# 🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨 # This file was automatically generated from src/transformers/models/cohere2/modular_cohere2.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_cohere2.py file directly. One of our CI enforces this. # 🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨 # coding=utf-8 # Copyright 2024 Cohere 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. from typing import Callable, List, Optional, Tuple, Union import torch import torch.nn as nn from ...activations import ACT2FN from ...cache_utils import Cache, HybridCache, StaticCache from ...generation import GenerationMixin from ...modeling_flash_attention_utils import FlashAttentionKwargs from ...modeling_layers import GradientCheckpointingLayer from ...modeling_outputs import BaseModelOutputWithPast, CausalLMOutputWithPast from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack from ...utils import LossKwargs, auto_docstring, can_return_tuple, is_torch_flex_attn_available, logging from ...utils.deprecation import deprecate_kwarg from .configuration_cohere2 import Cohere2Config if is_torch_flex_attn_available(): from torch.nn.attention.flex_attention import BlockMask from ...integrations.flex_attention import make_flex_block_causal_mask logger = logging.get_logger(__name__) class Cohere2RotaryEmbedding(nn.Module): def __init__(self, config: Cohere2Config, device=None): super().__init__() # 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")) 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.register_buffer("inv_freq", inv_freq, persistent=False) self.original_inv_freq = self.inv_freq @torch.no_grad() @dynamic_rope_update # power user: used with advanced RoPE types (e.g. dynamic rope) def forward(self, x, position_ids): inv_freq_expanded = self.inv_freq[None, :, None].float().expand(position_ids.shape[0], -1, 1) position_ids_expanded = position_ids[:, None, :].float() device_type = x.device.type if isinstance(x.device.type, str) and x.device.type != "mps" else "cpu" with torch.autocast(device_type=device_type, enabled=False): # Force float32 freqs = (inv_freq_expanded.float() @ position_ids_expanded.float()).transpose(1, 2) emb = torch.repeat_interleave(freqs, 2, dim=-1) # diff from Llama: we interleave() instead of cat() cos = emb.cos() * self.attention_scaling sin = emb.sin() * self.attention_scaling return cos.to(dtype=x.dtype), sin.to(dtype=x.dtype) class Cohere2LayerNorm(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) 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 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. """ 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 Cohere2Attention(nn.Module): """Multi-headed attention from 'Attention Is All You Need' paper""" def __init__(self, config: Cohere2Config, layer_idx: Optional[int] = None): 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 ) self.sliding_window = ( config.sliding_window if (self.layer_idx + 1) % self.config.sliding_window_pattern != 0 else None ) 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 if self.sliding_window is not None: query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin) if past_key_value is not None: cache_kwargs = { "sin": sin, "cos": cos, "sliding_window": self.sliding_window, "cache_position": cache_position, } key_states, value_states = past_key_value.update(key_states, value_states, self.layer_idx, cache_kwargs) # Here we need to slice as we use a static cache by default, but FA2 does not support it if attention_mask is not None and self.config._attn_implementation == "flash_attention_2": seq_len = attention_mask.shape[-1] key_states, value_states = key_states[:, :, :seq_len, :], value_states[:, :, :seq_len, :] 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, sliding_window=self.sliding_window, **kwargs, ) attn_output = attn_output.reshape(*input_shape, -1).contiguous() attn_output = self.o_proj(attn_output) return attn_output, attn_weights class Cohere2MLP(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 Cohere2DecoderLayer(GradientCheckpointingLayer): def __init__(self, config: Cohere2Config, layer_idx: int): super().__init__() self.hidden_size = config.hidden_size self.self_attn = Cohere2Attention(config, layer_idx) self.mlp = Cohere2MLP(config) self.input_layernorm = Cohere2LayerNorm(hidden_size=(config.hidden_size), eps=config.layer_norm_eps) self.config = config self.is_sliding = (layer_idx + 1) % self.config.sliding_window_pattern != 0 self.sliding_window = config.sliding_window @deprecate_kwarg("last_cache_position", version="4.53.0") def forward( self, hidden_states: torch.Tensor, position_embeddings: Tuple[torch.Tensor, torch.Tensor], attention_mask: Optional[torch.Tensor] = None, past_key_value: Optional[Cache] = None, output_attentions: Optional[bool] = False, use_cache: Optional[bool] = False, cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> 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)` position_embeddings (`Tuple[torch.FloatTensor, torch.FloatTensor]`): 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. 