (`str` or `function`, *optional*, defaults to `"gelu_pytorch_tanh"`):
+            The legacy activation function. It is overwritten by the `hidden_activation`.
+        hidden_activation (`str` or `function`, *optional*):
+            The non-linear activation function (function or string) in the decoder. Will default to `"gelu_pytorch_tanh"`
+            if not specified. `"gelu_pytorch_tanh"` uses an approximation of the `"gelu"` activation function.
         max_position_embeddings (`int`, *optional*, defaults to 8192):
             The maximum sequence length that this model might ever be used with.
         initializer_range (`float`, *optional*, defaults to 0.02):
             The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
-        layer_norm_eps (`float`, *optional*, defaults to 1e-05):
+        rms_norm_eps (`float`, *optional*, defaults to 1e-06):
-            The epsilon used by the layer normalization.
+            The epsilon used by the rms normalization layers.
         use_cache (`bool`, *optional*, defaults to `True`):
             Whether or not the model should return the last key/values attentions (not used by all models). Only
             relevant if `config.is_decoder=True`.
         pad_token_id (`int`, *optional*, defaults to 0):
             Padding token id.
-        bos_token_id (`int`, *optional*, defaults to 5):
+        eos_token_id (`int`, *optional*, defaults to 1):
-            Beginning of stream token id.
-        eos_token_id (`int`, *optional*, defaults to 255001):
             End of stream token id.
+        bos_token_id (`int`, *optional*, defaults to 2):
+            Beginning of stream token id.
         tie_word_embeddings (`bool`, *optional*, defaults to `True`):
             Whether to tie weight embeddings
         rope_theta (`float`, *optional*, defaults to 10000.0):
             The base period of the RoPE embeddings.
-        rope_scaling (`Dict`, *optional*):
-            Dictionary containing the scaling configuration for the RoPE embeddings. NOTE: if you apply new rope type
-            and you expect the model to work on longer `max_position_embeddings`, we recommend you to update this value
-            accordingly.
-            Expected contents:
-                `rope_type` (`str`):
-                    The sub-variant of RoPE to use. Can be one of ['default', 'linear', 'dynamic', 'yarn', 'longrope',
-                    'llama3'], with 'default' being the original RoPE implementation.
-                `factor` (`float`, *optional*):
-                    Used with all rope types except 'default'. The scaling factor to apply to the RoPE embeddings. In
-                    most scaling types, a `factor` of x will enable the model to handle sequences of length x *
-                    original maximum pre-trained length.
-                `original_max_position_embeddings` (`int`, *optional*):
-                    Used with 'dynamic', 'longrope' and 'llama3'. The original max position embeddings used during
-                    pretraining.
-                `attention_factor` (`float`, *optional*):
-                    Used with 'yarn' and 'longrope'. The scaling factor to be applied on the attention
-                    computation. If unspecified, it defaults to value recommended by the implementation, using the
-                    `factor` field to infer the suggested value.
-                `beta_fast` (`float`, *optional*):
-                    Only used with 'yarn'. Parameter to set the boundary for extrapolation (only) in the linear
-                    ramp function. If unspecified, it defaults to 32.
-                `beta_slow` (`float`, *optional*):
-                    Only used with 'yarn'. Parameter to set the boundary for interpolation (only) in the linear
-                    ramp function. If unspecified, it defaults to 1.
-                `short_factor` (`List[float]`, *optional*):
-                    Only used with 'longrope'. The scaling factor to be applied to short contexts (<
-                    `original_max_position_embeddings`). Must be a list of numbers with the same length as the hidden
-                    size divided by the number of attention heads divided by 2
-                `long_factor` (`List[float]`, *optional*):
-                    Only used with 'longrope'. The scaling factor to be applied to long contexts (<
-                    `original_max_position_embeddings`). Must be a list of numbers with the same length as the hidden
-                    size divided by the number of attention heads divided by 2
-                `low_freq_factor` (`float`, *optional*):
-                    Only used with 'llama3'. Scaling factor applied to low frequency components of the RoPE
-                `high_freq_factor` (`float`, *optional*):
-                    Only used with 'llama3'. Scaling factor applied to high frequency components of the RoPE
         attention_bias (`bool`, defaults to `False`, *optional*, defaults to `False`):
             Whether to use a bias in the query, key, value and output projection layers during self-attention.
         attention_dropout (`float`, *optional*, defaults to 0.0):
             The dropout ratio for the attention probabilities.
-        use_qk_norm (`bool`, *optional*, defaults to `False`):
-            Whether to use query-key normalization in the attention
-
     ```python
     >>> from transformers import CohereModel, CohereConfig
-
+    >>> # Initializing a Cohere cohere-7b style configuration
-    >>> # Initializing a Cohere model configuration
     >>> configuration = CohereConfig()
-
+    >>> # Initializing a model from the cohere-7b style configuration
-    >>> # Initializing a model from the Cohere configuration
+    >>> model = CohereModel(configuration)
-    >>> model = CohereModel(configuration) # doctest: +SKIP
-
     >>> # Accessing the model configuration
-    >>> configuration = model.config # doctest: +SKIP
+    >>> configuration = model.config
-    ```"""
+    ```
+        use_qk_norm (`bool`, *optional*, defaults to `False`):
+            Whether to use query-key normalization in the attention
+    """
 
     model_type = "cohere"
     keys_to_ignore_at_inference = ["past_key_values"]
@@ -145,51 +102,57 @@ class CohereConfig(PretrainedConfig):
         vocab_size=256000,
         hidden_size=8192,
         intermediate_size=22528,
-        logit_scale=0.0625,
         num_hidden_layers=40,
         num_attention_heads=64,
         num_key_value_heads=None,
+        head_dim=256,
         hidden_act="silu",
+        hidden_activation=None,
         max_position_embeddings=8192,
         initializer_range=0.02,
-        layer_norm_eps=1e-5,
+        rms_norm_eps=1e-05,
         use_cache=True,
         pad_token_id=0,
-        bos_token_id=5,
         eos_token_id=255001,
+        bos_token_id=5,
         tie_word_embeddings=True,
         rope_theta=10000.0,
-        rope_scaling=None,
         attention_bias=False,
         attention_dropout=0.0,
         use_qk_norm=False,
+        layer_norm_eps=1e-05,
+        logit_scale=0.0625,
         **kwargs,
     ):
         self.vocab_size = vocab_size
         self.max_position_embeddings = max_position_embeddings
+
         self.hidden_size = hidden_size
-        self.logit_scale = logit_scale
         self.intermediate_size = intermediate_size
         self.num_hidden_layers = num_hidden_layers
         self.num_attention_heads = num_attention_heads
-
+        self.head_dim = head_dim
-        # for backward compatibility
-        if num_key_value_heads is None:
-            num_key_value_heads = num_attention_heads
-
         self.num_key_value_heads = num_key_value_heads
+
         self.hidden_act = hidden_act
+        self.hidden_activation = hidden_activation
         self.initializer_range = initializer_range
-        self.layer_norm_eps = layer_norm_eps
+        self.rms_norm_eps = layer_norm_eps
         self.use_cache = use_cache
         self.rope_theta = rope_theta
-        self.rope_scaling = rope_scaling
         self.attention_bias = attention_bias
         self.attention_dropout = attention_dropout
         self.use_qk_norm = use_qk_norm
+        self.layer_norm_eps = layer_norm_eps
+        self.itermediate_size = intermediate_size
 
-        # Validate the correctness of rotary position embeddings parameters
+        self.bos_token_id = bos_token_id
-        rope_config_validation(self)
+        self.eos_token_id = eos_token_id
+
+        self.logit_scale = logit_scale
+
+        if num_key_value_heads is None:
+            self.num_key_value_heads = num_attention_heads
 
         super().__init__(
             pad_token_id=pad_token_id,

src/transformers/models/cohere/modeling_cohere.py

@@ -1,10 +1,11 @@
-# coding=utf-8
+#           🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨
-# Copyright 2024 Cohere team. All rights reserved.
+#               This file was automatically generated from <path_to_modular_file.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_xxx.py file directly. One of our CI enforces this
+#           🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨
+# Copyright 2024 Google Inc. HuggingFace Inc. team. All rights reserved.
 #
-# This code is based on EleutherAI's GPT-NeoX library and the GPT-NeoX
-# and OPT implementations in this library. It has been modified from its
-# original forms to accommodate minor architectural differences compared
-# to GPT-NeoX and OPT used by the Meta AI team that trained the model.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -17,11 +18,6 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
-
-# This file is based on the LLama model definition file in transformers
-
-"""PyTorch Cohere model."""
-
 import math
 from typing import List, Optional, Tuple, Union
 
@@ -38,31 +34,21 @@ from ...modeling_outputs import (
     BaseModelOutputWithPast,
     CausalLMOutputWithPast,
 )
-from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS
 from ...modeling_utils import PreTrainedModel
-from ...pytorch_utils import ALL_LAYERNORM_LAYERS
 from ...utils import (
     add_start_docstrings,
     add_start_docstrings_to_model_forward,
-    is_flash_attn_2_available,
-    is_flash_attn_greater_or_equal_2_10,
-    is_torchdynamo_compiling,
     logging,
     replace_return_docstrings,
 )
 from .configuration_cohere import CohereConfig
 
 
-if is_flash_attn_2_available():
-    from ...modeling_flash_attention_utils import _flash_attention_forward
-
-
 logger = logging.get_logger(__name__)
 
 _CONFIG_FOR_DOC = "CohereConfig"
 
 
-# Copied from transformers.models.llama.modeling_llama._prepare_4d_causal_attention_mask_with_cache_position
 def _prepare_4d_causal_attention_mask_with_cache_position(
     attention_mask: torch.Tensor,
     sequence_length: int,
@@ -116,7 +102,7 @@ def _prepare_4d_causal_attention_mask_with_cache_position(
     return causal_mask
 
 
-class CohereLayerNorm(nn.Module):
+class CohereRMSNorm(nn.Module):
     def __init__(self, hidden_size=None, eps=1e-5, bias=False):
         """The hidden size can be a tuple or an int. The tuple is used for QKNorm to normalize across head_dim"""
         super().__init__()
@@ -133,105 +119,45 @@ class CohereLayerNorm(nn.Module):
         return hidden_states.to(input_dtype)
 
 
-ALL_LAYERNORM_LAYERS.append(CohereLayerNorm)
+class CohereLayerNorm(CohereRMSNorm):
+    def __init__(self, *args, **kwargs):
+        logger.warning_once("CohereLayerNorm is deprecated by CohereRMSNorm and will be removed in v4.45.")
+        super().__init__(*args, **kwargs)
 
 
 class CohereRotaryEmbedding(nn.Module):
-    # Note: the forward pass of this RoPE is slightly different from Llama's, resulting in different `sin`/`cos` for
+    def __init__(self, dim, max_position_embeddings=2048, base=10000, device=None, scaling_factor=1.0):
-    # the same parameterization. The differences are highlighted with a comment.
-
-    def __init__(
-        self,
-        dim=None,
-        max_position_embeddings=2048,
-        base=10000,
-        device=None,
-        scaling_factor=1.0,
-        rope_type="default",
-        config: Optional[CohereConfig] = None,
-    ):
         super().__init__()
-        # TODO (joao): remove the `if` below, only used for BC
+        self.scaling_factor = scaling_factor
-        self.rope_kwargs = {}
+        self.dim = dim
-        if config is None:
+        self.max_position_embeddings = max_position_embeddings
-            logger.warning_once(
+        self.base = base
-                "`CohereRotaryEmbedding` can now be fully parameterized by passing the model config through the "
+        inv_freq = 1.0 / (self.base ** (torch.arange(0, self.dim, 2, dtype=torch.int64).float().to(device) / self.dim))
-                "`config` argument. All other arguments will be removed in v4.46"
-            )
-            self.rope_kwargs = {
-                "rope_type": rope_type,
-                "factor": scaling_factor,
-                "dim": dim,
-                "base": base,
-                "max_position_embeddings": max_position_embeddings,
-            }
-            self.rope_type = rope_type
-            self.max_seq_len_cached = max_position_embeddings
-            self.original_max_seq_len = max_position_embeddings
-        else:
-            # BC: "rope_type" was originally "type"
-            if config.rope_scaling is not None:
-                self.rope_type = config.rope_scaling.get("rope_type", config.rope_scaling.get("type"))
-            else:
-                self.rope_type = "default"
-            self.max_seq_len_cached = config.max_position_embeddings
-            self.original_max_seq_len = config.max_position_embeddings
-
-        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)
         self.register_buffer("inv_freq", inv_freq, persistent=False)
-        self.original_inv_freq = self.inv_freq
-
-    def _dynamic_frequency_update(self, position_ids, device):
-        """
-        dynamic RoPE layers should recompute `inv_freq` in the following situations:
-        1 - growing beyond the cached sequence length (allow scaling)
-        2 - the current sequence length is in the original scale (avoid losing precision with small sequences)
-        """
-        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
-            )
-            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
-            self.register_buffer("inv_freq", self.original_inv_freq, persistent=False)
-            self.max_seq_len_cached = self.original_max_seq_len
 
     @torch.no_grad()
     def forward(self, x, position_ids):
-        if "dynamic" in self.rope_type:
+        # x: [bs, num_attention_heads, seq_len, head_size]
-            self._dynamic_frequency_update(position_ids, device=x.device)
-
-        # Core RoPE block
         inv_freq_expanded = self.inv_freq[None, :, None].float().expand(position_ids.shape[0], -1, 1)
         position_ids_expanded = position_ids[:, None, :].float()
-        # Force float32 (see https://github.com/huggingface/transformers/pull/29285)
+
+        # Force float32 since bfloat16 loses precision on long contexts
+        # See https://github.com/huggingface/transformers/pull/29285
         device_type = x.device.type
         device_type = device_type if isinstance(device_type, str) and device_type != "mps" else "cpu"
         with torch.autocast(device_type=device_type, enabled=False):
             freqs = (inv_freq_expanded.float() @ position_ids_expanded.float()).transpose(1, 2)
-            emb = torch.repeat_interleave(freqs, 2, dim=-1)  # This line differs from Llama's implementation
+            emb = torch.repeat_interleave(freqs, 2, dim=-1)
             cos = emb.cos()
             sin = emb.sin()
-
+        return cos, sin
-        # Advanced RoPE types (e.g. yarn) apply a post-processing scaling factor, equivalent to scaling attention
-        cos = cos * self.attention_scaling
-        sin = sin * self.attention_scaling
-
-        return cos.to(dtype=x.dtype), sin.to(dtype=x.dtype)
 
 
 def rotate_half(x):
-    # Split and rotate. Note that this function is different from e.g. Llama.
