add phi-2 optimization (#10843)

python/llm/src/ipex_llm/transformers/convert.py

@@ -650,6 +650,9 @@ def _optimize_pre(model):
     if model.config.model_type == "starcoder2":
         from ipex_llm.transformers.models.starcoder2 import merge_qkv
         model.apply(merge_qkv)
+    if model.config.model_type == "phi":
+        from ipex_llm.transformers.models.phi import merge_qkv
+        model.apply(merge_qkv)
     if model.config.model_type == "qwen":
         rope_base = model.config.rotary_emb_base
         from accelerate.big_modeling import init_empty_weights
@@ -1415,7 +1418,14 @@ def _optimize_post(model, lightweight_bmm=False):
         from ipex_llm.transformers.models.starcoder2 import model_forward
         convert_forward(model, module.Starcoder2Attention, attention_forward)
         convert_forward(model, module.Starcoder2Model, model_forward)
-
+    elif model.config.model_type == 'phi':
+        # for phi-2
+        modeling_module_name = model.__class__.__module__
+        module = importlib.import_module(modeling_module_name)
+        from ipex_llm.transformers.models.phi import attention_forward
+        from ipex_llm.transformers.models.phi import model_forward
+        convert_forward(model, module.PhiAttention, attention_forward)
+        convert_forward(model, module.PhiModel, model_forward)
     elif model.config.model_type == 'yuan':
         modeling_module_name = model.__class__.__module__
         module = importlib.import_module(modeling_module_name)

python/llm/src/ipex_llm/transformers/models/phi.py

@@ -0,0 +1,192 @@
+#
+# Copyright 2016 The BigDL Authors.
+#
+# 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.
+#
+# Some parts of this file is adapted from
+# https://github.com/huggingface/transformers/blob/v4.37.0/src/transformers/models/phi/modeling_phi.py
+# which is licensed under Apache License 2.0:
+#
+# Copyright 2023 Microsoft and the HuggingFace Inc. team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import math
+import torch
+
+from ipex_llm.transformers.models.utils import apply_rotary_pos_emb_cache_freq_xpu
+from ipex_llm.transformers.kv import DynamicNormalCache
+from ipex_llm.utils.common.log4Error import invalidInputError
+
+from typing import Optional, Tuple, List
+from transformers.cache_utils import Cache
+from transformers.models.phi.modeling_phi import repeat_kv, apply_rotary_pos_emb
+from transformers.models.phi.modeling_phi import PhiModel
+
+
+def should_use_fuse_rope(self, hidden_states, position_ids):
+    use_fuse_rope = (
+        hidden_states.device.type == "xpu" and
+        hidden_states.numel() == hidden_states.size(-1) and
+        not (self.training and hidden_states.requires_grad) and
+        position_ids is not None
+    )
+    return use_fuse_rope
+
+
+def merge_qkv(module: torch.nn.Module):
+    if module.__class__.__name__ == "PhiAttention":
+        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
+
+        del module.q_proj, module.k_proj, module.v_proj
+
+
+def attention_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,
+) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
+    bsz, q_len, _ = hidden_states.size()
+
+    invalidInputError(not self.qk_layernorm, "`qk_layernorm` must be false")
+
+    qkv = self.qkv_proj(hidden_states)
+    qkv = qkv.view(bsz, q_len, self.num_heads + 2 * self.num_key_value_heads, self.head_dim)
+    qkv = qkv.transpose(1, 2)
+    query_states, key_states, value_states = qkv.split([self.num_heads,
+                                                        self.num_key_value_heads,
+                                                        self.num_key_value_heads], dim=1)
+
+    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)
+
+    # Partial rotary embedding
+    query_rot, query_pass = (
+        query_states[..., : self.rotary_emb.dim],
+        query_states[..., self.rotary_emb.dim:],
+    )
+    key_rot, key_pass = (
+        key_states[..., : self.rotary_emb.dim],
+        key_states[..., self.rotary_emb.dim:],
+    )
+
+    # IPEX-LLM OPT: fuse rope
+    use_fuse_rope = should_use_fuse_rope(self, hidden_states, position_ids)
+
+    # [batch_size, seq_length, num_heads, head_dim // config.partial_rotary_factor]
+    if use_fuse_rope:
+        query_rot, key_rot = apply_rotary_pos_emb_cache_freq_xpu(query_rot, key_rot, sin,
+                                                                 cos, "stablelm", position_ids)
+    else:
+        query_rot, key_rot = apply_rotary_pos_emb(query_rot, key_rot, cos, sin, position_ids)
+
+    # [batch_size, seq_length, num_heads, head_dim]
+    query_states = torch.cat((query_rot, query_pass), dim=-1)
+    key_states = torch.cat((key_rot, key_pass), dim=-1)
+
+    invalidInputError(past_key_value is not None,
+                      "`past_key_value` cannot be None")
+    key_states, value_states = past_key_value.update(key_states, value_states,
+                                                     self.layer_idx, None, new_layout=True)
+
+    key_states = repeat_kv(key_states, self.num_key_value_groups)
+    value_states = repeat_kv(value_states, self.num_key_value_groups)
+
+    # Queries and keys upcast to fp32 is required by Phi-2 to avoid overflow
+    attn_weights = torch.matmul(
+        query_states.to(torch.float32), key_states.to(torch.float32).transpose(2, 3)
+    ) / math.sqrt(self.head_dim)
+
+    if attention_mask is not None:
+        attn_weights = attn_weights + attention_mask
+
+    # upcast attention to fp32
+    attn_weights = torch.nn.functional.softmax(attn_weights, dim=-1,
+                                               dtype=torch.float32).to(value_states.dtype)
+    attn_weights = torch.nn.functional.dropout(attn_weights, p=self.attention_dropout,
+                                               training=self.training)
+
+    attn_output = torch.matmul(attn_weights, value_states)
+
+    attn_output = attn_output.transpose(1, 2).contiguous()
+    attn_output = attn_output.reshape(bsz, q_len, self.hidden_size)
+
+    attn_output = self.dense(attn_output)
+
+    if not output_attentions:
+        attn_weights = None
+
+    return attn_output, attn_weights, past_key_value
+
+
+def model_forward(
+    self,
+    input_ids: torch.LongTensor = None,
+    attention_mask: Optional[torch.Tensor] = None,
+    position_ids: Optional[torch.LongTensor] = None,
+    past_key_values: Optional[List[torch.FloatTensor]] = None,
+    inputs_embeds: Optional[torch.FloatTensor] = None,
+    use_cache: Optional[bool] = None,
+    output_attentions: Optional[bool] = None,
+    output_hidden_states: Optional[bool] = None,
+    return_dict: Optional[bool] = None,
+):
+    # IPEX-LLM OPT: kv cache but no sdp (its head_dim 80, cannot use sdp)
+    use_cache = use_cache if use_cache is not None else self.config.use_cache
+    if use_cache:
+        if not isinstance(past_key_values, DynamicNormalCache):
+            past_key_values = DynamicNormalCache.from_legacy_cache(past_key_values)
+    return PhiModel.forward(
+        self=self,
+        input_ids=input_ids,
+        attention_mask=attention_mask,
+        position_ids=position_ids,
+        past_key_values=past_key_values,
+        inputs_embeds=inputs_embeds,
+        use_cache=use_cache,
+        output_attentions=output_attentions,
+        output_hidden_states=output_hidden_states,
+        return_dict=return_dict,
+    )