# coding=utf-8 # Copyright 2024 The Qwen team, Alibaba Group and the 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. # 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. """PyTorch Qwen2MoE model.""" import inspect import math from typing import List, Optional, Tuple, Union import torch import torch.nn.functional as F import torch.utils.checkpoint from torch import nn from torch.nn import BCEWithLogitsLoss, CrossEntropyLoss, MSELoss from transformers.activations import ACT2FN from transformers.cache_utils import Cache, DynamicCache, StaticCache from transformers.modeling_attn_mask_utils import ( AttentionMaskConverter, ) from transformers.modeling_outputs import ( MoeCausalLMOutputWithPast, MoeModelOutputWithPast, ) from transformers.modeling_utils import PreTrainedModel from transformers.utils import ( is_flash_attn_2_available, is_flash_attn_greater_or_equal_2_10, logging, replace_return_docstrings, ) from configuration_upcycling_qwen2_moe import UpcyclingQwen2MoeConfig from transformers import AutoModelForCausalLM,AutoConfig,AutoModel if is_flash_attn_2_available(): from flash_attn import flash_attn_func, flash_attn_varlen_func from flash_attn.bert_padding import index_first_axis, pad_input, unpad_input # noqa _flash_supports_window_size = "window_size" in list(inspect.signature(flash_attn_func).parameters) logger = logging.get_logger(__name__) _CHECKPOINT_FOR_DOC = "UpcyclingQwen2MoE" _CONFIG_FOR_DOC = "UpcyclingQwen2MoeConfig" # Copied from transformers.models.mixtral.modeling_mixtral.load_balancing_loss_func def load_balancing_loss_func( gate_logits: torch.Tensor, num_experts: torch.Tensor = None, top_k=2, attention_mask: Optional[torch.Tensor] = None ) -> float: r""" Computes auxiliary load balancing loss as in Switch Transformer - implemented in Pytorch. See Switch Transformer (https://arxiv.org/abs/2101.03961) for more details. This function implements the loss function presented in equations (4) - (6) of the paper. It aims at penalizing cases where the routing between experts is too unbalanced. Args: gate_logits (Union[`torch.Tensor`, Tuple[torch.Tensor]): Logits from the `gate`, should be a tuple of model.config.num_hidden_layers tensors of shape [batch_size X sequence_length, num_experts]. attention_mask (`torch.Tensor`, None): The attention_mask used in forward function shape [batch_size X sequence_length] if not None. num_experts (`int`, *optional*): Number of experts Returns: The auxiliary loss. """ if gate_logits is None or not isinstance(gate_logits, tuple): return 0 if isinstance(gate_logits, tuple): compute_device = gate_logits[0].device concatenated_gate_logits = torch.cat([layer_gate.to(compute_device) for layer_gate in gate_logits], dim=0) routing_weights = torch.nn.functional.softmax(concatenated_gate_logits, dim=-1) _, selected_experts = torch.topk(routing_weights, top_k, dim=-1) expert_mask = torch.nn.functional.one_hot(selected_experts, num_experts) if attention_mask is None: # Compute the percentage of tokens routed to each experts tokens_per_expert = torch.mean(expert_mask.float(), dim=0) # Compute the average probability of routing to these experts router_prob_per_expert = torch.mean(routing_weights, dim=0) else: batch_size, sequence_length = attention_mask.shape num_hidden_layers = concatenated_gate_logits.shape[0] // (batch_size * sequence_length) # Compute the mask that masks all padding tokens as 0 with the same shape of expert_mask expert_attention_mask = ( attention_mask[None, :, :, None, None] .expand((num_hidden_layers, batch_size, sequence_length, top_k, num_experts)) .reshape(-1, top_k, num_experts) .to(compute_device) ) # Compute the percentage of tokens routed to each experts tokens_per_expert = torch.sum(expert_mask.float() * expert_attention_mask, dim=0) / torch.sum( expert_attention_mask, dim=0 ) # Compute the mask that masks all padding tokens as 0 with the same shape of tokens_per_expert router_per_expert_attention_mask = ( attention_mask[None, :, :, None] .expand((num_hidden_layers, batch_size, sequence_length, num_experts)) .reshape(-1, num_experts) .to(compute_device) ) # Compute the average probability of routing to these experts router_prob_per_expert = torch.sum(routing_weights * router_per_expert_attention_mask, dim=0) / torch.sum( router_per_expert_attention_mask, dim=0 ) overall_loss = torch.sum(tokens_per_expert * router_prob_per_expert.unsqueeze(0)) return overall_loss * num_experts # Copied from transformers.models.llama.modeling_llama._get_unpad_data def _get_unpad_data(attention_mask): seqlens_in_batch = attention_mask.sum(dim=-1, dtype=torch.int32) indices = torch.nonzero(attention_mask.flatten(), as_tuple=False).flatten() max_seqlen_in_batch = seqlens_in_batch.max().item() cu_seqlens = F.pad(torch.cumsum(seqlens_in_batch, dim=0, dtype=torch.int32), (1, 0)) return ( indices, cu_seqlens, max_seqlen_in_batch, ) # Copied from transformers.models.llama.modeling_llama.LlamaRMSNorm with Llama->Qwen2Moe class Qwen2MoeRMSNorm(nn.Module): def __init__(self, hidden_size, eps=1e-6): """ Qwen2MoeRMSNorm is equivalent to T5LayerNorm """ 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) variance = hidden_states.pow(2).mean(-1, keepdim=True) hidden_states = hidden_states * torch.rsqrt(variance + self.variance_epsilon) return self.weight * hidden_states.to(input_dtype) # Copied from transformers.models.mixtral.modeling_mixtral.MixtralRotaryEmbedding with Mixtral->Qwen2Moe class Qwen2MoeRotaryEmbedding(nn.Module): def __init__(self, dim, max_position_embeddings=2048, base=10000, device=None): super().