diff --git a/python/llm/src/ipex_llm/transformers/convert.py b/python/llm/src/ipex_llm/transformers/convert.py index f9b54663..3ad0bcd7 100644 --- a/python/llm/src/ipex_llm/transformers/convert.py +++ b/python/llm/src/ipex_llm/transformers/convert.py @@ -713,6 +713,9 @@ def _optimize_pre(model): if model.config.model_type == "qwen2": from ipex_llm.transformers.models.qwen2 import merge_qkv model.apply(merge_qkv) + if model.config.model_type == "qwen2_moe": + from ipex_llm.transformers.models.qwen2_moe import merge_qkv + model.apply(merge_qkv) if model.config.model_type == "stablelm": # For stablelm-zephyr-3b and stablelm-2-zephyr-1_6b from ipex_llm.transformers.models.stablelm import merge_qkv @@ -1305,8 +1308,8 @@ def _optimize_post(model, lightweight_bmm=False): modeling_module_name = model.__class__.__module__ module = importlib.import_module(modeling_module_name) from ipex_llm.transformers.models.qwen2_moe import qwen2moe_moeblock_forward - from ipex_llm.transformers.models.qwen2_moe import qwen2moe_attention_forward from ipex_llm.transformers.models.qwen2_moe import qwen2moe_model_forward + from ipex_llm.transformers.models.qwen2 import qwen2_attention_forward convert_forward(model, module.Qwen2MoeModel, qwen2moe_model_forward) @@ -1321,7 +1324,10 @@ def _optimize_post(model, lightweight_bmm=False): llama_mlp_forward) convert_forward(model, module.Qwen2MoeAttention, - qwen2moe_attention_forward) + qwen2_attention_forward) + convert_forward(model, + module.Qwen2MoeSdpaAttention, + qwen2_attention_forward) elif model.config.model_type == "cohere": # for CohereForAI/c4ai-command-r-v01 modeling_module_name = model.__class__.__module__ diff --git a/python/llm/src/ipex_llm/transformers/models/qwen2_moe.py b/python/llm/src/ipex_llm/transformers/models/qwen2_moe.py index 7d9d2b36..be159316 100644 --- a/python/llm/src/ipex_llm/transformers/models/qwen2_moe.py +++ b/python/llm/src/ipex_llm/transformers/models/qwen2_moe.py @@ -37,39 +37,20 @@ # limitations under the License. """ PyTorch Qwen2MoE model.""" -import math import torch import torch.nn.functional as F -import torch.nn as nn import torch.utils.checkpoint -import warnings -from typing import TYPE_CHECKING, Optional, Tuple, Union, Callable, List -from ipex_llm.transformers.models.llama import repeat_kv -from ipex_llm.transformers.models.utils import should_use_fuse_rope -from ipex_llm.transformers.models.utils import extend_kv_cache, append_kv_cache -from ipex_llm.transformers.models.utils import apply_rotary_pos_emb_cache_freq_xpu -from ipex_llm.transformers.models.utils import is_enough_kv_cache_room_4_36 -from transformers.models.qwen2.modeling_qwen2 import apply_rotary_pos_emb +from typing import Optional, Tuple, Union, List from ipex_llm.utils.common import invalidInputError -from ipex_llm.transformers.models.utils import decoding_fast_path_qtype_check -from ipex_llm.transformers.models.utils import use_flash_attention -from transformers.models.qwen2_moe.modeling_qwen2_moe import Qwen2MoeModel, apply_rotary_pos_emb -from ipex_llm.transformers.models.utils import use_quantize_kv_cache, restore_fp8_kv_cache -from ipex_llm.transformers.kv import DynamicFp8Cache +from ipex_llm.transformers.models.utils import use_quantize_kv_cache +from ipex_llm.transformers.kv import DynamicFp8Cache, DynamicNormalCache -import os - -KV_CACHE_ALLOC_BLOCK_LENGTH = int(os.environ.get("KV_CACHE_ALLOC_BLOCK_LENGTH", 256)) - -from transformers.models.qwen2.modeling_qwen2 import _prepare_4d_causal_attention_mask_for_sdpa -from transformers.models.qwen2.modeling_qwen2 import _prepare_4d_causal_attention_mask +from transformers.models.qwen2_moe.modeling_qwen2_moe