+    """Rotates half the hidden dims of the input."""
-    x1 = x[..., ::2]
+    x1 = x[..., : x.shape[-1] // 2]
-    x2 = x[..., 1::2]
+    x2 = x[..., x.shape[-1] // 2 :]
-    rot_x = torch.stack([-x2, x1], dim=-1).flatten(-2)
+    return torch.cat((-x2, x1), dim=-1)
-    return rot_x
 
 
 def apply_rotary_pos_emb(q, k, cos, sin, position_ids=None, unsqueeze_dim=1):
@@ -254,14 +180,11 @@ def apply_rotary_pos_emb(q, k, cos, sin, position_ids=None, unsqueeze_dim=1):
     Returns:
         `tuple(torch.Tensor)` comprising of the query and key tensors rotated using the Rotary Position Embedding.
     """
-    dtype = q.dtype
-    q = q.float()
-    k = k.float()
     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.to(dtype=dtype), k_embed.to(dtype=dtype)
+    return q_embed, k_embed
 
 
 class CohereMLP(nn.Module):
@@ -275,13 +198,11 @@ class CohereMLP(nn.Module):
         self.down_proj = nn.Linear(self.intermediate_size, self.hidden_size, bias=False)
         self.act_fn = ACT2FN[config.hidden_act]
 
-    # Ignore copy
     def forward(self, x):
         down_proj = self.down_proj(self.act_fn(self.gate_proj(x)) * self.up_proj(x))
         return down_proj
 
 
-# Copied from transformers.models.llama.modeling_llama.repeat_kv
 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,
@@ -336,9 +257,14 @@ class CohereAttention(nn.Module):
         self.k_proj = nn.Linear(self.hidden_size, self.num_key_value_heads * self.head_dim, bias=config.attention_bias)
         self.v_proj = nn.Linear(self.hidden_size, self.num_key_value_heads * self.head_dim, bias=config.attention_bias)
         self.o_proj = nn.Linear(self.hidden_size, self.hidden_size, bias=config.attention_bias)
+        self._init_rope()
 
-        # TODO (joao): remove in v4.46 (RoPE is computed in the model, not in the decoder layers)
+    def _init_rope(self):
-        self.rotary_emb = CohereRotaryEmbedding(config=self.config)
+        self.rotary_emb = CohereRotaryEmbedding(
+            self.head_dim,
+            max_position_embeddings=self.max_position_embeddings,
+            base=self.rope_theta,
+        )
 
     def forward(
         self,
@@ -349,7 +275,6 @@ class CohereAttention(nn.Module):
         output_attentions: bool = False,
         use_cache: bool = False,
         cache_position: Optional[torch.LongTensor] = None,
-        position_embeddings: Optional[Tuple[torch.Tensor, torch.Tensor]] = None,  # will become mandatory in v4.46
         **kwargs,
     ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
         bsz, q_len, _ = hidden_states.size()
@@ -368,16 +293,7 @@ class CohereAttention(nn.Module):
         key_states = key_states.transpose(1, 2)
         value_states = value_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
 
-        if position_embeddings is None:
-            logger.warning_once(
-                "The attention layers in this model are transitioning from computing the RoPE embeddings internally "
-                "through `position_ids` (2D tensor with the indexes of the tokens), to using externally computed "
-                "`position_embeddings` (Tuple of tensors, containing cos and sin). In v4.46 `position_ids` will be "
-                "removed and `position_embeddings` will be mandatory."
-            )
         cos, sin = self.rotary_emb(value_states, position_ids)
-        else:
-            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:
@@ -417,136 +333,7 @@ class CohereAttention(nn.Module):
         return attn_output, attn_weights, past_key_value
 
 
-# Copied from transformers.models.llama.modeling_llama.LlamaFlashAttention2 with Llama->Cohere
-class CohereFlashAttention2(CohereAttention):
-    """
-    Cohere flash attention module. This module inherits from `CohereAttention` as the weights of the module stays
-    untouched. The only required change would be on the forward pass where it needs to correctly call the public API of
-    flash attention and deal with padding tokens in case the input contains any of them.
-    """
-
-    def __init__(self, *args, **kwargs):
-        super().__init__(*args, **kwargs)
-
-        # TODO: Should be removed once Flash Attention for RoCm is bumped to 2.1.
-        # flash_attn<2.1 generates top-left aligned causal mask, while what is needed here is bottom-right alignement, that was made default for flash_attn>=2.1. This attribute is used to handle this difference. Reference: https://github.com/Dao-AILab/flash-attention/releases/tag/v2.1.0.
-        # Beware that with flash_attn<2.1, using q_seqlen != k_seqlen (except for the case q_seqlen == 1) produces a wrong mask (top-left).
-        self._flash_attn_uses_top_left_mask = not is_flash_attn_greater_or_equal_2_10()
-
-    # Ignore copy
-    def forward(
-        self,
-        hidden_states: torch.Tensor,
-        attention_mask: Optional[torch.LongTensor] = None,
-        position_ids: Optional[torch.LongTensor] = None,
-        past_key_value: Optional[Cache] = None,
-        output_attentions: bool = False,
-        use_cache: bool = False,
-        cache_position: Optional[torch.LongTensor] = None,
-        position_embeddings: Optional[Tuple[torch.Tensor, torch.Tensor]] = None,  # will become mandatory in v4.46
-        **kwargs,
-    ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
-        if isinstance(past_key_value, StaticCache):
-            raise ValueError(
-                "`static` cache implementation is not compatible with `attn_implementation==flash_attention_2` "
-                "make sure to use `sdpa` in the mean time, and open an issue at https://github.com/huggingface/transformers"
-            )
-        output_attentions = False
-
-        bsz, q_len, _ = hidden_states.size()
-
-        query_states = self.q_proj(hidden_states)
-        key_states = self.k_proj(hidden_states)
-        value_states = self.v_proj(hidden_states)
-
-        query_states = query_states.view(bsz, q_len, self.num_heads, self.head_dim)
-        key_states = key_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim)
-        if self.use_qk_norm:
-            query_states = self.q_norm(query_states)
-            key_states = self.k_norm(key_states)
-
-        query_states = query_states.transpose(1, 2)
-        key_states = key_states.transpose(1, 2)
-        value_states = value_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
-
-        if position_embeddings is None:
-            logger.warning_once(
-                "The attention layers in this model are transitioning from computing the RoPE embeddings internally "
-                "through `position_ids` (2D tensor with the indexes of the tokens), to using externally computed "
-                "`position_embeddings` (Tuple of tensors, containing cos and sin). In v4.46 `position_ids` will be "
-                "removed and `position_embeddings` will be mandatory."
-            )
-            cos, sin = self.rotary_emb(value_states, position_ids)
-        else:
-            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; position_ids 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)
-
-        # TODO: These transpose are quite inefficient but Flash Attention requires the layout [batch_size, sequence_length, num_heads, head_dim]. We would need to refactor the KV cache
-        # to be able to avoid many of these transpose/reshape/view.
-        query_states = query_states.transpose(1, 2)
-        key_states = key_states.transpose(1, 2)
-        value_states = value_states.transpose(1, 2)
-
-        dropout_rate = self.attention_dropout if self.training else 0.0
-
-        # In PEFT, usually we cast the layer norms in float32 for training stability reasons
-        # therefore the input hidden states gets silently casted in float32. Hence, we need
-        # cast them back in the correct dtype just to be sure everything works as expected.
