Remove changes that should no be propagatted just yet

huggingface · Feb 15, 2024 · c9f3c82 · c9f3c82
1 parent 80148ab
commit c9f3c82
Show file tree

Hide file tree

Showing 3 changed files with 0 additions and 15 deletions.
diff --git a/src/transformers/models/persimmon/modeling_persimmon.py b/src/transformers/models/persimmon/modeling_persimmon.py
@@ -863,11 +863,6 @@ def prepare_inputs_for_generation(
             if past_key_values:
                 position_ids = position_ids[:, -input_ids.shape[1] :]
 
-        if past_key_value := getattr(self.model.layers[0].self_attn, "past_key_value", None):
-            # generation with static cache
-            seen_tokens = past_key_value.get_seq_length()
-            input_ids = input_ids[:, seen_tokens:]
-            position_ids = position_ids[:, seen_tokens:]
 
         # if `inputs_embeds` are passed, we only want to use them in the 1st generation step
         if inputs_embeds is not None and past_key_values is None:

diff --git a/src/transformers/models/phi/modeling_phi.py b/src/transformers/models/phi/modeling_phi.py
@@ -1125,11 +1125,6 @@ def prepare_inputs_for_generation(
             if past_key_values:
                 position_ids = position_ids[:, -input_ids.shape[1] :]
 
-        if past_key_value := getattr(self.model.layers[0].self_attn, "past_key_value", None):
-            # generation with static cache
-            seen_tokens = past_key_value.get_seq_length()
-            input_ids = input_ids[:, seen_tokens:]
-            position_ids = position_ids[:, seen_tokens:]
 
         # if `inputs_embeds` are passed, we only want to use them in the 1st generation step
         if inputs_embeds is not None and past_key_values is None:

diff --git a/src/transformers/models/stablelm/modeling_stablelm.py b/src/transformers/models/stablelm/modeling_stablelm.py
@@ -1088,11 +1088,6 @@ def prepare_inputs_for_generation(
             if past_key_values:
                 position_ids = position_ids[:, -input_ids.shape[1] :]
 
-        if past_key_value := getattr(self.model.layers[0].self_attn, "past_key_value", None):
-            # generation with static cache
-            seen_tokens = past_key_value.get_seq_length()
-            input_ids = input_ids[:, seen_tokens:]
-            position_ids = position_ids[:, seen_tokens:]
 
         # if `inputs_embeds` are passed, we only want to use them in the 1st generation step
         if inputs_embeds is not None and past_key_values is None: