Llama et al. / FSDP : Fix breaking change in 4.40 for FSDP (huggingfa…

…ce#31161) * fix llama fsdp * fixup * adding FSDP tests for CPU offloading * fixes * fix tests * fix tests * add it for mixtral * propagate the changes on other models * Update src/transformers/models/phi/modeling_phi.py * Delete utils/testing_scripts/fsdp_cpu_offloading.py Remove script - FSDP + CPU offloading it tested in the test suite * Delete utils/testing_scripts/dummy_fsdp_config.yml * Update + add cache_positions docstring --------- Co-authored-by: amyeroberts <[email protected]>
SangbumChoi · Jun 26, 2024 · 3f93fd0 · 3f93fd0
1 parent ac52084
commit 3f93fd0
Show file tree

Hide file tree

Showing 13 changed files with 71 additions and 1 deletion.
diff --git a/src/transformers/models/falcon/modeling_falcon.py b/src/transformers/models/falcon/modeling_falcon.py
@@ -782,6 +782,7 @@ def forward(
         head_mask: Optional[torch.Tensor] = None,
         use_cache: bool = False,
         output_attentions: bool = False,
+        **kwargs,
     ):
         residual = hidden_states
 

diff --git a/src/transformers/models/gemma/modeling_gemma.py b/src/transformers/models/gemma/modeling_gemma.py
@@ -628,6 +628,7 @@ def forward(
         output_attentions: Optional[bool] = False,
         use_cache: Optional[bool] = False,
         cache_position: Optional[torch.LongTensor] = None,
+        **kwargs,
     ) -> Tuple[torch.FloatTensor, Optional[Tuple[torch.FloatTensor, torch.FloatTensor]]]:
         """
         Args:
@@ -642,6 +643,11 @@ def forward(
                 If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding
                 (see `past_key_values`).
             past_key_value (`Tuple(torch.FloatTensor)`, *optional*): cached past key and value projection states
+            cache_position (`torch.LongTensor` of shape `(sequence_length)`, *optional*):
+                Indices depicting the position of the input sequence tokens in the sequence
+            kwargs (`dict`, *optional*):
+                Arbitrary kwargs to be ignored, used for FSDP and other methods that injects code
+                into the model
         """
         residual = hidden_states
 

diff --git a/src/transformers/models/gpt_bigcode/modeling_gpt_bigcode.py b/src/transformers/models/gpt_bigcode/modeling_gpt_bigcode.py
@@ -706,6 +706,7 @@ def forward(
         encoder_attention_mask: Optional[torch.Tensor] = None,
         use_cache: Optional[bool] = False,
         output_attentions: Optional[bool] = False,
+        **kwargs,
     ) -> Union[
         Tuple[torch.Tensor], Tuple[torch.Tensor, torch.Tensor], Tuple[torch.Tensor, torch.Tensor, torch.Tensor]
     ]:

diff --git a/src/transformers/models/llama/modeling_llama.py b/src/transformers/models/llama/modeling_llama.py
@@ -301,6 +301,7 @@ def forward(
         output_attentions: bool = False,
         use_cache: bool = False,
         cache_position: Optional[torch.LongTensor] = None,
+        **kwargs,
     ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
         bsz, q_len, _ = hidden_states.size()
 
@@ -590,6 +591,7 @@ def forward(
         output_attentions: bool = False,
         use_cache: bool = False,
         cache_position: Optional[torch.LongTensor] = None,
+        **kwargs,
     ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
         if output_attentions:
             # TODO: Improve this warning with e.g. `model.config.attn_implementation = "manual"` once this is implemented.
@@ -687,6 +689,7 @@ def forward(
         output_attentions: Optional[bool] = False,
         use_cache: Optional[bool] = False,
         cache_position: Optional[torch.LongTensor] = None,
+        **kwargs,
     ) -> Tuple[torch.FloatTensor, Optional[Tuple[torch.FloatTensor, torch.FloatTensor]]]:
         """
         Args:
@@ -701,6 +704,11 @@ def forward(
                 If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding
                 (see `past_key_values`).
             past_key_value (`Tuple(torch.FloatTensor)`, *optional*): cached past key and value projection states
+            cache_position (`torch.LongTensor` of shape `(sequence_length)`, *optional*):
+                Indices depicting the position of the input sequence tokens in the sequence
+            kwargs (`dict`, *optional*):
+                Arbitrary kwargs to be ignored, used for FSDP and other methods that injects code
+                into the model
         """
         residual = hidden_states
 