-        # This might slowdown training & inference so it is recommended to not cast the LayerNorms
-        # in fp32. (CohereLayerNorm handles it correctly)
-
-        input_dtype = query_states.dtype
-        if input_dtype == torch.float32:
-            if torch.is_autocast_enabled():
-                target_dtype = torch.get_autocast_gpu_dtype()
-            # Handle the case where the model is quantized
-            elif hasattr(self.config, "_pre_quantization_dtype"):
-                target_dtype = self.config._pre_quantization_dtype
-            else:
-                target_dtype = self.q_proj.weight.dtype
-
-            logger.warning_once(
-                f"The input hidden states seems to be silently casted in float32, this might be related to"
-                f" the fact you have upcasted embedding or layer norm layers in float32. We will cast back the input in"
-                f" {target_dtype}."
-            )
-
-            query_states = query_states.to(target_dtype)
-            key_states = key_states.to(target_dtype)
-            value_states = value_states.to(target_dtype)
-
-        attn_output = _flash_attention_forward(
-            query_states,
-            key_states,
-            value_states,
-            attention_mask,
-            q_len,
-            dropout=dropout_rate,
-            use_top_left_mask=self._flash_attn_uses_top_left_mask,
-            is_causal=self.is_causal,
-        )
-
-        attn_output = attn_output.reshape(bsz, q_len, self.hidden_size).contiguous()
-        attn_output = self.o_proj(attn_output)
-
-        if not output_attentions:
-            attn_weights = None
-
-        return attn_output, attn_weights, past_key_value
-
-
 class CohereSdpaAttention(CohereAttention):
-    """
-    Cohere attention module using torch.nn.functional.scaled_dot_product_attention. This module inherits from
-    `CohereAttention` as the weights of the module stays untouched. The only changes are on the forward pass to adapt to
-    SDPA API.
-    """
-
     def forward(
         self,
         hidden_states: torch.Tensor,
@@ -556,7 +343,6 @@ class CohereSdpaAttention(CohereAttention):
         output_attentions: bool = False,
         use_cache: bool = False,
         cache_position: Optional[torch.LongTensor] = None,
-        position_embeddings: Optional[Tuple[torch.Tensor, torch.Tensor]] = None,  # will become mandatory in v4.46
     ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
         if output_attentions:
             # TODO: Improve this warning with e.g. `model.config.attn_implementation = "manual"` once this is implemented.
@@ -590,16 +376,7 @@ class CohereSdpaAttention(CohereAttention):
         key_states = key_states.transpose(1, 2)
         value_states = value_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
 
-        if position_embeddings is None:
-            logger.warning_once(
-                "The attention layers in this model are transitioning from computing the RoPE embeddings internally "
-                "through `position_ids` (2D tensor with the indexes of the tokens), to using externally computed "
-                "`position_embeddings` (Tuple of tensors, containing cos and sin). In v4.46 `position_ids` will be "
-                "removed and `position_embeddings` will be mandatory."
-            )
         cos, sin = self.rotary_emb(value_states, position_ids)
-        else:
-            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:
@@ -643,6 +420,97 @@ class CohereSdpaAttention(CohereAttention):
         return attn_output, None, past_key_value
 
 
+class CohereFlashAttention2(CohereAttention):
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: Optional[torch.LongTensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_value: Optional[Cache] = None,
+        output_attentions: bool = False,
+        use_cache: bool = False,
+        cache_position: Optional[torch.LongTensor] = None,
+        **kwargs,
+    ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
+        if isinstance(past_key_value, StaticCache):
+            raise ValueError(
+                "`static` cache implementation is not compatible with `attn_implementation==flash_attention_2` "
+                "make sure to use `sdpa` in the mean time, and open an issue at https://github.com/huggingface/transformers"
+            )
+        output_attentions = False
+
+        bsz, q_len, _ = hidden_states.size()
+
+        query_states = self.q_proj(hidden_states)
+        key_states = self.k_proj(hidden_states)
+        value_states = self.v_proj(hidden_states)
+
+        query_states = query_states.view(bsz, q_len, self.num_heads, self.head_dim)
+        key_states = key_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim)
+        if self.use_qk_norm:
+            query_states = self.q_norm(query_states)
+            key_states = self.k_norm(key_states)
+
+        query_states = query_states.transpose(1, 2)
+        key_states = key_states.transpose(1, 2)
+        value_states = value_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
+
+        cos, sin = self.rotary_emb(value_states, position_ids)
+        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; position_ids 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)
+
+        # TODO: These transpose are quite inefficient but Flash Attention requires the layout [batch_size, sequence_length, num_heads, head_dim]. We would need to refactor the KV cache
+        # to be able to avoid many of these transpose/reshape/view.
+        query_states = query_states.transpose(1, 2)
+        key_states = key_states.transpose(1, 2)
+        value_states = value_states.transpose(1, 2)
+
+        dropout_rate = self.attention_dropout if self.training else 0.0
+
+        # Ignore copy
+        # In PEFT, usually we cast the layer norms in float32 for training stability reasons
+        # therefore the input hidden states gets silently casted in float32. Hence, we need
+        # cast them back in the correct dtype just to be sure everything works as expected.
+        # This might slowdown training & inference so it is recommended to not cast the LayerNorms
+        # in fp32. (CohereLayerNorm handles it correctly)
+
+        input_dtype = query_states.dtype
+        if input_dtype == torch.float32:
+            if torch.is_autocast_enabled():
+                target_dtype = torch.get_autocast_gpu_dtype()
+            # Handle the case where the model is quantized
+            elif hasattr(self.config, "_pre_quantization_dtype"):
+                target_dtype = self.config._pre_quantization_dtype
+            else:
+                target_dtype = self.q_proj.weight.dtype
+
+            logger.warning_once(
+                f"The input hidden states seems to be silently casted in float32, this might be related to"
+                f" the fact you have upcasted embedding or layer norm layers in float32. We will cast back the input in"
+                f" {target_dtype}."
+            )
+
+            query_states = query_states.to(target_dtype)
+            key_states = key_states.to(target_dtype)
+            value_states = value_states.to(target_dtype)
+
+        attn_output = self._flash_attention_forward(
+            query_states, key_states, value_states, attention_mask, q_len, dropout=dropout_rate
+        )
+
+        attn_output = attn_output.reshape(bsz, q_len, self.hidden_size).contiguous()
+        attn_output = self.o_proj(attn_output)
+
+        if not output_attentions:
+            attn_weights = None
+
+        return attn_output, attn_weights, past_key_value
+
+
 COHERE_ATTENTION_CLASSES = {
     "eager": CohereAttention,
     "flash_attention_2": CohereFlashAttention2,
@@ -669,7 +537,6 @@ class CohereDecoderLayer(nn.Module):
         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,  # will become mandatory in v4.46
     ) -> Tuple[torch.FloatTensor, Optional[Tuple[torch.FloatTensor, torch.FloatTensor]]]:
         """
         Args:
@@ -684,11 +551,6 @@ class CohereDecoderLayer(nn.Module):
                 If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding
                 (see `past_key_values`).
             past_key_value (`Tuple(torch.FloatTensor)`, *optional*): cached past key and value projection states
-            cache_position (`torch.LongTensor` of shape `(sequence_length)`, *optional*):
-                Indices depicting the position of the input sequence tokens in the sequence
-            position_embeddings (`Tuple[torch.FloatTensor, torch.FloatTensor]`, *optional*):
-                Tuple containing the cosine and sine positional embeddings of shape `(batch_size, seq_len, head_dim)`,
-                with `head_dim` being the embedding dimension of each attention head.