import ( + _prepare_4d_causal_attention_mask_for_sdpa, _prepare_4d_causal_attention_mask, + Qwen2MoeAttention, +) from transformers.modeling_outputs import MoeModelOutputWithPast - -try: - from transformers.cache_utils import Cache, DynamicCache -except ImportError: - Cache = Tuple[torch.Tensor] -import logging +from transformers.cache_utils import Cache, DynamicCache from transformers import logging @@ -90,9 +71,12 @@ def qwen2moe_model_forward( return_dict: Optional[bool] = None, ): use_cache = use_cache if use_cache is not None else self.config.use_cache - if use_cache and use_quantize_kv_cache(self.layers[0].mlp.shared_expert.up_proj, input_ids): - if not isinstance(past_key_values, DynamicFp8Cache): + use_quantize_kv = use_quantize_kv_cache(self.layers[0].mlp.shared_expert.up_proj, input_ids) + if use_cache: + if use_quantize_kv and not isinstance(past_key_values, DynamicFp8Cache): past_key_values = DynamicFp8Cache.from_legacy_cache(past_key_values) + if not use_quantize_kv and not isinstance(past_key_values, DynamicNormalCache): + past_key_values = DynamicNormalCache.from_legacy_cache(past_key_values) return qwen2_moe_model_forward_internal( self=self, input_ids=input_ids, @@ -290,452 +274,27 @@ def qwen2_moe_model_forward_internal( ) -def qwen2moe_attention_forward( - self, - hidden_states: torch.Tensor, - attention_mask: Optional[torch.Tensor] = None, - position_ids: Optional[torch.LongTensor] = None, - past_key_value: Optional[Tuple[torch.Tensor]] = None, - output_attentions: bool = False, - use_cache: bool = False, - **kwargs, -) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]: - if use_quantize_kv_cache(self.q_proj, hidden_states): - forward_function = qwen2moe_attention_forward_quantized - elif hidden_states.device.type == "cpu": - forward_function = qwen2moe_attention_forward_sdpa - else: - forward_function = qwen2moe_attention_forward_origin - return forward_function( - self=self, - 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, - **kwargs, - ) +def merge_qkv(module: torch.nn.Module): + if isinstance(module, Qwen2MoeAttention): + new_weight = torch.cat([ + module.q_proj.weight.data, + module.k_proj.weight.data, + module.v_proj.weight.data, + ], dim=0) + new_bias = torch.cat([ + module.q_proj.bias.data, + module.k_proj.bias.data, + module.v_proj.bias.data, + ], dim=-1) + qkv_proj = torch.nn.Linear(0, 0, bias=True) + qkv_proj.weight = torch.nn.Parameter(new_weight, requires_grad=False) + qkv_proj.bias = torch.nn.Parameter(new_bias, requires_grad=False) + qkv_proj.in_features = new_weight.size(1) + qkv_proj.out_features = new_weight.size(0) + module.qkv_proj = qkv_proj -def qwen2moe_attention_forward_quantized( - self, - hidden_states: torch.Tensor, - attention_mask: Optional[torch.Tensor] = None, - position_ids: Optional[torch.LongTensor] = None, - past_key_value: Optional[Tuple[torch.Tensor]] = None, - output_attentions: bool = False, - use_cache: bool = False, - **kwargs, -) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]: - if "padding_mask" in kwargs: - warnings.warn( - "Passing `padding_mask` is deprecated and will be removed in v4.37." - "Please make sure use `attention_mask` instead.