diff --git a/src/transformers/models/mistral/modeling_mistral.py b/src/transformers/models/mistral/modeling_mistral.py
@@ -591,6 +591,7 @@ def forward(
         output_attentions: bool = False,
         use_cache: bool = False,
         cache_position: Optional[torch.LongTensor] = None,
+        **kwargs,
     ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
         if output_attentions:
             # TODO: Improve this warning with e.g. `model.config.attn_implementation = "manual"` once this is implemented.
@@ -689,6 +690,7 @@ def forward(
         output_attentions: Optional[bool] = False,
         use_cache: Optional[bool] = False,
         cache_position: Optional[torch.LongTensor] = None,
+        **kwargs,
     ) -> Tuple[torch.FloatTensor, Optional[Tuple[torch.FloatTensor, torch.FloatTensor]]]:
         """
         Args:
@@ -703,8 +705,12 @@ def forward(
                 If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding
                 (see `past_key_values`).
             past_key_value (`Tuple(torch.FloatTensor)`, *optional*): cached past key and value projection states
+            cache_position (`torch.LongTensor` of shape `(sequence_length)`, *optional*):
+                Indices depicting the position of the input sequence tokens in the sequence
+            kwargs (`dict`, *optional*):
+                Arbitrary kwargs to be ignored, used for FSDP and other methods that injects code
+                into the model
         """
-
         residual = hidden_states
 
         hidden_states = self.input_layernorm(hidden_states)

diff --git a/src/transformers/models/mixtral/modeling_mixtral.py b/src/transformers/models/mixtral/modeling_mixtral.py
@@ -888,6 +888,7 @@ def forward(
         output_router_logits: Optional[bool] = False,
         use_cache: Optional[bool] = False,
         cache_position: Optional[torch.LongTensor] = None,
+        **kwargs,
     ) -> Tuple[torch.FloatTensor, Optional[Tuple[torch.FloatTensor, torch.FloatTensor]]]:
         """
         Args:
@@ -906,6 +907,9 @@ def forward(
                 (see `past_key_values`).
             cache_position (`torch.LongTensor` of shape `(sequence_length)`, *optional*):
                 Indices depicting the position of the input sequence tokens in the sequence.
+            kwargs (`dict`, *optional*):
+                Arbitrary kwargs to be ignored, used for FSDP and other methods that injects code
+                into the model
         """
 
         residual = hidden_states

diff --git a/src/transformers/models/olmo/modeling_olmo.py b/src/transformers/models/olmo/modeling_olmo.py
@@ -666,6 +666,7 @@ def forward(
         output_attentions: Optional[bool] = False,
         use_cache: Optional[bool] = False,
         cache_position: Optional[torch.LongTensor] = None,
+        **kwargs,
     ) -> Tuple[torch.FloatTensor, Optional[Tuple[torch.FloatTensor, torch.FloatTensor]]]:
         """
         Args:
@@ -680,6 +681,11 @@ def forward(
                 If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding
                 (see `past_key_values`).
             past_key_value (`Tuple(torch.FloatTensor)`, *optional*): cached past key and value projection states
+            cache_position (`torch.LongTensor` of shape `(sequence_length)`, *optional*):
+                Indices depicting the position of the input sequence tokens in the sequence
+            kwargs (`dict`, *optional*):
+                Arbitrary kwargs to be ignored, used for FSDP and other methods that injects code
+                into the model
         """
         residual = hidden_states
 

diff --git a/src/transformers/models/phi/modeling_phi.py b/src/transformers/models/phi/modeling_phi.py
@@ -771,6 +771,7 @@ def forward(
         use_cache: Optional[bool] = False,
         past_key_value: Optional[Tuple[torch.Tensor]] = None,
         cache_position: Optional[torch.LongTensor] = None,
+        **kwargs,
     ) -> Tuple[torch.FloatTensor, Optional[Tuple[torch.FloatTensor, torch.FloatTensor]]]:
         """
         Args:
@@ -790,6 +791,9 @@ def forward(
             past_key_value (`Tuple(torch.FloatTensor)`, *optional*): cached past key and value projection states
             cache_position (`torch.LongTensor` of shape `(sequence_length)`, *optional*):
                 Indices depicting the position of the input sequence tokens in the sequence
+            kwargs (`dict`, *optional*):
+                Arbitrary kwargs to be ignored, used for FSDP and other methods that injects code
+                into the model
         """
 