         """
         residual = hidden_states
 
@@ -703,7 +565,6 @@ class CohereDecoderLayer(nn.Module):
             output_attentions=output_attentions,
             use_cache=use_cache,
             cache_position=cache_position,
-            position_embeddings=position_embeddings,
         )
 
         # Fully Connected
@@ -744,7 +605,6 @@ COHERE_START_DOCSTRING = r"""
     "The bare Cohere Model outputting raw hidden-states without any specific head on top.",
     COHERE_START_DOCSTRING,
 )
-# Copied from transformers.models.llama.modeling_llama.LlamaPreTrainedModel with Llama->Cohere
 class CoherePreTrainedModel(PreTrainedModel):
     config_class = CohereConfig
     base_model_prefix = "model"
@@ -837,6 +697,10 @@ COHERE_INPUTS_DOCSTRING = r"""
             more detail.
         return_dict (`bool`, *optional*):
             Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple.
+        cache_position (`torch.LongTensor` of shape `(sequence_length)`, *optional*):
+            Indices depicting the position of the input sequence tokens in the sequence. Contrarily to `position_ids`,
+            this tensor is not affected by padding. It is used to update the cache in the correct position and to infer
+            the complete sequence length.
 """
 
 
@@ -844,7 +708,6 @@ COHERE_INPUTS_DOCSTRING = r"""
     "The bare Cohere Model outputting raw hidden-states without any specific head on top.",
     COHERE_START_DOCSTRING,
 )
-# Copied from transformers.models.llama.modeling_llama.LlamaModel with Llama->Cohere, LLAMA->COHERE
 class CohereModel(CoherePreTrainedModel):
     """
     Transformer decoder consisting of *config.num_hidden_layers* layers. Each layer is a [`CohereDecoderLayer`]
@@ -853,7 +716,6 @@ class CohereModel(CoherePreTrainedModel):
         config: CohereConfig
     """
 
-    # Ignore copy
     def __init__(self, config: CohereConfig):
         super().__init__(config)
         self.padding_idx = config.pad_token_id
@@ -863,7 +725,7 @@ class CohereModel(CoherePreTrainedModel):
         self.layers = nn.ModuleList(
             [CohereDecoderLayer(config, layer_idx) for layer_idx in range(config.num_hidden_layers)]
         )
-        self.norm = CohereLayerNorm(hidden_size=(config.hidden_size), eps=config.layer_norm_eps)
+        self.norm = CohereRMSNorm(config.hidden_size, eps=config.rms_norm_eps)
         self.rotary_emb = CohereRotaryEmbedding(config=config)
         self.gradient_checkpointing = False
 
@@ -1068,18 +930,15 @@ class CohereModel(CoherePreTrainedModel):
         return causal_mask
 
 
-# Copied from transformers.models.llama.modeling_llama.LlamaForCausalLM with Llama->Cohere
 class CohereForCausalLM(CoherePreTrainedModel, GenerationMixin):
     _tied_weights_keys = ["lm_head.weight"]
 
-    # Ignore copy
     def __init__(self, config):
         super().__init__(config)
         self.model = CohereModel(config)
         self.vocab_size = config.vocab_size
         self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)
-        self.logit_scale = config.logit_scale
+
-        self.tie_word_embeddings = config.tie_word_embeddings
         # Initialize weights and apply final processing
         self.post_init()
 
@@ -1101,7 +960,6 @@ class CohereForCausalLM(CoherePreTrainedModel, GenerationMixin):
     def get_decoder(self):
         return self.model
 
-    # Ignore copy
     @add_start_docstrings_to_model_forward(COHERE_INPUTS_DOCSTRING)
     @replace_return_docstrings(output_type=CausalLMOutputWithPast, config_class=_CONFIG_FOR_DOC)
     def forward(
@@ -1117,7 +975,6 @@ class CohereForCausalLM(CoherePreTrainedModel, GenerationMixin):
         output_hidden_states: Optional[bool] = None,
         return_dict: Optional[bool] = None,
         cache_position: Optional[torch.LongTensor] = None,
-        num_logits_to_keep: int = 0,
     ) -> Union[Tuple, CausalLMOutputWithPast]:
         r"""
         Args:
@@ -1126,11 +983,6 @@ class CohereForCausalLM(CoherePreTrainedModel, GenerationMixin):
                 config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are ignored
                 (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`.
 
-            num_logits_to_keep (`int`, *optional*):
-                Calculate logits for the last `num_logits_to_keep` tokens. If `0`, calculate logits for all
-                `input_ids` (special case). Only last token logits are needed for generation, and calculating them only for that
-                token can save memory, which becomes pretty significant for long sequences or large vocabulary size.
-
         Returns:
 
         Example:
@@ -1170,19 +1022,12 @@ class CohereForCausalLM(CoherePreTrainedModel, GenerationMixin):
         )
 
         hidden_states = outputs[0]
-        if labels is None and not is_torchdynamo_compiling():
+        logits = self.lm_head(hidden_states)
-            logger.warning_once(
+        logits = logits * self.config.logit_scale
-                "Starting from v4.46, the `logits` model output will have the same type as the model (except at train time, where it will always be FP32)"
+        logits = logits.float()
-            )
-        # Only compute necessary logits, and do not upcast them to float if we are not computing the loss
-        # TODO: remove the float() operation in v4.46
-        logits = self.lm_head(hidden_states[:, -num_logits_to_keep:, :]).float()
-        logits = logits * self.logit_scale
 
         loss = None
         if labels is not None:
-            # Upcast to float if we need to compute the loss to avoid potential precision issues
-            logits = logits.float()
             # Shift so that tokens < n predict n
             shift_logits = logits[..., :-1, :].contiguous()
             shift_labels = labels[..., 1:].contiguous()
@@ -1279,3 +1124,17 @@ class CohereForCausalLM(CoherePreTrainedModel, GenerationMixin):
             }
         )
         return model_inputs
+
+    @property
+    def logit_scale(self):
+        logger.warning(
+            "`logit_scale` attribute is going to be deprecated in future versions, please use `model.config.logit_scale` instead."
+        )
+        return self.config.logit_scale
+
+    @property
+    def tie_word_embeddings(self):
+        logger.warning(
+            "`tie_word_embeddings` attribute is going to be deprecated in future versions, please use `model.config.tie_word_embeddings` instead."
+        )
+        return self.config.tie_word_embeddings

src/transformers/models/cohere/modular_cohere.py

@@ -0,0 +1,663 @@
+# Copyright 2024 Google Inc. HuggingFace Inc. team. All rights reserved.
+#
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+import math
+from typing import List, Optional, Tuple, Union
+
+import torch
+import torch.utils.checkpoint
+from torch import nn
+from torch.nn import CrossEntropyLoss
+
+from transformers.models.gemma.configuration_gemma import GemmaConfig
+from transformers.models.llama.modeling_llama import (
+    LlamaFlashAttention2,
+    LlamaForCausalLM,
+    LlamaSdpaAttention,
+    repeat_kv,
+)
+
+from ...activations import ACT2FN
+from ...cache_utils import Cache, StaticCache
+from ...modeling_outputs import CausalLMOutputWithPast
+from ...pytorch_utils import ALL_LAYERNORM_LAYERS
+from ...utils import logging
+
+
+logger = logging.get_logger(__name__)
+
+
+class CohereConfig(GemmaConfig):
+    r"""
+    use_qk_norm (`bool`, *optional*, defaults to `False`):
+        Whether to use query-key normalization in the attention
+    """
+
+    def __init__(
+        self,
+        hidden_size=8192,
+        intermediate_size=22528,
+        num_hidden_layers=40,
+        hidden_act="silu",
+        use_qk_norm=False,
+        layer_norm_eps=1e-05,
+        num_attention_heads=64,
+        num_key_value_heads=None,
+        logit_scale=0.0625,
+        rms_norm_eps=1e-05,
+        bos_token_id=5,
+        eos_token_id=255001,
+        **super_kwargs,
+    ):
+        super().__init__(self, **super_kwargs)
+        self.use_qk_norm = use_qk_norm
+        self.layer_norm_eps = layer_norm_eps
+        self.rms_norm_eps = layer_norm_eps
+
+        self.hidden_size = hidden_size
+        self.itermediate_size = intermediate_size
+        self.num_hidden_layers = num_hidden_layers
+
+        self.hidden_act = hidden_act
+
+        self.bos_token_id = bos_token_id
+        self.eos_token_id = eos_token_id
+
+        self.logit_scale = logit_scale
+
+
+        if num_key_value_heads is None:
+            self.num_key_value_heads = num_attention_heads
+
+
+class CohereRMSNorm(nn.Module):
+    def __init__(self, hidden_size=None, eps=1e-5, bias=False):
+        """The hidden size can be a tuple or an int. The tuple is used for QKNorm to normalize across head_dim"""
+        super().__init__()
+        self.weight = nn.Parameter(torch.ones(hidden_size))
+        self.variance_epsilon = eps
+
+    def forward(self, hidden_states):
+        input_dtype = hidden_states.dtype
+        hidden_states = hidden_states.to(torch.float32)
+        mean = hidden_states.mean(-1, keepdim=True)
+        variance = (hidden_states - mean).pow(2).mean(-1, keepdim=True)
+        hidden_states = (hidden_states - mean) * torch.rsqrt(variance + self.variance_epsilon)
+        hidden_states = self.weight.to(torch.float32) * hidden_states
+        return hidden_states.to(input_dtype)
+
+
+ALL_LAYERNORM_LAYERS.append(CohereRMSNorm)
+
+
+class CohereLayerNorm(CohereRMSNorm):
+    def __init__(self, *args, **kwargs):
+        logger.warning_once("CohereLayerNorm is deprecated by CohereRMSNorm and will be removed in v4.45.")
+        super().__init__(*args, **kwargs)
+
+
+class CohereRotaryEmbedding(nn.Module):
+    def __init__(self, dim, max_position_embeddings=2048, base=10000, device=None, scaling_factor=1.0):
+        super().__init__()
+        self.scaling_factor = scaling_factor
+        self.dim = dim
+        self.max_position_embeddings = max_position_embeddings
+        self.base = base
+        inv_freq = 1.0 / (self.base ** (torch.arange(0, self.dim, 2, dtype=torch.int64).float().to(device) / self.dim))
+        self.register_buffer("inv_freq", inv_freq, persistent=False)
+
+    @torch.no_grad()
+    def forward(self, x, position_ids):
+        # x: [bs, num_attention_heads, seq_len, head_size]
+        inv_freq_expanded = self.inv_freq[None, :, None].float().expand(position_ids.shape[0], -1, 1)
+        position_ids_expanded = position_ids[:, None, :].float()
+
+        # Force float32 since bfloat16 loses precision on long contexts
+        # See https://github.com/huggingface/transformers/pull/29285
+        device_type = x.device.type
+        device_type = device_type if isinstance(device_type, str) and device_type != "mps" else "cpu"
+        with torch.autocast(device_type=device_type, enabled=False):
+            freqs = (inv_freq_expanded.float() @ position_ids_expanded.float()).transpose(1, 2)
+            emb = torch.repeat_interleave(freqs, 2, dim=-1)
+            cos = emb.cos()
+            sin = emb.sin()
+        return cos, sin
+
+
+def rotate_half(x):
+    # Split and rotate
+    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.
+    """
+    dtype = q.dtype
+    q = q.float()
+    k = k.float()
+    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.to(dtype=dtype), k_embed.to(dtype=dtype)
+
+
+class CohereMLP(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        self.config = config
+        self.hidden_size = config.hidden_size
+        self.intermediate_size = config.intermediate_size
+        self.gate_proj = nn.Linear(self.hidden_size, self.intermediate_size, bias=False)
+        self.up_proj = nn.Linear(self.hidden_size, self.intermediate_size, bias=False)
+        self.down_proj = nn.Linear(self.intermediate_size, self.hidden_size, bias=False)
+        self.act_fn = ACT2FN[config.hidden_act]
+
+    def forward(self, x):
+        down_proj = self.down_proj(self.act_fn(self.gate_proj(x)) * self.up_proj(x))
+        return down_proj
+
+
+class CohereAttention(nn.Module):
+    """Multi-headed attention from 'Attention Is All You Need' paper"""
+
+    def __init__(self, config: CohereConfig, layer_idx: Optional[int] = None):
+        super().__init__()
+        self.config = config
+        self.layer_idx = layer_idx
+        if layer_idx is None:
+            logger.warning_once(
+                f"Instantiating {self.__class__.__name__} without passing a `layer_idx` is not recommended and will "
+                "lead to errors during the forward call if caching is used. Please make sure to provide a `layer_idx` "
+                "when creating this class."
+            )
+
+        self.attention_dropout = config.attention_dropout
+        self.hidden_size = config.hidden_size
+        self.num_heads = config.num_attention_heads
+        self.head_dim = self.hidden_size // self.num_heads
+        self.num_key_value_heads = config.num_key_value_heads
+        self.num_key_value_groups = self.num_heads // self.num_key_value_heads
+        self.max_position_embeddings = config.max_position_embeddings
+        self.rope_theta = config.rope_theta
+        self.is_causal = True
+        self.use_qk_norm = config.use_qk_norm
+
+        if (self.head_dim * self.num_heads) != self.hidden_size:
+            raise ValueError(
+                f"hidden_size must be divisible by num_heads (got `hidden_size`: {self.hidden_size}"
+                f" and `num_heads`: {self.num_heads})."