__init__() 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) # Build here to make `torch.jit.trace` work. self._set_cos_sin_cache( seq_len=max_position_embeddings, device=self.inv_freq.device, dtype=torch.get_default_dtype() ) def _set_cos_sin_cache(self, seq_len, device, dtype): self.max_seq_len_cached = seq_len t = torch.arange(self.max_seq_len_cached, device=device, dtype=torch.int64).type_as(self.inv_freq) freqs = torch.outer(t, self.inv_freq) # Different from paper, but it uses a different permutation in order to obtain the same calculation emb = torch.cat((freqs, freqs), dim=-1) self.register_buffer("cos_cached", emb.cos().to(dtype), persistent=False) self.register_buffer("sin_cached", emb.sin().to(dtype), persistent=False) def forward(self, x, seq_len=None): # x: [bs, num_attention_heads, seq_len, head_size] if seq_len > self.max_seq_len_cached: self._set_cos_sin_cache(seq_len=seq_len, device=x.device, dtype=x.dtype) return ( self.cos_cached[:seq_len].to(dtype=x.dtype), self.sin_cached[:seq_len].to(dtype=x.dtype), ) # Copied from transformers.models.llama.modeling_llama.rotate_half def rotate_half(x): """Rotates half the hidden dims of the input.""" x1 = x[..., : x.shape[-1] // 2] x2 = x[..., x.shape[-1] // 2 :] return torch.cat((-x2, x1), dim=-1) # Copied from transformers.models.mixtral.modeling_mixtral.apply_rotary_pos_emb def apply_rotary_pos_emb(q, k, cos, sin, position_ids, 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`): The position indices of the tokens corresponding to the query and key tensors. For example, this can be used to pass offsetted position ids when working with a KV-cache. unsqueeze_dim (`int`, *optional*, defaults to 1): The 'unsqueeze_dim' argument specifies the dimension along which to unsqueeze cos[position_ids] and sin[position_ids] so that they can be properly broadcasted to the dimensions of q and k. For example, note that cos[position_ids] and sin[position_ids] have the shape [batch_size, seq_len, head_dim]. Then, if q and k have the shape [batch_size, heads, seq_len, head_dim], then setting unsqueeze_dim=1 makes cos[position_ids] and sin[position_ids] broadcastable to the shapes of q and k. Similarly, if q and k have the shape [batch_size, seq_len, heads, head_dim], then set unsqueeze_dim=2. Returns: `tuple(torch.Tensor)` comprising of the query and key tensors rotated using the Rotary Position Embedding. """ cos = cos[position_ids].unsqueeze(unsqueeze_dim) sin = sin[position_ids].unsqueeze(unsqueeze_dim) q_embed = (q * cos) + (rotate_half(q) * sin) k_embed = (k * cos) + (rotate_half(k) * sin) return q_embed, k_embed # Modified from transformers.models.mistral.modeling_mistral.MistralMLP with Mistral->Qwen2Moe class Qwen2MoeMLP(nn.Module): def __init__(self, config, intermediate_size=None): super().__init__() self.config = config self.hidden_size = config.hidden_size self.intermediate_size = 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,language_ids:Optional[torch.LongTensor]=None): return self.down_proj(self.act_fn(self.gate_proj(x)) * self.up_proj(x)) # 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, 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) # Copied from transformers.models.qwen2.modeling_qwen2.Qwen2Attention with Qwen2->Qwen2Moe class Qwen2MoeAttention(nn.Module): """ Multi-headed attention from 'Attention Is All You Need' paper. Modified to use sliding window attention: Longformer and "Generating Long Sequences with Sparse Transformers". """ def __init__(self, config: UpcyclingQwen2MoeConfig, 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 `layer_idx` is not recommended and will " "to errors during the forward call, if caching is used. Please make sure to provide a `layer_idx` " "when creating this class." ) 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.attention_dropout = config.attention_dropout 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})." ) self.q_proj = nn.Linear(self.hidden_size, self.num_heads * self.head_dim, bias=True) self.k_proj = nn.Linear(self.hidden_size, self.num_key_value_heads * self.head_dim, bias=True) self.v_proj = nn.Linear(self.hidden_size, self.num_key_value_heads * self.head_dim, bias=True) self.o_proj = nn.Linear(self.num_heads * self.head_dim, self.hidden_size, bias=False) self.rotary_emb = Qwen2MoeRotaryEmbedding( 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, ) -> 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).transpose(1, 2) key_states = key_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2) value_states = value_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2) kv_seq_len = key_states.shape[-2] if past_key_value is not None: if self.layer_idx is None: raise ValueError( f"The cache structure has changed since version v4.36. If you are using {self.__class__.__name__} " "for auto-regressive decoding with k/v caching, please make sure to initialize the attention class " "with a layer index." ) kv_seq_len += past_key_value.get_usable_length(kv_seq_len, self.layer_idx) cos, sin = self.rotary_emb(value_states, seq_len=kv_seq_len) query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin, position_ids) if past_key_value is not None: cache_kwargs = {"sin": sin, "cos": cos, "cache_position": cache_position} # Specific to RoPE models key_states, value_states = past_key_value.update(key_states, value_states, self.layer_idx, cache_kwargs) # repeat k/v heads if n_kv_heads < n_heads 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 attn_weights.size() != (bsz, self.num_heads, q_len, kv_seq_len): raise ValueError( f"Attention weights should be of size {(bsz, self.num_heads, q_len, kv_seq_len)}, but is" f" {attn_weights.size()}" ) 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 # Copied from transformers.models.qwen2.modeling_qwen2.Qwen2FlashAttention2 with Qwen2->Qwen2Moe class Qwen2MoeFlashAttention2(Qwen2MoeAttention): """ Qwen2Moe flash attention module, following Qwen2Moe attention module. This module inherits from `Qwen2MoeAttention` 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. Additionally, for sliding window attention, we apply SWA only to the bottom config.max_window_layers layers. """ # Copied from transformers.models.llama.modeling_llama.LlamaFlashAttention2.__init__ 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() 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, ): 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).transpose(1, 2) key_states = key_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2) value_states = value_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2) kv_seq_len = key_states.shape[-2] if past_key_value is not None: if self.layer_idx is None: raise ValueError( f"The cache structure has changed since version v4.36. If you are using {self.__class__.