Expose Llama Fused OPs control from run_lora_clm.py

examples/language-modeling/run_lora_clm.py

@@ -156,6 +156,14 @@ class ModelArguments:
             )
         },
     )
+    use_fused_rope: bool = field(
+        default=True,
+        metadata={
+            "help": (
+                "Whether to use Habana fused-rope for fine-tuning. The current support is limited to Llama only.",
+            )
+        },
+    )
     load_meta_device: bool = field(
         default=False,
         metadata={
@@ -540,6 +548,8 @@ def main():
         if model_args.use_flash_attention:
             model.generation_config.use_flash_attention = True
             model.generation_config.flash_attention_recompute = model_args.flash_attention_recompute
+        if model_args.use_fused_rope is False:
+            model.generation_config.use_fused_rope = False
 
     if hasattr(model.generation_config, "pad_token_id") and model.generation_config.pad_token_id is not None:
         tokenizer.pad_token_id = model.generation_config.pad_token_id

optimum/habana/transformers/generation/configuration_utils.py

@@ -51,3 +51,4 @@ def __init__(self, **kwargs):
         self.kv_cache_fp8 = kwargs.get("kv_cache_fp8", None)
         self.use_flash_attention = kwargs.get("use_flash_attention", None)
         self.flash_attention_recompute = kwargs.get("flash_attention_recompute", None)
+        self.use_fused_rope = kwargs.get("use_fused_rope", None)

optimum/habana/transformers/generation/utils.py

@@ -683,6 +683,7 @@ def generate(
         # determine whether flash attention needs to be used
         model_kwargs["use_flash_attention"] = generation_config.use_flash_attention
         model_kwargs["flash_attention_recompute"] = True if generation_config.flash_attention_recompute else False
+        model_kwargs["use_fused_rope"] = False if generation_config.use_fused_rope is False else True
 
         if not self.config.is_encoder_decoder:
             calculated_max_length = input_ids.shape[-1]

optimum/habana/transformers/models/llama/modeling_llama.py

@@ -208,6 +208,7 @@ def pre_attn_forward(
         use_flash_attention: Optional[bool] = False,
         flash_attention_recompute: Optional[bool] = False,
         cache_idx: int = None,
+        use_fused_rope: Optional[bool] = True,
         **kwargs,
     ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
         """
@@ -274,7 +275,9 @@ def pre_attn_forward(
                     kv_seq_len = past_key_value[0].shape[-2]
 
         cos, sin = self.rotary_emb(value_states, seq_len=kv_seq_len)
-        query_states, key_states = apply_customized_rope(query_states, key_states, cos, sin, position_ids)
+        query_states, key_states = apply_customized_rope(
+            query_states, key_states, cos, sin, position_ids, use_fused_rope=use_fused_rope
+        )
 
         if past_key_value is not None or reuse_cache:
             # reuse k, v, self_attention
@@ -456,6 +459,7 @@ def forward(
         use_flash_attention: Optional[bool] = False,
         flash_attention_recompute: Optional[bool] = False,
         cache_idx: int = None,
+        use_fused_rope: Optional[bool] = True,
         **kwargs,
     ) -> Tuple[torch.FloatTensor, Optional[Tuple[torch.FloatTensor, torch.FloatTensor]]]:
         """
@@ -486,6 +490,7 @@ def forward(
             use_flash_attention=use_flash_attention,
             flash_attention_recompute=flash_attention_recompute,
             cache_idx=cache_idx,
+            use_fused_rope=use_fused_rope,
             **kwargs,
         )
         self.self_attn.attention_all_reduce(output_pre_attn)
@@ -516,6 +521,7 @@ def pre_attn(
         use_flash_attention: Optional[bool] = False,
         flash_attention_recompute: Optional[bool] = False,
         cache_idx: int = None,
+        use_fused_rope: Optional[bool] = True,
     ) -> Tuple[torch.FloatTensor, Optional[Tuple[torch.FloatTensor, torch.FloatTensor]]]:
         hidden_states = self.input_layernorm(hidden_states)
         output_attn, attn_weights, present_key_value = self.self_attn.pre_attn_forward(
@@ -531,6 +537,7 @@ def pre_attn(
             use_flash_attention,
             flash_attention_recompute,
             cache_idx=cache_idx,
+            use_fused_rope=use_fused_rope,
         )
         return output_attn, attn_weights, present_key_value
 
@@ -580,6 +587,7 @@ def forward(
         use_flash_attention: Optional[bool] = False,
         flash_attention_recompute: Optional[bool] = False,
         cache_idx: int = None,
+        use_fused_rope: Optional[bool] = True,
     ) -> Union[Tuple, BaseModelOutputWithPast]:
         """
         Copied from LlamaModel.forward: https://github.com/huggingface/transformers/blob/main/src/transformers/models/llama/modeling_llama.py
@@ -682,6 +690,7 @@ def forward(
                     False,
                     use_flash_attention,
                     flash_attention_recompute,
+                    use_fused_rope,
                 )
             else:
                 layer_outputs = decoder_layer(
@@ -697,6 +706,7 @@ def forward(
                     use_flash_attention=use_flash_attention,
                     flash_attention_recompute=flash_attention_recompute,
                     cache_idx=cache_idx,
+                    use_fused_rope=use_fused_rope,
                 )
 
             hidden_states = layer_outputs[0]
@@ -771,6 +781,7 @@ def forward(
         use_flash_attention: Optional[bool] = False,
         flash_attention_recompute: Optional[bool] = False,
         cache_idx: int = None,
+        use_fused_rope: Optional[bool] = True,
     ) -> Union[Tuple, CausalLMOutputWithPast]:
         output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
         output_hidden_states = (
@@ -794,6 +805,7 @@ def forward(
             use_flash_attention=use_flash_attention,
             flash_attention_recompute=flash_attention_recompute,
             cache_idx=cache_idx,
+            use_fused_rope=use_fused_rope,
         )
         hidden_states = outputs[0]
         _, seq_len, _ = hidden_states.shape
@@ -911,8 +923,8 @@ def prepare_inputs_for_generation(
         return model_inputs
 
 
-def apply_customized_rope(q, k, cos, sin, position_ids):
-    if q.device.type == "hpu" and FusedRoPE:
+def apply_customized_rope(q, k, cos, sin, position_ids, use_fused_rope=True):
+    if q.device.type == "hpu" and FusedRoPE and use_fused_rope:
         # TODO: remove `.clone()` when it is fixed in SynapseAI
         return FusedRoPE.apply(
             q, cos.unsqueeze(0).unsqueeze(0).clone(), sin.unsqueeze(0).unsqueeze(0).clone(), position_ids

optimum/habana/transformers/trainer.py

@@ -909,6 +909,8 @@ def hpu_deepspeed_checkpointing(function, *checkpoint_args):
                             inputs["use_flash_attention"] = True
                         if self.model.generation_config.flash_attention_recompute:
                             inputs["flash_attention_recompute"] = True
+                        if self.model.generation_config.use_fused_rope is False:
+                            inputs["use_fused_rope"] = False
 
                 # TODO: keep syncs for fast DDP?
                 with self.accelerator.accumulate(model):
@@ -1684,6 +1686,8 @@ def evaluation_loop(
                         inputs["use_flash_attention"] = True
                     if self.model.generation_config.flash_attention_recompute:
                         inputs["flash_attention_recompute"] = True
+                    if self.model.generation_config.use_fused_rope is False:
+                        inputs["use_fused_rope"] = False
 
             # Prediction step
             loss, logits, labels = self.prediction_step(model, inputs, prediction_loss_only, ignore_keys=ignore_keys)