+            )
+
+        if self.use_qk_norm:
+            # When sharding the model using Tensor Parallelism, need to be careful to use n_local_heads
+            self.q_norm = CohereLayerNorm(hidden_size=(self.num_heads, self.head_dim), eps=config.layer_norm_eps)
+            self.k_norm = CohereLayerNorm(
+                hidden_size=(self.num_key_value_heads, self.head_dim), eps=config.layer_norm_eps
+            )
+
+        self.q_proj = nn.Linear(self.hidden_size, self.num_heads * self.head_dim, bias=config.attention_bias)
+        self.k_proj = nn.Linear(self.hidden_size, self.num_key_value_heads * self.head_dim, bias=config.attention_bias)
+        self.v_proj = nn.Linear(self.hidden_size, self.num_key_value_heads * self.head_dim, bias=config.attention_bias)
+        self.o_proj = nn.Linear(self.hidden_size, self.hidden_size, bias=config.attention_bias)
+        self._init_rope()
+
+    def _init_rope(self):
+        self.rotary_emb = CohereRotaryEmbedding(
+            self.head_dim,
+            max_position_embeddings=self.max_position_embeddings,
+            base=self.rope_theta,
+        )
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_value: Optional[Cache] = None,
+        output_attentions: bool = False,
+        use_cache: bool = False,
+        cache_position: Optional[torch.LongTensor] = None,
+        **kwargs,
+    ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
+        bsz, q_len, _ = hidden_states.size()
+
+        query_states = self.q_proj(hidden_states)
+        key_states = self.k_proj(hidden_states)
+        value_states = self.v_proj(hidden_states)
+
+        query_states = query_states.view(bsz, q_len, self.num_heads, self.head_dim)
+        key_states = key_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim)
+        if self.use_qk_norm:
+            query_states = self.q_norm(query_states)
+            key_states = self.k_norm(key_states)
+
+        query_states = query_states.transpose(1, 2)
+        key_states = key_states.transpose(1, 2)
+        value_states = value_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
+
+        cos, sin = self.rotary_emb(value_states, position_ids)
+        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; position_ids 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)
+
+        key_states = repeat_kv(key_states, self.num_key_value_groups)
+        value_states = repeat_kv(value_states, self.num_key_value_groups)
+
+        attn_weights = torch.matmul(query_states, key_states.transpose(2, 3)) / math.sqrt(self.head_dim)
+
+        if attention_mask is not None:  # no matter the length, we just slice it
+            causal_mask = attention_mask[:, :, :, : key_states.shape[-2]]
+            attn_weights = attn_weights + causal_mask
+
+        # upcast attention to fp32
+        attn_weights = nn.functional.softmax(attn_weights, dim=-1, dtype=torch.float32).to(query_states.dtype)
+        attn_weights = nn.functional.dropout(attn_weights, p=self.attention_dropout, training=self.training)
+        attn_output = torch.matmul(attn_weights, value_states)
+
+        if attn_output.size() != (bsz, self.num_heads, q_len, self.head_dim):
+            raise ValueError(
+                f"`attn_output` should be of size {(bsz, self.num_heads, q_len, self.head_dim)}, but is"
+                f" {attn_output.size()}"
+            )
+
+        attn_output = attn_output.transpose(1, 2).contiguous()
+
+        attn_output = attn_output.reshape(bsz, q_len, self.hidden_size)
+
+        attn_output = self.o_proj(attn_output)
+
+        if not output_attentions:
+            attn_weights = None
+
+        return attn_output, attn_weights, past_key_value
+
+
+class CohereSdpaAttention(CohereAttention):
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_value: Optional[Cache] = None,
+        output_attentions: bool = False,
+        use_cache: bool = False,
+        cache_position: Optional[torch.LongTensor] = None,
+    ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
+        if output_attentions:
+            # TODO: Improve this warning with e.g. `model.config.attn_implementation = "manual"` once this is implemented.
+            logger.warning_once(
+                "CohereModel is using CohereSdpaAttention, but `torch.nn.functional.scaled_dot_product_attention` does not support `output_attentions=True`. Falling back to the manual attention implementation, "
+                'but specifying the manual implementation will be required from Transformers version v5.0.0 onwards. This warning can be removed using the argument `attn_implementation="eager"` when loading the model.'
+            )
+            return super().forward(
+                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,
+            )
+
+        bsz, q_len, _ = hidden_states.size()
+
+        query_states = self.q_proj(hidden_states)
+        key_states = self.k_proj(hidden_states)
+        value_states = self.v_proj(hidden_states)
+
+        query_states = query_states.view(bsz, q_len, self.num_heads, self.head_dim)
+        key_states = key_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim)
+        if self.use_qk_norm:
+            query_states = self.q_norm(query_states)
+            key_states = self.k_norm(key_states)
+
+        query_states = query_states.transpose(1, 2)
+        key_states = key_states.transpose(1, 2)
+        value_states = value_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
+
+        cos, sin = self.rotary_emb(value_states, position_ids)
+        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)
+
+        key_states = repeat_kv(key_states, self.num_key_value_groups)
+        value_states = repeat_kv(value_states, self.num_key_value_groups)
+
+        causal_mask = attention_mask
+        # if attention_mask is not None and cache_position is not None:
+        if attention_mask is not None:
+            causal_mask = causal_mask[:, :, :, : key_states.shape[-2]]
+
+        # SDPA with memory-efficient backend is currently (torch==2.1.2) bugged with non-contiguous inputs with custom attn_mask,
+        # Reference: https://github.com/pytorch/pytorch/issues/112577.
+        if query_states.device.type == "cuda" and causal_mask is not None:
+            query_states = query_states.contiguous()
+            key_states = key_states.contiguous()
+            value_states = value_states.contiguous()
+
+        # We dispatch to SDPA's Flash Attention or Efficient kernels via this `is_causal` if statement instead of an inline conditional assignment
+        # in SDPA to support both torch.compile's dynamic shapes and full graph options. An inline conditional prevents dynamic shapes from compiling.
+        is_causal = True if causal_mask is None and q_len > 1 else False
+
+        attn_output = torch.nn.functional.scaled_dot_product_attention(
+            query_states,
+            key_states,
+            value_states,
+            attn_mask=causal_mask,
+            dropout_p=self.attention_dropout if self.training else 0.0,
+            is_causal=is_causal,
+        )
+
+        attn_output = attn_output.transpose(1, 2).contiguous()
+        attn_output = attn_output.view(bsz, q_len, self.hidden_size)
+
+        attn_output = self.o_proj(attn_output)
+
+        return attn_output, None, past_key_value
+
+
+class CohereFlashAttention2(CohereAttention):
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: Optional[torch.LongTensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_value: Optional[Cache] = None,
+        output_attentions: bool = False,
+        use_cache: bool = False,
+        cache_position: Optional[torch.LongTensor] = None,
+        **kwargs,
+    ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
+        if isinstance(past_key_value, StaticCache):
+            raise ValueError(
+                "`static` cache implementation is not compatible with `attn_implementation==flash_attention_2` "
+                "make sure to use `sdpa` in the mean time, and open an issue at https://github.com/huggingface/transformers"
+            )
+        output_attentions = False
+
+        bsz, q_len, _ = hidden_states.size()
+
+        query_states = self.q_proj(hidden_states)
+        key_states = self.k_proj(hidden_states)
+        value_states = self.v_proj(hidden_states)
+
+        query_states = query_states.view(bsz, q_len, self.num_heads, self.head_dim)
+        key_states = key_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim)
+        if self.use_qk_norm:
+            query_states = self.q_norm(query_states)
+            key_states = self.k_norm(key_states)
+
+        query_states = query_states.transpose(1, 2)
+        key_states = key_states.transpose(1, 2)
+        value_states = value_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
+
+        cos, sin = self.rotary_emb(value_states, position_ids)
+        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; position_ids 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)
+
+        # TODO: These transpose are quite inefficient but Flash Attention requires the layout [batch_size, sequence_length, num_heads, head_dim]. We would need to refactor the KV cache
+        # to be able to avoid many of these transpose/reshape/view.