__name__} " "for auto-regressive decoding with k/v caching, please make sure to initialize the attention class " "with a layer index." ) kv_seq_len += past_key_value.get_usable_length(kv_seq_len, self.layer_idx) # Because the input can be padded, the absolute sequence length depends on the max position id. rotary_seq_len = max(kv_seq_len, position_ids[:, -1].max().item()) + 1 cos, sin = self.rotary_emb(value_states, seq_len=rotary_seq_len) query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin, position_ids) use_sliding_windows = ( _flash_supports_window_size and getattr(self.config, "sliding_window", None) is not None and kv_seq_len > self.config.sliding_window and self.config.use_sliding_window ) if not _flash_supports_window_size: logger.warning_once( "The current flash attention version does not support sliding window attention, for a more memory efficient implementation" " make sure to upgrade flash-attn library." ) if past_key_value is not None: # Activate slicing cache only if the config has a value `sliding_windows` attribute cache_has_contents = past_key_value.get_seq_length(self.layer_idx) > 0 if ( getattr(self.config, "sliding_window", None) is not None and kv_seq_len > self.config.sliding_window and cache_has_contents ): slicing_tokens = 1 - self.config.sliding_window past_key = past_key_value[self.layer_idx][0] past_value = past_key_value[self.layer_idx][1] past_key = past_key[:, :, slicing_tokens:, :].contiguous() past_value = past_value[:, :, slicing_tokens:, :].contiguous() if past_key.shape[-2] != self.config.sliding_window - 1: raise ValueError( f"past key must have a shape of (`batch_size, num_heads, self.config.sliding_window-1, head_dim`), got" f" {past_key.shape}" ) if attention_mask is not None: attention_mask = attention_mask[:, slicing_tokens:] attention_mask = torch.cat([attention_mask, torch.ones_like(attention_mask[:, -1:])], dim=-1) cache_kwargs = {"sin": sin, "cos": cos, "cache_position": cache_position} # Specific to RoPE models key_states, value_states = past_key_value.update(key_states, value_states, self.layer_idx, cache_kwargs) # repeat k/v heads if n_kv_heads < n_heads key_states = repeat_kv(key_states, self.num_key_value_groups) value_states = repeat_kv(value_states, self.num_key_value_groups) dropout_rate = 0.0 if not self.training else self.attention_dropout # 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 float16 just to be sure everything works as expected. 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) # Reashape to the expected shape for Flash Attention query_states = query_states.transpose(1, 2) key_states = key_states.transpose(1, 2) value_states = value_states.transpose(1, 2) attn_output = self._flash_attention_forward( query_states, key_states, value_states, attention_mask, q_len, dropout=dropout_rate, use_sliding_windows=use_sliding_windows, ) 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 def _flash_attention_forward( self, query_states, key_states, value_states, attention_mask, query_length, dropout=0.0, softmax_scale=None, use_sliding_windows=False, ): """ Calls the forward method of Flash Attention - if the input hidden states contain at least one padding token first unpad the input, then computes the attention scores and pad the final attention scores. Args: query_states (`torch.Tensor`): Input query states to be passed to Flash Attention API key_states (`torch.Tensor`): Input key states to be passed to Flash Attention API value_states (`torch.Tensor`): Input value states to be passed to Flash Attention API attention_mask (`torch.Tensor`): The padding mask - corresponds to a tensor of size `(batch_size, seq_len)` where 0 stands for the position of padding tokens and 1 for the position of non-padding tokens. dropout (`float`): Attention dropout softmax_scale (`float`, *optional*): The scaling of QK^T before applying softmax. Default to 1 / sqrt(head_dim) use_sliding_windows (`bool`, *optional*): Whether to activate sliding window attention. """ if not self._flash_attn_uses_top_left_mask: causal = self.is_causal else: # TODO: Remove the `query_length != 1` check once Flash Attention for RoCm is bumped to 2.1. For details, please see the comment in LlamaFlashAttention2 __init__. causal = self.is_causal and query_length != 1 # Decide whether to use SWA or not by layer index. if use_sliding_windows and self.layer_idx >= self.config.max_window_layers: use_sliding_windows = False # Contains at least one padding token in the sequence if attention_mask is not None: batch_size = query_states.shape[0] query_states, key_states, value_states, indices_q, cu_seq_lens, max_seq_lens = self._upad_input( query_states, key_states, value_states, attention_mask, query_length ) cu_seqlens_q, cu_seqlens_k = cu_seq_lens max_seqlen_in_batch_q, max_seqlen_in_batch_k = max_seq_lens if not use_sliding_windows: attn_output_unpad = flash_attn_varlen_func( query_states, key_states, value_states, cu_seqlens_q=cu_seqlens_q, cu_seqlens_k=cu_seqlens_k, max_seqlen_q=max_seqlen_in_batch_q, max_seqlen_k=max_seqlen_in_batch_k, dropout_p=dropout, softmax_scale=softmax_scale, causal=causal, ) else: attn_output_unpad = flash_attn_varlen_func( query_states, key_states, value_states, cu_seqlens_q=cu_seqlens_q, cu_seqlens_k=cu_seqlens_k, max_seqlen_q=max_seqlen_in_batch_q, max_seqlen_k=max_seqlen_in_batch_k, dropout_p=dropout, softmax_scale=softmax_scale, causal=causal, window_size=(self.config.sliding_window, self.config.sliding_window), ) attn_output = pad_input(attn_output_unpad, indices_q, batch_size, query_length) else: if not use_sliding_windows: attn_output = flash_attn_func( query_states, key_states, value_states, dropout, softmax_scale=softmax_scale, causal=causal, ) else: attn_output = flash_attn_func( query_states, key_states, value_states, dropout, softmax_scale=softmax_scale, causal=causal, window_size=(self.config.sliding_window, self.config.sliding_window), ) return attn_output # Copied from transformers.models.mistral.modeling_mistral.MistralFlashAttention2._upad_input def _upad_input(self, query_layer, key_layer, value_layer, attention_mask, query_length): batch_size, kv_seq_len, num_heads, head_dim = key_layer.shape # On the first iteration we need to properly re-create the padding mask # by slicing it on the proper place if kv_seq_len != attention_mask.shape[-1]: attention_mask_num_tokens = attention_mask.shape[-1] attention_mask = attention_mask[:, attention_mask_num_tokens - kv_seq_len :] indices_k, cu_seqlens_k, max_seqlen_in_batch_k = _get_unpad_data(attention_mask) key_layer = index_first_axis(key_layer.reshape(batch_size * kv_seq_len, num_heads, head_dim), indices_k) value_layer = index_first_axis(value_layer.reshape(batch_size * kv_seq_len, num_heads, head_dim), indices_k) if query_length == kv_seq_len: query_layer = index_first_axis( query_layer.reshape(batch_size * kv_seq_len, num_heads, head_dim), indices_k ) cu_seqlens_q = cu_seqlens_k