`" - ) - use_fuse_rope = should_use_fuse_rope(hidden_states, position_ids, self.training) - 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: - invalidInputError(self.layer_idx is not None, - "The cache structure has changed since version v4.36. " - f"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) - - if use_fuse_rope: - query_states, key_states = apply_rotary_pos_emb_cache_freq_xpu(query_states, key_states, - sin, cos, "qwen2_moe", - position_ids) - else: - 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} # Specific to RoPE models - key_states, value_states = past_key_value.update(key_states, value_states, - self.layer_idx, cache_kwargs) - if q_len == 1 and query_states.device.type == 'xpu' and not self.training \ - and not hidden_states.requires_grad: - import xe_addons - attn_weights = xe_addons.query_key_fp8_matmul(query_states, key_states) - else: - key_states, value_states = restore_fp8_kv_cache(key_states, - value_states, query_states.dtype) - # 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)) - - attn_weights = attn_weights / math.sqrt(self.head_dim) - - invalidInputError(attn_weights.size() == (bsz, self.num_heads, q_len, kv_seq_len), - ("Attention weights should be of size " - f"{(bsz, self.num_heads, q_len, kv_seq_len)}," - "but is {attn_weights.size()}")) - - if attention_mask is not None: - invalidInputError(attention_mask.size() == (bsz, 1, q_len, kv_seq_len), - (f"Attention mask should be of size {(bsz, 1, q_len, kv_seq_len)}" - f" but is {attention_mask.size()}")) - - attn_weights = attn_weights + attention_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) - if q_len == 1 and query_states.device.type == 'xpu' and not self.training \ - and not hidden_states.requires_grad: - import xe_addons - attn_output = xe_addons.attn_value_fp8_matmul(attn_weights, value_states) - else: - attn_output = torch.matmul(attn_weights, value_states) - - invalidInputError(attn_output.size() == (bsz, self.num_heads, q_len, self.head_dim), - "`attn_output` should be of size " - f"{(bsz, self.num_heads, q_len, self.head_dim)}," - f" but is {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 - - -def qwen2moe_attention_forward_origin( - self, - hidden_states: torch.Tensor, - attention_mask: Optional[torch.Tensor] = None, - position_ids: Optional[torch.LongTensor] = None, - past_key_value: Optional[Tuple[torch.Tensor]] = None, - output_attentions: bool = False, - use_cache: bool = False, - **kwargs, -) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]: - use_fuse_rope = should_use_fuse_rope(hidden_states, position_ids, self.training) - - if "padding_mask" in kwargs: - warnings.warn( - "Passing `padding_mask` is deprecated and will be removed in v4.37. " - "Please make sure use `attention_mask` instead.`" - ) - bsz, q_len, _ = hidden_states.size() - device = hidden_states.device - - enough_kv_room = is_enough_kv_cache_room_4_36(past_key_value, self.layer_idx) - - qtype_check = decoding_fast_path_qtype_check(self.q_proj) - decoding_fast_path = (qtype_check and use_fuse_rope - and enough_kv_room and bsz * q_len == 1) - decoding_fast_path = decoding_fast_path and not self.q_proj.enable_xetla - if decoding_fast_path: - hidden_states = hidden_states.view(1, -1) - cache_k = past_key_value.key_cache[self.layer_idx] - cache_v = past_key_value.value_cache[self.layer_idx] - kv_seq_len = cache_k.shape[-2] - import xe_linear - args = [hidden_states, self.q_proj.weight, self.k_proj.weight, self.v_proj.weight, - self.q_proj.bias, self.k_proj.bias, self.v_proj.bias, position_ids, cache_k, - cache_v, self.q_proj.weight.qtype, self.v_proj.weight.qtype, kv_seq_len, - self.head_dim, self.rotary_emb.base] - query_states, key_states, value_states = xe_linear.forward_qkv_bias(*args) - kv_seq_len += 1 - if self.layer_idx == 0: - past_key_value._seen_tokens = kv_seq_len - past_key_value.key_cache[self.layer_idx] = key_states - past_key_value.value_cache[self.layer_idx] = value_states - else: - 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: - invalidInputError( - False, - "The cache structure has changed since version