+        query_states = query_states.transpose(1, 2)
+        key_states = key_states.transpose(1, 2)
+        value_states = value_states.transpose(1, 2)
+
+        dropout_rate = self.attention_dropout if self.training else 0.0
+
+        # Ignore copy
+        # In PEFT, usually we cast the layer norms in float32 for training stability reasons
+        # therefore the input hidden states gets silently casted in float32. Hence, we need
+        # cast them back in the correct dtype just to be sure everything works as expected.
+        # This might slowdown training & inference so it is recommended to not cast the LayerNorms
+        # in fp32. (CohereLayerNorm handles it correctly)
+
+        input_dtype = query_states.dtype
+        if input_dtype == torch.float32:
+            if torch.is_autocast_enabled():
+                target_dtype = torch.get_autocast_gpu_dtype()
+            # Handle the case where the model is quantized
+            elif hasattr(self.config, "_pre_quantization_dtype"):
+                target_dtype = self.config._pre_quantization_dtype
+            else:
+                target_dtype = self.q_proj.weight.dtype
+
+            logger.warning_once(
+                f"The input hidden states seems to be silently casted in float32, this might be related to"
+                f" the fact you have upcasted embedding or layer norm layers in float32. We will cast back the input in"
+                f" {target_dtype}."
+            )
+
+            query_states = query_states.to(target_dtype)
+            key_states = key_states.to(target_dtype)
+            value_states = value_states.to(target_dtype)
+
+        attn_output = self._flash_attention_forward(
+            query_states, key_states, value_states, attention_mask, q_len, dropout=dropout_rate
+        )
+
+        attn_output = attn_output.reshape(bsz, q_len, self.hidden_size).contiguous()
+        attn_output = self.o_proj(attn_output)
+
+        if not output_attentions:
+            attn_weights = None
+
+        return attn_output, attn_weights, past_key_value
+
+
+COHERE_ATTENTION_CLASSES = {
+    "eager": CohereAttention,
+    "flash_attention_2": CohereFlashAttention2,
+    "sdpa": CohereSdpaAttention,
+}
+
+
+class CohereDecoderLayer(nn.Module):
+    def __init__(self, config: CohereConfig, layer_idx: int):
+        super().__init__()
+        self.hidden_size = config.hidden_size
+
+        self.self_attn = COHERE_ATTENTION_CLASSES[config._attn_implementation](config=config, layer_idx=layer_idx)
+
+        self.mlp = CohereMLP(config)
+        self.input_layernorm = CohereLayerNorm(hidden_size=(config.hidden_size), eps=config.layer_norm_eps)
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_value: Optional[Tuple[torch.Tensor]] = None,
+        output_attentions: Optional[bool] = False,
+        use_cache: Optional[bool] = False,
+        cache_position: Optional[torch.LongTensor] = None,
+    ) -> Tuple[torch.FloatTensor, Optional[Tuple[torch.FloatTensor, torch.FloatTensor]]]:
+        """
+        Args:
+            hidden_states (`torch.FloatTensor`): input to the layer of shape `(batch, seq_len, embed_dim)`
+            attention_mask (`torch.FloatTensor`, *optional*):
+                attention mask of size `(batch_size, sequence_length)` if flash attention is used or `(batch_size, 1,
+                query_sequence_length, key_sequence_length)` if default attention is used.
+            output_attentions (`bool`, *optional*):
+                Whether or not to return the attentions tensors of all attention layers. See `attentions` under
+                returned tensors for more detail.
+            use_cache (`bool`, *optional*):
+                If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding
+                (see `past_key_values`).
+            past_key_value (`Tuple(torch.FloatTensor)`, *optional*): cached past key and value projection states
+        """
+        residual = hidden_states
+
+        hidden_states = self.input_layernorm(hidden_states)
+
+        # Self Attention
+        hidden_states_attention, self_attn_weights, present_key_value = 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,
+        )
+
+        # Fully Connected
+        hidden_states_mlp = self.mlp(hidden_states)
+
+        # Add everything together
+        hidden_states = residual + hidden_states_attention + hidden_states_mlp
+
+        outputs = (hidden_states,)
+
+        if output_attentions:
+            outputs += (self_attn_weights,)
+
+        if use_cache:
+            outputs += (present_key_value,)
+
+        return outputs
+
+
+class CohereForCausalLM(LlamaForCausalLM):
+    def forward(
+        self,
+        input_ids: torch.LongTensor = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[List[torch.FloatTensor]] = None,
+        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,
+        return_dict: Optional[bool] = None,
+        cache_position: Optional[torch.LongTensor] = None,
+    ) -> Union[Tuple, CausalLMOutputWithPast]:
+        r"""
+        Args:
+            labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
+                Labels for computing the masked language modeling loss. Indices should either be in `[0, ...,
+                config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are ignored
+                (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`.
+
+        Returns:
+
+        Example:
+
+        ```python
+        >> from transformers import AutoTokenizer, CohereForCausalLM
+
+        >> model = CohereForCausalLM.from_pretrained("CohereForAI/c4ai-command-r-v01")
+        >> tokenizer = AutoTokenizer.from_pretrained("CohereForAI/c4ai-command-r-v01")
+
+        >> prompt = "Hey, are you conscious? Can you talk to me?"
+        >> inputs = tokenizer(prompt, return_tensors="pt")
+
+        >> # Generate
+        >> generate_ids = model.generate(inputs.input_ids, max_length=30)
+        >> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]
+        "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you."
+        ```"""
+        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
+        )
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn)
+        outputs = self.model(
+            input_ids=input_ids,
+            attention_mask=attention_mask,
+            position_ids=position_ids,
+            past_key_values=past_key_values,
+            inputs_embeds=inputs_embeds,
+            use_cache=use_cache,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+            cache_position=cache_position,
+        )
+
+        hidden_states = outputs[0]
+        logits = self.lm_head(hidden_states)
+        logits = logits * self.config.logit_scale
+        logits = logits.float()
+
+        loss = None
+        if labels is not None:
+            # Shift so that tokens < n predict n
+            shift_logits = logits[..., :-1, :].contiguous()
+            shift_labels = labels[..., 1:].contiguous()
+            # Flatten the tokens
+            loss_fct = CrossEntropyLoss()
+            shift_logits = shift_logits.view(-1, self.config.vocab_size)
+            shift_labels = shift_labels.view(-1)
+            # Enable model parallelism
+            shift_labels = shift_labels.to(shift_logits.device)
+            loss = loss_fct(shift_logits, shift_labels)
+
+        if not return_dict:
+            output = (logits,) + outputs[1:]
+            return (loss,) + output if loss is not None else output
+
+        return CausalLMOutputWithPast(
+            loss=loss,
+            logits=logits,
+            past_key_values=outputs.past_key_values,
+            hidden_states=outputs.hidden_states,
+            attentions=outputs.attentions,
+        )
+
+    @property
+    def logit_scale(self):
+        logger.warning(
+            "`logit_scale` attribute is going to be deprecated in future versions, please use `model.config.logit_scale` instead."
+        )
+        return self.config.logit_scale
+
+    @property
+    def tie_word_embeddings(self):
+        logger.warning(
+            "`tie_word_embeddings` attribute is going to be deprecated in future versions, please use `model.config.tie_word_embeddings` instead."
+        )
+        return self.config.tie_word_embeddings