rebase

Signed-off-by: jiqing-feng <jiqing.feng@intel.com>

Author: jiqing-feng

optimum/exporters/ipex/cache_utils.py

@@ -5,6 +5,10 @@
 from transformers import Cache, PretrainedConfig
 
 
+# May need to tune based on sequence length and different models but default to 16 currently.
+BLOCK_SIZE = 16
+
+
 class IPEXPagedCache(Cache):
     """
     A PagedCache that grows dynamically as more tokens are generated. everytime it grows block-size memory, vendor could set the pageCache memory layout.
@@ -44,7 +48,7 @@ def __init__(
         self.batch_size = batch_size
         # Used in `generate` to keep tally of how many tokens the cache has seen
         self._seen_tokens = torch.zeros([batch_size], dtype=torch.int32, device=device)
-        self.block_size = 64
+        self.block_size = BLOCK_SIZE
         self.num_blocks = (max_cache_len // self.block_size + (max_cache_len % self.block_size != 0)) * batch_size
         self.block_tables = -1 * torch.ones([self.num_blocks], dtype=torch.int32, device=device).reshape(
             batch_size, -1

optimum/exporters/ipex/model_patcher.py

@@ -15,6 +15,7 @@
 from transformers.models.bert.modeling_bert import BertIntermediate
 from transformers.models.falcon.modeling_falcon import FalconDecoderLayer, FalconModel
 from transformers.models.gpt2.modeling_gpt2 import GPT2MLP, GPT2Attention, GPT2Block, GPT2Model
+from transformers.models.gpt2.modeling_gpt2 import GPT2MLP, GPT2Attention, GPT2Block, GPT2Model
 from transformers.models.llama.modeling_llama import (
     LlamaDecoderLayer,
     LlamaModel,
@@ -33,6 +34,7 @@
 from .modeling_utils import (
     _IPEX_MINIMUM_VERSION_FOR_PATCHING,
     _IPEXGPT2MLP,
+    _IPEXGPT2MLP,
     _falcon_model_forward,
     _gpt2_block_forward,
     _gpt2_model_forward,

optimum/exporters/ipex/modeling_utils.py

@@ -769,19 +769,17 @@ def attention_interface(
                 is_causal=True,
             )
             self.use_sdpa = True
-        elif self.has_flash_attn(query):
+        elif self.has_flash_attn(query) and past_len == 0:
             attn_output = torch.empty_like(query)
             seq_len_tensor = torch.cat((input_lens.new_tensor([0]), input_lens.cumsum(-1).int()))
-            query_len_tensor = seq_len_tensor if past_len == 0 else torch.arange(seq_len_tensor.shape[0]).int()
-            query_max_len = input_lens.max() if past_len == 0 else 1
             PagedAttention.flash_attn_varlen_func(
                 attn_output,
                 query.contiguous() if query.device.type == "xpu" else query,
                 key_cache.contiguous() if key_cache.device.type == "xpu" else key_cache,
                 value_cache.contiguous() if value_cache.device.type == "xpu" else value_cache,
-                query_len_tensor,
                 seq_len_tensor,
-                query_max_len,
+                seq_len_tensor,
+                input_lens.max(),
                 input_lens.max(),
                 1.0 / math.sqrt(self.head_dim),
                 True,

optimum/exporters/openvino/convert.py

@@ -45,6 +45,7 @@
     compare_versions,
     is_diffusers_version,
     is_openvino_tokenizers_version,
+    is_openvino_version,
     is_tokenizers_version,
     is_transformers_version,
 )
@@ -366,6 +367,7 @@ def export_pytorch(
     import torch
     from torch.utils._pytree import tree_map
 
+    from openvino.frontend.pytorch.ts_decoder import TorchScriptPythonDecoder
     from optimum.exporters.utils import check_dummy_inputs_are_allowed
 
     logger.info(f"Using framework PyTorch: {torch.__version__}")
@@ -428,15 +430,20 @@ def ts_patched_forward(*args, **kwargs):
 
             patcher.patched_forward = ts_patched_forward
 
+            ts_decoder_kwargs = {}
+            if library_name == "diffusers" and is_openvino_version(">=", "2025.0"):
+                ts_decoder_kwargs["trace_kwargs"] = {"check_trace": False}
+
             with patcher:
                 if patch_16bit_model:
                     from openvino.frontend.pytorch.patch_model import __make_16bit_traceable
 
                     __make_16bit_traceable(model)
                 check_dummy_inputs_are_allowed(model, dummy_inputs)
                 input_info = _get_input_info(model, config, dummy_inputs)
+                ts_decoder = TorchScriptPythonDecoder(model, example_input=dummy_inputs, **ts_decoder_kwargs)
                 ov_model = convert_model(
-                    model,
+                    ts_decoder,
                     example_input=dummy_inputs,
                     input=[(item.shape, item.type) for item in input_info],
                 )