v4.36. " - f"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) - if use_fuse_rope: - query_states, key_states = apply_rotary_pos_emb_cache_freq_xpu(query_states, key_states, - sin, cos, "qwen2_moe", - position_ids) - else: - query_states, key_states = apply_rotary_pos_emb(query_states, key_states, - cos, sin, position_ids) - if past_key_value is not None: - if self.layer_idx == 0: - past_key_value._seen_tokens += key_states.shape[-2] - - if len(past_key_value.key_cache) <= self.layer_idx: - past_key_value.key_cache.append(key_states) - past_key_value.value_cache.append(value_states) - else: - cache_k = past_key_value.key_cache[self.layer_idx] - cache_v = past_key_value.value_cache[self.layer_idx] - - if not enough_kv_room: - # allocate new - new_c_k, new_c_v = extend_kv_cache(bsz, - self.num_key_value_heads, # Support GQA - self.head_dim, - cache_k.size(2), - kv_seq_len + KV_CACHE_ALLOC_BLOCK_LENGTH, - dtype=cache_k.dtype, - device=device) - - new_c_k[:] = cache_k - new_c_v[:] = cache_v - cache_k = new_c_k - cache_v = new_c_v - - key_states, value_states = append_kv_cache(cache_k, - cache_v, - key_states, - value_states) - - # update past_key_value - past_key_value.key_cache[self.layer_idx] = key_states - past_key_value.value_cache[self.layer_idx] = value_states - # 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) - - if not self.training and not hidden_states.requires_grad and \ - use_flash_attention(query_states, key_states, attention_mask): - attn_output = F.scaled_dot_product_attention(query_states.to(device, dtype=torch.float16), - key_states.to(device, dtype=torch.float16), - value_states.to(device, dtype=torch.float16), - is_causal=True) - attn_weights = None - else: - attn_weights = torch.matmul(query_states, - key_states.transpose(2, 3)) / math.sqrt(self.head_dim) - - invalidInputError(attn_weights.size() == (bsz, self.num_heads, q_len, kv_seq_len), - ("Attention weights should be of size " - f"{(bsz, self.num_heads, q_len, kv_seq_len)}," - "but is {attn_weights.size()}")) - - if attention_mask is not None: - invalidInputError(attention_mask.size() == (bsz, 1, q_len, kv_seq_len), - (f"Attention mask should be of size {(bsz, 1, q_len, kv_seq_len)}" - f" but is {attention_mask.size()}")) - - attn_weights = attn_weights + attention_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) - - invalidInputError(attn_output.size() == (bsz, self.num_heads, q_len, self.head_dim), - "`attn_output` should be of size " - f"{(bsz, self.num_heads, q_len, self.head_dim)}," - f" but is {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.to(hidden_states.dtype), attn_weights, past_key_value - - -def qwen2moe_attention_forward_sdpa( - self, - hidden_states: torch.Tensor, - attention_mask: Optional[torch.Tensor] = None, - position_ids: Optional[torch.LongTensor] = None, - past_key_value: Optional[Tuple[torch.Tensor]] = None, - output_attentions: bool = False, - use_cache: bool = False, - **kwargs, -) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]: - use_fuse_rope = should_use_fuse_rope(hidden_states, position_ids, self.training) - - if "padding_mask" in kwargs: - warnings.warn( - "Passing `padding_mask` is deprecated and will be removed in v4.37. " - "Please make sure use `attention_mask` instead.