max_seqlen_in_batch_q = max_seqlen_in_batch_k indices_q = indices_k elif query_length == 1: max_seqlen_in_batch_q = 1 cu_seqlens_q = torch.arange( batch_size + 1, dtype=torch.int32, device=query_layer.device ) # There is a memcpy here, that is very bad. indices_q = cu_seqlens_q[:-1] query_layer = query_layer.squeeze(1) else: # The -q_len: slice assumes left padding. attention_mask = attention_mask[:, -query_length:] query_layer, indices_q, cu_seqlens_q, max_seqlen_in_batch_q = unpad_input(query_layer, attention_mask) return ( query_layer, key_layer, value_layer, indices_q, (cu_seqlens_q, cu_seqlens_k), (max_seqlen_in_batch_q, max_seqlen_in_batch_k), ) # Copied from transformers.models.mixtral.modeling_mixtral.MixtralSdpaAttention with Mixtral->Qwen2Moe class Qwen2MoeSdpaAttention(Qwen2MoeAttention): """ Qwen2Moe attention module using torch.nn.functional.scaled_dot_product_attention. This module inherits from `Qwen2MoeAttention` as the weights of the module stays untouched. The only changes are on the forward pass to adapt to SDPA API. """ # Adapted from Qwen2MoeAttention.forward 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( "Qwen2MoeModel is using Qwen2MoeSdpaAttention, 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, ) 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).transpose(1, 2) key_states = key_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2) value_states = value_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2) kv_seq_len = key_states.shape[-2] if past_key_value is not None: kv_seq_len += past_key_value.get_usable_length(kv_seq_len, self.layer_idx) cos, sin = self.rotary_emb(value_states, seq_len=kv_seq_len) query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin, position_ids) if past_key_value is not None: cache_kwargs = {"sin": sin, "cos": cos, "cache_position": cache_position} # Specific to RoPE models 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: # no matter the length, we just slice it causal_mask = attention_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 attention_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. # The q_len > 1 is necessary to match with AttentionMaskConverter.to_causal_4d that does not create a causal mask in case q_len == 1. 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 QWEN2MOE_ATTENTION_CLASSES = { "eager": Qwen2MoeAttention, "flash_attention_2": Qwen2MoeFlashAttention2, "sdpa": Qwen2MoeSdpaAttention, } class Qwen2MoeSparseMoeBlock(nn.Module): def __init__(self, config): super().__init__() self.num_experts = config.num_experts self.top_k = config.num_experts_per_tok self.norm_topk_prob = config.norm_topk_prob # gating self.gate = nn.Linear(config.hidden_size, config.num_experts, bias=False) self.experts = nn.ModuleList( [Qwen2MoeMLP(config, intermediate_size=config.moe_intermediate_size) for _ in range(self.num_experts)] ) #share self.share_flag=config.share_flag if self.share_flag: self.shared_expert = Qwen2MoeMLP(config, intermediate_size=config.shared_expert_intermediate_size) self.shared_expert_gate = torch.nn.Linear(config.hidden_size, 1, bias=False) #language-specific self.language_gate=config.language_gate def forward(self, hidden_states: torch.Tensor,language_ids:Optional[torch.LongTensor] = None) -> torch.Tensor: batch_size, sequence_length, hidden_dim = hidden_states.shape hidden_states = hidden_states.view(-1, hidden_dim) if self.language_gate and self.training : if language_ids is None: raise ValueError('language_ids is not initialized') language_ids=language_ids.view(batch_size*sequence_length,-1) # router_logits: (batch * sequence_length, n_experts) router_logits = self.gate(hidden_states) routing_weights = F.softmax(router_logits, dim=1, dtype=torch.float) _, selected_experts = torch.topk(routing_weights, self.top_k, dim=-1) #language specific select one expert if self.language_gate and self.training: if language_ids is None: raise ValueError('language_ids is not initialized') assert language_ids.shape[0]==selected_experts.shape[0],f'{language_ids.shape},{selected_experts.shape}' language_experts=language_ids.to(selected_experts.dtype) mask=torch.sum((language_experts==selected_experts).int(),dim=1,keepdims=True).bool() selected_experts[:,-1]=torch.where(mask.squeeze(),selected_experts[:,-1],language_experts.squeeze()) routing_weights=torch.gather(routing_weights,1,selected_experts) else: routing_weights, selected_experts = torch.topk(routing_weights, self.top_k, dim=-1) if self.norm_topk_prob: routing_weights /= routing_weights.sum(dim=-1, keepdim=True) # we cast back to the input dtype routing_weights = routing_weights.to(hidden_states.dtype) final_hidden_states = torch.zeros( (batch_size * sequence_length, hidden_dim), dtype=hidden_states.dtype, device=hidden_states.device ) # One hot encode the selected experts to create an expert mask # this will be used to easily index which expert is going to be sollicitated expert_mask = torch.nn.functional.one_hot(selected_experts, num_classes=self.num_experts).permute(2, 1, 0) # Loop over all available experts in the model and perform the computation on each expert for expert_idx in range(self.num_experts): expert_layer = self.experts[expert_idx] idx, top_x = torch.where(expert_mask[expert_idx]) # Index the correct hidden states and compute the expert hidden state for # the current expert. We need to make sure to multiply the output hidden # states by `routing_weights` on the corresponding tokens (top-1 and top-2) current_state = hidden_states[None, top_x].reshape(-1, hidden_dim) current_hidden_states = expert_layer(current_state) * routing_weights[top_x, idx, None] # However `index_add_` only support torch tensors for indexing so we'll use # the `top_x` tensor here. final_hidden_states.index_add_(0, top_x, current_hidden_states.to(hidden_states.dtype)) if self.share_flag: shared_expert_output = self.shared_expert(hidden_states) shared_expert_output = F.sigmoid(self.shared_expert_gate(hidden_states)) * shared_expert_output final_hidden_states = final_hidden_states + shared_expert_output final_hidden_states = final_hidden_states.reshape(batch_size, sequence_length, hidden_dim) return final_hidden_states, router_logits class Qwen2MoeDecoderLayer(nn.Module): def __init__(self, config: UpcyclingQwen2MoeConfig, layer_idx: int): super().__init__() self.hidden_size = config.hidden_size self.self_attn = QWEN2MOE_ATTENTION_CLASSES[config._attn_implementation](config, layer_idx) if (layer_idx not in config.mlp_only_layers) and ( config.num_experts > 0 and (layer_idx + 1) % config.decoder_sparse_step == 0 ): self.mlp = Qwen2MoeSparseMoeBlock(config) else: self.mlp = Qwen2MoeMLP(config, intermediate_size=config.intermediate_size) self.input_layernorm = Qwen2MoeRMSNorm(config.hidden_size, eps=config.rms_norm_eps) self.post_attention_layernorm = Qwen2MoeRMSNorm(config.hidden_size, eps=config.rms_norm_eps) def forward( self, hidden_states: torch.Tensor, language_ids:Optional[torch.LongTensor] = None, 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, output_router_logits: 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, sequence_length)` where padding elements are indicated by 0. output_attentions (`bool`, *optional*): Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned tensors for more detail. output_router_logits (`bool`, *optional*): Whether or not to return the logits of all the routers. They are useful for computing the router loss, and should not be returned during inference. 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 cache_position (`torch.LongTensor` of shape `(sequence_length)`, *optional*): Indices depicting the position of the input sequence tokens in the sequence. """ residual = hidden_states hidden_states = self.input_layernorm(hidden_states) # Self Attention hidden_states, 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, ) hidden_states = residual + hidden_states # Fully Connected residual = hidden_states hidden_states = self.post_attention_layernorm(hidden_states) hidden_states = self.mlp(hidden_states,language_ids) if isinstance(hidden_states, tuple): hidden_states, router_logits = hidden_states else: router_logits = None hidden_states = residual + hidden_states outputs = (hidden_states,) if output_attentions: outputs += (self_attn_weights,) if use_cache: outputs += (present_key_value,) if output_router_logits: outputs += (router_logits,) return outputs class UpcyclingQwen2MoePreTrainedModel(PreTrainedModel): config_class = UpcyclingQwen2MoeConfig base_model_prefix = "model" supports_gradient_checkpointing = True _no_split_modules = ["Qwen2MoeDecoderLayer"] _skip_keys_device_placement = "past_key_values" _supports_flash_attn_2 = True _supports_sdpa = True _supports_cache_class = 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_() @classmethod def from_qwen(cls, pretrained_model_name_or_path, *model_args, **kwargs): share_flag=kwargs.pop('share_flag') attn_init_change=kwargs.pop('attn_init_change') language_gate=kwargs.pop('language_gate') config = cls.config_class.from_pretrained(pretrained_model_name_or_path) config.share_flag=True if isinstance(share_flag,bool) and share_flag else False config.attn_init_change=True if isinstance(attn_init_change,bool) and attn_init_change else False config.language_gate=True if isinstance(language_gate,bool) and language_gate else False print('share_flag',config.share_flag) print('attn_init_change',config.attn_init_change) print('language_gate',config.language_gate) config.num_experts_per_tok = config.num_experts_per_tok if not config.share_flag else config.num_experts_per_tok-1 config.num_experts = config.num_experts if not config.share_flag else config.num_experts-1 base_model = AutoModelForCausalLM.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs) base_cls = type(base_model) print(cls.config_class,cls) #create auto_map #allows you to use your custom model with the auto-API (but doesn’t share any custom code with other users). cls.config_class.register_for_auto_class() cls.register_for_auto_class('AutoModelForCausalLM') # assert base_cls.__name__ == "Qwen2ForCausalLM", f"Invalid convert base model type: {base_cls}" model = cls(config) print(f"converting {base_cls.__name__} to {cls.__name__}") #MoE architechture model_dict=model.state_dict() base_model_dict = base_model.state_dict() #lm_head print('lm_head.weight',model_dict['lm_head.weight'],base_model_dict['lm_head.weight']) shared_keys=set(model_dict)&set(base_model_dict) init_keys=[] #attention for k in shared_keys: if k not in init_keys and 'self_attn' in k: init_keys.append(k) if not config.attn_init_change: model_dict[k]=base_model_dict[k] if config.attn_init_change: #initilization with upper and lower for layer_id in range(config.num_hidden_layers): if layer_id ==0 or config.num_hidden_layers-1: model_dict[f'model.layers.{layer_id}.self_attn.q_proj.bias']=base_model_dict[f'model.layers.{layer_id}.self_attn.q_proj.bias'] model_dict[f'model.layers.{layer_id}.self_attn.q_proj.weight']=base_model_dict[f'model.layers.{layer_id}.self_attn.q_proj.weight'] model_dict[f'model.layers.{layer_id}.self_attn.k_proj.bias']=base_model_dict[f'model.layers.{layer_id}.self_attn.k_proj.bias'] model_dict[f'model.layers.{layer_id}.self_attn.k_proj.weight']=base_model_dict[f'model.layers.{layer_id}.self_attn.k_proj.weight'] model_dict[f'model.layers.{layer_id}.self_attn.v_proj.bias']=base_model_dict[f'model.layers.{layer_id}.self_attn.v_proj.bias'] model_dict[f'model.layers.{layer_id}.self_attn.v_proj.weight']=base_model_dict[f'model.layers.{layer_id}.self_attn.v_proj.weight'] model_dict[f'model.layers.{layer_id}.self_attn.o_proj.weight']=base_model_dict[f'model.layers.{layer_id}.self_attn.o_proj.weight'] else: model_dict[f'model.layers.{layer_id}.self_attn.q_proj.bias']=1/3*(base_model_dict[f'model.layers.{layer_id}.self_attn.q_proj.bias']+base_model_dict[f'model.layers.{layer_id+1}.self_attn.q_proj.bias']+base_model_dict[f'model.layers.{layer_id-1}.self_attn.q_proj.bias']) model_dict[f'model.layers.{layer_id}.self_attn.q_proj.weight']=1/3*(base_model_dict[f'model.layers.{layer_id}.self_attn.q_proj.weight']+base_model_dict[f'model.layers.{layer_id+1}.self_attn.q_proj.weight']+base_model_dict[f'model.layers.{layer_id-1}.self_attn.q_proj.weight']) model_dict[f'model.layers.{layer_id}.self_attn.k_proj.bias']=1/3*(base_model_dict[f'model.layers.{layer_id}.self_attn.k_proj.bias']+base_model_dict[f'model.layers.{layer_id+1}.self_attn.k_proj.bias']+base_model_dict[f'model.layers.{layer_id-1}.self_attn.k_proj.bias']) model_dict[f'model.layers.{layer_id}.self_attn.k_proj.weight']=1/3*(base_model_dict[f'model.layers.{layer_id}.self_attn.k_proj.weight']+base_model_dict[f'model.layers.{layer_id+1}.self_attn.k_proj.weight']+base_model_dict[f'model.layers.{layer_id-1}.self_attn.k_proj.weight']) model_dict[f'model.layers.{layer_id}.self_attn.v_proj.bias']=1/3*(base_model_dict[f'model.layers.