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. past_key_value (`Tuple(torch.FloatTensor)`, *optional*): cached past key and value projection states 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`). cache_position (`torch.LongTensor` of shape `(sequence_length)`, *optional*): Indices depicting the position of the input sequence tokens in the sequence """ if self.is_sliding and attention_mask is not None: # efficient SDPA and no padding # In prefill, we may be larger than sliding window effective_seq_len = max(cache_position.shape[0], self.sliding_window) # For FA2, the mask is 2D and is of shape [bs, processed_tokens] (not [bs, max_cache_len]), # thus we must slice from the right (at most `effective_seq_len` elements) if self.config._attn_implementation == "flash_attention_2": attention_mask = attention_mask[:, -effective_seq_len:] # Otherwise, the mask is 4D of shape [bs, 1, query_len, max_cache_len] thus we must slice # from the left, with an offset if we are beyond the sliding window else: min_dtype = torch.finfo(hidden_states.dtype).min sliding_window_mask = torch.tril( torch.ones_like(attention_mask, dtype=torch.bool), diagonal=-self.sliding_window ) attention_mask = torch.where(sliding_window_mask, min_dtype, attention_mask) # In case we are beyond the sliding window, we need to correctly offset the mask slicing offset = cache_position[-1] - effective_seq_len + 1 # Should only be used when beyond the sliding window (i.e. offset > 0) offset = torch.clamp(offset, min=0) # equivalent to: `attention_mask = attention_mask[:, :, :, offset : offset + effective_seq_len]`, # but without data-dependent slicing (i.e. torch.compile friendly) mask_indexes = torch.arange( min(effective_seq_len, attention_mask.shape[-1]), device=attention_mask.device ) mask_indexes += offset attention_mask = attention_mask[:, :, :, mask_indexes] residual = hidden_states hidden_states = self.input_layernorm(hidden_states) # Self Attention hidden_states_attention, self_attn_weights = self.self_attn( hidden_states=hidden_states, position_embeddings=position_embeddings, attention_mask=attention_mask, past_key_value=past_key_value, output_attentions=output_attentions, use_cache=use_cache, cache_position=cache_position, **kwargs, ) # 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,) return outputs @auto_docstring class Cohere2PreTrainedModel(PreTrainedModel): config_class = Cohere2Config base_model_prefix = "model" supports_gradient_checkpointing = True _no_split_modules = ["Cohere2DecoderLayer"] _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 _supports_attention_backend = True def _init_weights(self, module): std = self.config.initializer_range if isinstance(module, nn.Linear): module.weight.data.normal_(mean=0.0, std=std) if module.bias is not None: module.bias.data.zero_() elif isinstance(module, nn.Embedding): module.weight.data.normal_(mean=0.0, std=std) if module.padding_idx is not None: module.weight.data[module.padding_idx].zero_() elif isinstance(module, Cohere2LayerNorm): module.weight.data.fill_(1.0) @auto_docstring class Cohere2Model(Cohere2PreTrainedModel): def __init__(self, config: Cohere2Config): super().__init__(config) self.padding_idx = config.pad_token_id self.vocab_size = config.vocab_size self.embed_tokens = nn.Embedding(config.vocab_size, config.hidden_size, self.padding_idx) self.layers = nn.ModuleList( [Cohere2DecoderLayer(config, layer_idx) for layer_idx in range(config.num_hidden_layers)] ) self.norm = Cohere2LayerNorm(hidden_size=(config.hidden_size), eps=config.layer_norm_eps) self.rotary_emb = Cohere2RotaryEmbedding(config=config) self.gradient_checkpointing = False # Initialize weights and apply final processing self.post_init() def get_input_embeddings(self): return self.embed_tokens def set_input_embeddings(self, value): self.embed_tokens = value @can_return_tuple @auto_docstring def forward( self, input_ids: Optional[torch.LongTensor] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, past_key_values: Optional[HybridCache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, **flash_attn_kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): raise ValueError("You must specify exactly one of input_ids or inputs_embeds") if self.gradient_checkpointing and self.training and use_cache: logger.warning_once( "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`." ) use_cache = False if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) if use_cache and past_key_values is None and not self.training: batch_size, seq_len, _ = inputs_embeds.shape # NOTE: ideally, `HybridCache` should be initialized outside the model with `layer_device_map` past_key_values = HybridCache( self.config, max_batch_size=batch_size, max_cache_len=seq_len, dtype=inputs_embeds.dtype, device=self.device, ) if cache_position is None: past_seen_tokens = past_key_values.get_seq_length() if past_key_values is not None else 0 cache_position = torch.arange( past_seen_tokens, past_seen_tokens + inputs_embeds.shape[1], device=inputs_embeds.device ) if position_ids is None: position_ids = cache_position.unsqueeze(0) causal_mask = self._update_causal_mask( attention_mask, inputs_embeds, cache_position, past_key_values, output_attentions ) hidden_states = inputs_embeds # create position embeddings to be shared across the decoder layers position_embeddings = self.rotary_emb(hidden_states, position_ids) # decoder layers all_hidden_states = () if output_hidden_states else None all_self_attns = () if output_attentions else None for decoder_layer in self.layers: if output_hidden_states: all_hidden_states += (hidden_states,) layer_outputs = decoder_layer( hidden_states, position_embeddings=position_embeddings, attention_mask=causal_mask, past_key_value=past_key_values, output_attentions=output_attentions, use_cache=use_cache, cache_position=cache_position, **flash_attn_kwargs, ) hidden_states = layer_outputs[0] if output_attentions: all_self_attns += (layer_outputs[1],) hidden_states = self.norm(hidden_states) # add hidden states from the last decoder layer if output_hidden_states: all_hidden_states += (hidden_states,) return BaseModelOutputWithPast( last_hidden_state=hidden_states, past_key_values=past_key_values, hidden_states=all_hidden_states, attentions=all_self_attns, ) @torch.no_grad() def _update_causal_mask( self, attention_mask: Union[torch.Tensor, "BlockMask"], input_tensor: torch.Tensor, cache_position: torch.Tensor, past_key_values: HybridCache, output_attentions: bool = False, ): # Flash Attention currently doesn't support static cache but Cohere2 work only with static cache. # So we will pass in attention mask as is in any case, not only when ther's padding. Then we'll use its shape # to cut out keys/values trailing 0 used in static cache. This workaround should be compile compatible # as it doesn't cause dynamic control issues. if self.config._attn_implementation == "flash_attention_2": return attention_mask if self.config._attn_implementation == "flex_attention": if isinstance(attention_mask, torch.Tensor): attention_mask = make_flex_block_causal_mask(attention_mask) return attention_mask dtype, device = input_tensor.dtype, input_tensor.device sequence_length = input_tensor.shape[1] if isinstance(past_key_values, (HybridCache, StaticCache)): target_length = past_key_values.get_max_cache_shape() else: target_length = attention_mask.shape[-1] if attention_mask is not None else input_tensor.shape[1] # In case the provided `attention` mask is 2D, we generate a causal mask here (4D). causal_mask = self._prepare_4d_causal_attention_mask_with_cache_position( attention_mask, sequence_length=sequence_length, target_length=target_length, dtype=dtype, device=device, cache_position=cache_position, batch_size=input_tensor.shape[0], ) return causal_mask @staticmethod def _prepare_4d_causal_attention_mask_with_cache_position( attention_mask: torch.Tensor, sequence_length: int, target_length: int, dtype: torch.dtype, cache_position: torch.Tensor, batch_size: int, **kwargs, ): """ Creates a causal 4D mask of shape `(batch_size, 1, query_length, key_value_length)` from a 2D mask of shape `(batch_size, key_value_length)`, or if the input `attention_mask` is already 4D, do nothing. Args: attention_mask (`torch.Tensor`): A 2D attention mask of shape `(batch_size, key_value_length)` or a 4D attention mask of shape `(batch_size, 1, query_length, key_value_length)`. sequence_length (`int`): The sequence length being processed. target_length (`int`): The target length: when generating with static cache, the mask should be as long as the static cache, to account for the 0 padding, the part of the cache that is not filled yet. dtype (`torch.dtype`): The dtype to use for the 4D attention mask. cache_position (`torch.Tensor`): Indices depicting the position of the input sequence tokens in the sequence. batch_size (`torch.Tensor`): Batch size. """ if attention_mask is not None and attention_mask.dim() == 4: # In this case we assume that the mask comes already in inverted form and requires no inversion or slicing. causal_mask = attention_mask else: min_dtype = torch.finfo(dtype).min causal_mask = torch.full( (sequence_length, target_length), fill_value=min_dtype, dtype=dtype, device=cache_position.device ) if sequence_length != 1: causal_mask = torch.triu(causal_mask, diagonal=1) causal_mask *= torch.arange(target_length, device=cache_position.device) > cache_position.reshape(-1, 1) causal_mask = causal_mask[None, None, :, :].expand(batch_size, 1, -1, -1) if attention_mask is not None: causal_mask = causal_mask.clone() # copy to contiguous memory for in-place edit mask_length = attention_mask.shape[-1] padding_mask = causal_mask[:, :, :, :mask_length] + attention_mask[:, None, None, :].to( causal_mask.device ) padding_mask = padding_mask == 0 causal_mask[:, :, :, :mask_length] = causal_mask[:, :, :, :mask_length].masked_fill( padding_mask, min_dtype ) return causal_mask class KwargsForCausalLM(FlashAttentionKwargs, LossKwargs): ... @auto_docstring class Cohere2ForCausalLM(Cohere2PreTrainedModel, GenerationMixin): _tied_weights_keys = ["lm_head.weight"] _tp_plan = {"lm_head": "colwise_rep"} _pp_plan = {"lm_head": (["hidden_states"], ["logits"])} def __init__(self, config: Cohere2Config): super().