`" - ) - bsz, q_len, _ = hidden_states.size() - device = hidden_states.device - - enough_kv_room = is_enough_kv_cache_room_4_36(past_key_value, self.layer_idx) - - qtype_check = decoding_fast_path_qtype_check(self.q_proj) - decoding_fast_path = (qtype_check and use_fuse_rope - and enough_kv_room and bsz * q_len == 1) - decoding_fast_path = decoding_fast_path and not self.q_proj.enable_xetla - if decoding_fast_path: - hidden_states = hidden_states.view(1, -1) - cache_k = past_key_value.key_cache[self.layer_idx] - cache_v = past_key_value.value_cache[self.layer_idx] - kv_seq_len = cache_k.shape[-2] - import xe_linear - args = [hidden_states, self.q_proj.weight, self.k_proj.weight, self.v_proj.weight, - self.q_proj.bias, self.k_proj.bias, self.v_proj.bias, position_ids, cache_k, - cache_v, self.q_proj.weight.qtype, self.v_proj.weight.qtype, kv_seq_len, - self.head_dim, self.rotary_emb.base] - query_states, key_states, value_states = xe_linear.forward_qkv_bias(*args) - kv_seq_len += 1 - if self.layer_idx == 0: - past_key_value._seen_tokens = kv_seq_len - past_key_value.key_cache[self.layer_idx] = key_states - past_key_value.value_cache[self.layer_idx] = value_states - else: - 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: - invalidInputError( - False, - "The cache structure has changed since version v4.36. " - f"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) - if use_fuse_rope: - query_states, key_states = apply_rotary_pos_emb_cache_freq_xpu(query_states, key_states, - sin, cos, "qwen2_moe", - position_ids) - else: - query_states, key_states = apply_rotary_pos_emb(query_states, key_states, - cos, sin, position_ids) - if past_key_value is not None: - if self.layer_idx == 0: - past_key_value._seen_tokens += key_states.shape[-2] - - if len(past_key_value.key_cache) <= self.layer_idx: - past_key_value.key_cache.append(key_states) - past_key_value.value_cache.append(value_states) - else: - cache_k = past_key_value.key_cache[self.layer_idx] - cache_v = past_key_value.value_cache[self.layer_idx] - - if not enough_kv_room: - # allocate new - new_c_k, new_c_v = extend_kv_cache(bsz, - self.num_key_value_heads, # Support GQA - self.head_dim, - cache_k.size(2), - kv_seq_len + KV_CACHE_ALLOC_BLOCK_LENGTH, - dtype=cache_k.dtype, - device=device) - - new_c_k[:] = cache_k - new_c_v[:] = cache_v - cache_k = new_c_k - cache_v = new_c_v - - key_states, value_states = append_kv_cache(cache_k, - cache_v, - key_states, - value_states) - - # update past_key_value - past_key_value.key_cache[self.layer_idx] = key_states - past_key_value.value_cache[self.layer_idx] = value_states - # 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) - - if output_attentions: - attn_weights = torch.matmul(query_states, - key_states.transpose(2, 3)) / math.sqrt(self.head_dim) - - invalidInputError(attn_weights.size() == (bsz, self.num_heads, q_len, kv_seq_len), - ("Attention weights should be of size " - f"{(bsz, self.num_heads, q_len, kv_seq_len)}," - "but is {attn_weights.size()}")) - - if attention_mask is not None: - invalidInputError(attention_mask.size() == (bsz, 1, q_len, kv_seq_len), - (f"Attention mask should be of size {(bsz, 1, q_len, kv_seq_len)}" - f" but is {attention_mask.size()}")) - - attn_weights = attn_weights + attention_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) - else: - attn_weights = None - - from torch.nn.functional import scaled_dot_product_attention as sdpa - attn_output = sdpa(query_states, - key_states, - value_states, - attn_mask=attention_mask, - dropout_p=self.attention_dropout if self.training else 0.0, - is_causal=self.is_causal and attention_mask is None and q_len > 1) - - invalidInputError(attn_output.size() == (bsz, self.num_heads, q_len, self.head_dim), - "`attn_output` should be of size " - f"{(bsz, self.num_heads, q_len, self.head_dim)}," - f" but is {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) - - return attn_output, attn_weights, past_key_value + del module.q_proj, module.k_proj, module.v_proj def qwen2moe_moeblock_forward(self, hidden_states: torch.Tensor):