{layer_id}.self_attn.v_proj.bias']+base_model_dict[f'model.layers.{layer_id+1}.self_attn.v_proj.bias']+base_model_dict[f'model.layers.{layer_id-1}.self_attn.v_proj.bias']) model_dict[f'model.layers.{layer_id}.self_attn.v_proj.weight']=1/3*(base_model_dict[f'model.layers.{layer_id}.self_attn.v_proj.weight']+base_model_dict[f'model.layers.{layer_id+1}.self_attn.v_proj.weight']+base_model_dict[f'model.layers.{layer_id-1}.self_attn.v_proj.weight']) model_dict[f'model.layers.{layer_id}.self_attn.o_proj.weight']=1/3*(base_model_dict[f'model.layers.{layer_id}.self_attn.o_proj.weight']+base_model_dict[f'model.layers.{layer_id+1}.self_attn.o_proj.weight']+base_model_dict[f'model.layers.{layer_id-1}.self_attn.o_proj.weight']) #mlp if config.mlp_only_layers: for layer_id in config.mlp_only_layers: key_mapping=sum([ [ (f'model.layers.{layer_id}.mlp.down_proj.weight',f'model.layers.{layer_id}.mlp.down_proj.weight'), (f'model.layers.{layer_id}.mlp.gate_proj.weight',f'model.layers.{layer_id}.mlp.gate_proj.weight'), (f'model.layers.{layer_id}.mlp.up_proj.weight',f'model.layers.{layer_id}.mlp.up_proj.weight'), ]] ,[]) for model_key,base_model_key in key_mapping: model_dict[model_key]=base_model_dict[base_model_key] init_keys.append(model_key) moe_only_layers=list(set(range(config.num_hidden_layers))-set(config.mlp_only_layers)) if config.mlp_only_layers else config.num_hidden_layers #moe-mlp-expert for layer_id in moe_only_layers: key_mapping=sum([ [ (f'model.layers.{layer_id}.mlp.experts.{expert_id}.down_proj.weight',f'model.layers.{layer_id}.mlp.down_proj.weight'), (f'model.layers.{layer_id}.mlp.experts.{expert_id}.gate_proj.weight',f'model.layers.{layer_id}.mlp.gate_proj.weight'), (f'model.layers.{layer_id}.mlp.experts.{expert_id}.up_proj.weight',f'model.layers.{layer_id}.mlp.up_proj.weight'), ] for expert_id in range(config.num_experts)] ,[]) for model_key,base_model_key in key_mapping: model_dict[model_key]=base_model_dict[base_model_key] init_keys.append(model_key) #model_dict[f'model.layers.{layer_id}.mlp.gate.weight'] #share expert if config.share_flag: shared_key_mapping=sum([[ (f'model.layers.{layer_id}.mlp.shared_expert.down_proj.weight',f'model.layers.{layer_id}.mlp.down_proj.weight'), (f'model.layers.{layer_id}.mlp.shared_expert.gate_proj.weight',f'model.layers.{layer_id}.mlp.gate_proj.weight'), (f'model.layers.{layer_id}.mlp.shared_expert.up_proj.weight',f'model.layers.{layer_id}.mlp.up_proj.weight'), ]for layer_id in range(config.num_hidden_layers)], []) for model_key,base_model_key in shared_key_mapping: model_dict[model_key]=base_model_dict[base_model_key] init_keys.append(model_key) # model_dict[f'model.layers.{layer_id}.mlp.shared_expert_gate.weight'] #norm for k in shared_keys: if k not in init_keys: #input_layernorm.weight,post_attention_layernorm.weight,norm.weight # embed_token.weight,lm_head.weight model_dict[k]=base_model_dict[k] init_keys.append(k) gate_initialized = False shared_gate_initilizaed=False for key in model_dict.keys(): if key in init_keys: continue if "mlp.gate.weight" in key: if gate_initialized: continue gate_initialized = True print(f"{cls.__name__} key [{cls.base_model_prefix}.layers.[0-{config.num_hidden_layers-1}].mlp.gate.weight] is not initialized from {base_cls.__name__}. e.g, {key}") continue if 'shared_expert_gate.weight' in key: if shared_gate_initilizaed: continue shared_gate_initilizaed = True print(f"{cls.__name__} key [{cls.base_model_prefix}.layers.[0-{config.num_hidden_layers-1}].mlp.shared_expert_gate.weight] is not initialized from {base_cls.__name__}. e.g, {key}") continue raise NotImplementedError(f"{cls.__name__} key [{key}] is not correctly initilized from {base_cls.__name__}.") model.load_state_dict(model_dict) print(f"Done converted, alreadly check all parameters of {cls.__name__} are initialized from {base_cls.__name__}.") del base_model return model @classmethod def from_btx(cls, pretrained_model_name_or_path, *model_args, **kwargs): share_flag=kwargs.pop('share_flag') attn_init_change=kwargs.pop('attn_init_change') language_gate=kwargs.pop('language_gate') config = cls.config_class.from_pretrained(pretrained_model_name_or_path) config.share_flag=True if isinstance(share_flag,bool) and share_flag else False config.attn_init_change=True if isinstance(attn_init_change,bool) and attn_init_change else False config.language_gate=True if isinstance(language_gate,bool) and language_gate else False print('share_flag',config.share_flag) print('attn_init_change',config.attn_init_change) print('language_gate',config.language_gate) config.num_experts_per_tok = config.num_experts_per_tok if not config.share_flag else config.num_experts_per_tok-1 config.num_experts = config.num_experts if not config.share_flag else config.num_experts-1 base_model = AutoModelForCausalLM.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs) base_cls = type(base_model) print(cls.config_class,cls) #create auto_map #allows you to use your custom model with the auto-API (but doesn’t share any custom code with other users). cls.config_class.register_for_auto_class() cls.register_for_auto_class('AutoModelForCausalLM') # assert base_cls.__name__ == "Qwen2ForCausalLM", f"Invalid convert base model type: {base_cls}" model = cls(config) print(f"converting {base_cls.__name__} to {cls.__name__}") #MoE architechture model_dict=model.state_dict() base_model_dict = base_model.state_dict() #lm_head print('lm_head.weight',model_dict['lm_head.weight'],base_model_dict['lm_head.weight']) shared_keys=set(model_dict)&set(base_model_dict) init_keys=[] #attention for k in shared_keys: init_keys.append(k) model_dict[k]=base_model_dict[k] gate_initialized = False shared_gate_initilizaed=False for key in model_dict.keys(): if key in init_keys: continue if "mlp.gate.weight" in key: if gate_initialized: continue gate_initialized = True print(f"{cls.__name__} key [{cls.base_model_prefix}.layers.[0-{config.num_hidden_layers-1}].mlp.gate.weight] is not initialized from {base_cls.__name__}. e.g, {key}") continue if 'shared_expert_gate.weight' in key: if shared_gate_initilizaed: continue shared_gate_initilizaed = True print(f"{cls.__name__} key [{cls.base_model_prefix}.layers.[0-{config.num_hidden_layers-1}].mlp.shared_expert_gate.weight] is not initialized from {base_cls.__name__}. e.g, {key}") continue raise NotImplementedError(f"{cls.__name__} key [{key}] is not correctly initilized from {base_cls.