__init__(config) self.model = Cohere2Model(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() def get_input_embeddings(self): return self.model.embed_tokens def set_input_embeddings(self, value): self.model.embed_tokens = value def get_output_embeddings(self): return self.lm_head def set_output_embeddings(self, new_embeddings): self.lm_head = new_embeddings def set_decoder(self, decoder): self.model = decoder def get_decoder(self): return self.model @can_return_tuple @auto_docstring def forward( self, input_ids: Optional[torch.LongTensor] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, past_key_values: Optional[Union[Cache, 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, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs: Unpack[KwargsForCausalLM], ) -> CausalLMOutputWithPast: r""" 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]`. Example: ```python >> from transformers import AutoTokenizer, Cohere2ForCausalLM >> model = Cohere2ForCausalLM.from_pretrained("Cohere2ForAI/c4ai-command-r-v01") >> tokenizer = AutoTokenizer.from_pretrained("Cohere2ForAI/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 ) # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: BaseModelOutputWithPast = 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, cache_position=cache_position, **kwargs, ) hidden_states = outputs.last_hidden_state # Only compute necessary logits, and do not upcast them to float if we are not computing the loss slice_indices = slice(-logits_to_keep, None) if isinstance(logits_to_keep, int) else logits_to_keep logits = self.lm_head(hidden_states[:, slice_indices, :]) logits = logits * self.logit_scale # main diff from Llama loss = None if labels is not None: loss = self.loss_function(logits=logits, labels=labels, vocab_size=self.config.vocab_size, **kwargs) return CausalLMOutputWithPast( loss=loss, logits=logits, past_key_values=outputs.past_key_values, hidden_states=outputs.hidden_states, attentions=outputs.attentions, ) def prepare_inputs_for_generation( self, input_ids, past_key_values=None, attention_mask=None, inputs_embeds=None, cache_position=None, position_ids=None, use_cache=True, logits_to_keep=None, **kwargs, ): # Overwritten: has a special cache type, `HybridCache` # If we have cache: let's slice `input_ids` through `cache_position`, to keep only the unprocessed tokens # Exception 1: when passing input_embeds, input_ids may be missing entries # Exception 2: some generation methods do special slicing of input_ids, so we don't need to do it here # Exception 3: with synced GPUs cache_position may go out of bounds, but we only want dummy token in that case. # (we can't check exception 3 while compiling) if past_key_values is not None: if ( inputs_embeds is not None # Exception 1 or cache_position[-1] >= input_ids.shape[1] # Exception 3 ): input_ids = input_ids[:, -cache_position.shape[0] :] elif input_ids.shape[1] != cache_position.shape[0]: # Default case (the "else", a no op, is Exception 2) input_ids = input_ids[:, cache_position] if attention_mask is not None and position_ids is None: # create position_ids on the fly for batch generation position_ids = attention_mask.long().cumsum(-1) - 1 position_ids.masked_fill_(attention_mask == 0, 1) if past_key_values: position_ids = position_ids[:, -input_ids.shape[1] :] # This `clone` call is needed to avoid recapturing cuda graphs with `torch.compile`'s # `mode="reduce-overhead`, as otherwise the input `position_ids` would have various stride # during the decoding. Here, simply using `.contiguous()` is not sufficient as in the # batch size = 1 case, `position_ids` is already contiguous but with varying stride # which retriggers a capture. position_ids = position_ids.clone(memory_format=torch.contiguous_format) # if `inputs_embeds` are passed, we only want to use them in the 1st generation step if inputs_embeds is not None and cache_position[0] == 0: model_inputs = {"inputs_embeds": inputs_embeds, "input_ids": None} else: # The clone here is for the same reason as for `position_ids`. model_inputs = {"input_ids": input_ids.clone(memory_format=torch.contiguous_format), "inputs_embeds": None} if ( isinstance(past_key_values, HybridCache) and attention_mask.ndim == 2 and not self.config._attn_implementation == "flash_attention_2" ): if model_inputs["inputs_embeds"] is not None: batch_size, sequence_length, _ = model_inputs["inputs_embeds"].shape device = model_inputs["inputs_embeds"].device else: batch_size, sequence_length = model_inputs["input_ids"].shape device = model_inputs["input_ids"].device attention_mask = self.model._prepare_4d_causal_attention_mask_with_cache_position( attention_mask, sequence_length=sequence_length, target_length=past_key_values.get_max_cache_shape(), dtype=self.lm_head.weight.dtype, device=device, cache_position=cache_position, batch_size=batch_size, ) if logits_to_keep is not None: model_inputs["logits_to_keep"] = logits_to_keep model_inputs.update( { "position_ids": position_ids, "cache_position": cache_position, "past_key_values": past_key_values, "use_cache": use_cache, "attention_mask": attention_mask, } ) return model_inputs __all__ = ["Cohere2ForCausalLM", "Cohere2Model", "Cohere2PreTrainedModel"]