__name__}.") model.load_state_dict(model_dict) print(f"Done converted, alreadly check all parameters of {cls.__name__} are initialized from {base_cls.__name__}.") del base_model return model class UpcyclingQwen2MoeModel(UpcyclingQwen2MoePreTrainedModel): """ Transformer decoder consisting of *config.num_hidden_layers* layers. Each layer is a [`Qwen2MoeDecoderLayer`] Args: config: Qwen2MoeConfig """ def __init__(self, config: UpcyclingQwen2MoeConfig): 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( [Qwen2MoeDecoderLayer(config, layer_idx) for layer_idx in range(config.num_hidden_layers)] ) self._attn_implementation = config._attn_implementation self.norm = Qwen2MoeRMSNorm(config.hidden_size, eps=config.rms_norm_eps) 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 def forward( self, input_ids: torch.LongTensor = None, language_ids :Optional[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, use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, output_router_logits: Optional[bool] = None, return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[Tuple, MoeModelOutputWithPast]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_router_logits = ( output_router_logits if output_router_logits is not None else self.config.output_router_logits ) 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 return_dict = return_dict if return_dict is not None else self.config.use_return_dict if (input_ids is None) ^ (inputs_embeds is not None): raise ValueError( "You cannot specify both input_ids and inputs_embeds at the same time, and must specify either one" ) if self.gradient_checkpointing and self.training: if use_cache: logger.warning_once( "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..." ) use_cache = False use_legacy_cache = False if use_cache and not isinstance(past_key_values, Cache): use_legacy_cache = True past_key_values = DynamicCache.from_legacy_cache(past_key_values) logger.warning_once( "We detected that you are passing `past_key_values` as a tuple and this is deprecated and will be removed in v4.43. " "Please use an appropriate `Cache` class (https://huggingface.co/docs/transformers/v4.41.3/en/internal/generation_utils#transformers.Cache)" ) if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) 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 # decoder layers all_hidden_states = () if output_hidden_states else None all_self_attns = () if output_attentions else None all_router_logits = () if output_router_logits else None next_decoder_cache = None for decoder_layer in self.layers: if output_hidden_states: all_hidden_states += (hidden_states,) if self.gradient_checkpointing and self.training: layer_outputs = self._gradient_checkpointing_func( decoder_layer.__call__, hidden_states, language_ids, causal_mask, position_ids, past_key_values, output_attentions, output_router_logits, use_cache, cache_position, ) else: layer_outputs = decoder_layer( hidden_states, language_ids, attention_mask=causal_mask, position_ids=position_ids, past_key_value=past_key_values, output_attentions=output_attentions, output_router_logits=output_router_logits, use_cache=use_cache, cache_position=cache_position, ) hidden_states = layer_outputs[0] if use_cache: next_decoder_cache = layer_outputs[2 if output_attentions else 1] if output_attentions: all_self_attns += (layer_outputs[1],) if output_router_logits and layer_outputs[-1] is not None: all_router_logits += (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,) next_cache = None if use_cache: next_cache = next_decoder_cache.to_legacy_cache() if use_legacy_cache else next_decoder_cache if not return_dict: return tuple( v for v in [hidden_states, next_cache, all_hidden_states, all_self_attns, all_router_logits] if v is not None ) return MoeModelOutputWithPast( last_hidden_state=hidden_states, past_key_values=next_cache, hidden_states=all_hidden_states, attentions=all_self_attns, router_logits=all_router_logits, ) # Copied from transformers.models.llama.modeling_llama.LlamaModel._update_causal_mask def _update_causal_mask( self, attention_mask: torch.Tensor, input_tensor: torch.Tensor, cache_position: torch.Tensor, past_key_values: Cache, output_attentions: bool, ): # TODO: As of torch==2.2.0, the `attention_mask` passed to the model in `generate` is 2D and of dynamic length even when the static # KV cache is used. This is an issue for torch.compile which then recaptures cudagraphs at each decode steps due to the dynamic shapes. # (`recording cudagraph tree for symint key 13`, etc.), which is VERY slow. A workaround is `@torch.compiler.disable`, but this prevents using # `fullgraph=True`. See more context in https://github.com/huggingface/transformers/pull/29114 if self.config._attn_implementation == "flash_attention_2": if attention_mask is not None and 0.0 in attention_mask: return attention_mask return None # For SDPA, when possible, we will rely on its `is_causal` argument instead of its `attn_mask` argument, in # order to dispatch on Flash Attention 2. This feature is not compatible with static cache, as SDPA will fail # to infer the attention mask. past_seen_tokens = past_key_values.get_seq_length() if past_key_values is not None else 0 using_static_cache = isinstance(past_key_values, StaticCache) # When output attentions is True, sdpa implementation's forward method calls the eager implementation's forward if self.config._attn_implementation == "sdpa" and not using_static_cache and not output_attentions: if AttentionMaskConverter._ignore_causal_mask_sdpa( attention_mask, inputs_embeds=input_tensor, past_key_values_length=past_seen_tokens, is_training=self.training, ): return None dtype, device = input_tensor.dtype, input_tensor.device min_dtype = torch.finfo(dtype).min sequence_length = input_tensor.shape[1] if using_static_cache: target_length = past_key_values.get_max_length() else: target_length = ( attention_mask.shape[-1] if isinstance(attention_mask, torch.Tensor) else past_seen_tokens + sequence_length + 1 ) 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 if attention_mask.max() != 0: raise ValueError("Custom 4D attention mask should be passed in inverted form with max==0`") causal_mask = attention_mask else: causal_mask = torch.full( (sequence_length, target_length), fill_value=min_dtype, dtype=dtype, device=device ) if sequence_length != 1: causal_mask = torch.triu(causal_mask, diagonal=1) causal_mask *= torch.arange(target_length, device=device) > cache_position.reshape(-1, 1) causal_mask = causal_mask[None, None, :, :].expand(input_tensor.shape[0], 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, :] padding_mask = padding_mask == 0 causal_mask[:, :, :, :mask_length] = causal_mask[:, :, :, :mask_length].masked_fill( padding_mask, min_dtype ) if ( self.config._attn_implementation == "sdpa" and attention_mask is not None and attention_mask.device.type == "cuda" and not output_attentions ): # Attend to all tokens in fully masked rows in the causal_mask, for example the relevant first rows when # using left padding. This is required by F.scaled_dot_product_attention memory-efficient attention path. # Details: https://github.com/pytorch/pytorch/issues/110213 causal_mask = AttentionMaskConverter._unmask_unattended(causal_mask, min_dtype) return causal_mask class UpcyclingQwen2MoeForCausalLM(UpcyclingQwen2MoePreTrainedModel): _tied_weights_keys = ["lm_head.weight"] def __init__(self, config): super().__init__(config) self.model = UpcyclingQwen2MoeModel(config) self.vocab_size = config.vocab_size self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False) self.router_aux_loss_coef = config.router_aux_loss_coef self.num_experts = config.num_experts self.num_experts_per_tok = config.num_experts_per_tok self.language_gate=config.language_gate # 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 # @replace_return_docstrings(output_type=MoeCausalLMOutputWithPast, config_class=_CONFIG_FOR_DOC) def forward( self, input_ids: torch.LongTensor = None, language_ids: Optional[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, output_router_logits: Optional[bool] = None, return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[Tuple, MoeCausalLMOutputWithPast]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_router_logits = ( output_router_logits if output_router_logits is not None else self.config.output_router_logits ) 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, language_ids=language_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, output_router_logits=output_router_logits, return_dict=return_dict, cache_position=cache_position, ) hidden_states = outputs[0] logits = self.lm_head(hidden_states) 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) aux_loss = None if output_router_logits: aux_loss = load_balancing_loss_func( outputs.router_logits if return_dict else outputs[-1], self.num_experts, self.num_experts_per_tok, attention_mask, ) if labels is not None: loss += self.router_aux_loss_coef * aux_loss.to(loss.device) # make sure to reside in the same device if not return_dict: output = (logits,) + outputs[1:] if output_router_logits: output = (aux_loss,) + output return (loss,) + output if loss is not None else output return MoeCausalLMOutputWithPast( loss=loss, aux_loss=aux_loss, logits=logits, past_key_values=outputs.past_key_values, hidden_states=outputs.hidden_states, attentions=outputs.attentions, router_logits=outputs.router_logits, ) def prepare_inputs_for_generation( self, input_ids, past_key_values=None, attention_mask=None, inputs_embeds=None, cache_position=None, use_cache=True, **kwargs, ): past_length = 0 # ##### by own if past_key_values is not None: if isinstance(past_key_values,Cache): # Past key values are always initialized with a `Cache` object -> no need for if-else anymore past_length = cache_position[0] if cache_position is not None else past_key_values.get_seq_length() max_cache_length = ( torch.tensor(past_key_values.get_max_length(), device=input_ids.device) if past_key_values.get_max_length() is not None else None ) cache_length = past_length if max_cache_length is None else torch.min(max_cache_length, past_length) else: cache_length=past_length=past_key_values[0][0].shape[2] max_cache_length=None # # ##### # Omit tokens covered by past_key_values # if past_key_values is not None: # # Past key values are always initialized with a `Cache` object -> no need for if-else anymore # past_length = cache_position[0] if cache_position is not None else past_key_values.get_seq_length() # max_cache_length = ( # torch.tensor(past_key_values.get_max_length(), device=input_ids.device) # if past_key_values.get_max_length() is not None # else None # ) # cache_length = past_length if max_cache_length is None else torch.min(max_cache_length, past_length) # Keep only the unprocessed tokens: # 1 - If the length of the attention_mask exceeds the length of input_ids, then we are in a setting where # some of the inputs are exclusively passed as part of the cache (e.g. when passing input_embeds as # input) if attention_mask is not None and attention_mask.shape[1] > input_ids.shape[1]: input_ids = input_ids[:, -(attention_mask.shape[1] - past_length) :] # 2 - If the past_length is smaller than input_ids', then input_ids holds all input tokens. We can discard # input_ids based on the past_length. elif past_length < input_ids.shape[1]: input_ids = input_ids[:, past_length:] # 3 - Otherwise (past_length >= input_ids.shape[1]), let's assume input_ids only has unprocessed tokens. # If we are about to go beyond the maximum cache length, we need to crop the input attention mask. if ( max_cache_length is not None and attention_mask is not None and cache_length + input_ids.shape[1] > max_cache_length ): attention_mask = attention_mask[:, -max_cache_length:] position_ids = kwargs.get("position_ids", None) 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] :] # if `inputs_embeds` are passed, we only want to use them in the 1st generation step if inputs_embeds is not None and past_length == 0: model_inputs = {"inputs_embeds": inputs_embeds} else: model_inputs = {"input_ids": input_ids} input_length = position_ids.shape[-1] if position_ids is not None else input_ids.shape[-1] if cache_position is None: cache_position = torch.arange(past_length, past_length + input_length, device=input_ids.device) elif use_cache: cache_position = cache_position[-input_length:] model_inputs.update( { "position_ids": position_ids, "past_key_values": past_key_values, "use_cache": use_cache, "attention_mask": attention_mask, "cache_position": cache_position, } ) return model_inputs @staticmethod def _reorder_cache(past_key_values, beam_idx): reordered_past = () for layer_past in past_key_values: reordered_past += ( tuple(past_state.index_select(0, beam_idx.to(past_state.device)) for past_state in layer_past), ) return reordered_past