         residual = hidden_states

diff --git a/src/transformers/models/phi3/modeling_phi3.py b/src/transformers/models/phi3/modeling_phi3.py
@@ -830,6 +830,7 @@ def forward(
         output_attentions: Optional[bool] = False,
         use_cache: Optional[bool] = False,
         cache_position: Optional[torch.LongTensor] = None,
+        **kwargs,
     ) -> Tuple[torch.FloatTensor, Optional[Tuple[torch.FloatTensor, torch.FloatTensor]]]:
         """
         Args:
@@ -847,6 +848,11 @@ def forward(
                 If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding
                 (see `past_key_values`).
             past_key_value (`Tuple(torch.FloatTensor)`, *optional*): cached past key and value projection states
+            cache_position (`torch.LongTensor` of shape `(sequence_length)`, *optional*):
+                Indices depicting the position of the input sequence tokens in the sequence
+            kwargs (`dict`, *optional*):
+                Arbitrary kwargs to be ignored, used for FSDP and other methods that injects code
+                into the model
         """
 
         residual = hidden_states

diff --git a/src/transformers/models/qwen2/modeling_qwen2.py b/src/transformers/models/qwen2/modeling_qwen2.py
@@ -734,6 +734,7 @@ def forward(
         output_attentions: Optional[bool] = False,
         use_cache: Optional[bool] = False,
         cache_position: Optional[torch.LongTensor] = None,
+        **kwargs,
     ) -> Tuple[torch.FloatTensor, Optional[Tuple[torch.FloatTensor, torch.FloatTensor]]]:
         """
         Args:
@@ -749,6 +750,9 @@ def forward(
             past_key_value (`Tuple(torch.FloatTensor)`, *optional*): cached past key and value projection states
             cache_position (`torch.LongTensor` of shape `(sequence_length)`, *optional*):
                 Indices depicting the position of the input sequence tokens in the sequence.
+            kwargs (`dict`, *optional*):
+                Arbitrary kwargs to be ignored, used for FSDP and other methods that injects code
+                into the model
         """
 
         residual = hidden_states

diff --git a/src/transformers/models/qwen2_moe/modeling_qwen2_moe.py b/src/transformers/models/qwen2_moe/modeling_qwen2_moe.py
@@ -876,6 +876,7 @@ def forward(
         output_router_logits: Optional[bool] = False,
         use_cache: Optional[bool] = False,
         cache_position: Optional[torch.LongTensor] = None,
+        **kwargs,
     ) -> Tuple[torch.FloatTensor, Optional[Tuple[torch.FloatTensor, torch.FloatTensor]]]:
         """
         Args:
@@ -894,6 +895,9 @@ def forward(
             past_key_value (`Tuple(torch.FloatTensor)`, *optional*): cached past key and value projection states
             cache_position (`torch.LongTensor` of shape `(sequence_length)`, *optional*):
                 Indices depicting the position of the input sequence tokens in the sequence.
+            kwargs (`dict`, *optional*):
+                Arbitrary kwargs to be ignored, used for FSDP and other methods that injects code
+                into the model
         """
 
         residual = hidden_states

diff --git a/src/transformers/models/starcoder2/modeling_starcoder2.py b/src/transformers/models/starcoder2/modeling_starcoder2.py
@@ -713,6 +713,7 @@ def forward(
         output_attentions: Optional[bool] = False,
         use_cache: Optional[bool] = False,
         cache_position: Optional[torch.LongTensor] = None,
+        **kwargs,
     ) -> Tuple[torch.FloatTensor, Optional[Tuple[torch.FloatTensor, torch.FloatTensor]]]:
         """
         Args:
@@ -728,6 +729,9 @@ def forward(
             past_key_value (`Tuple(torch.FloatTensor)`, *optional*): cached past key and value projection states
             cache_position (`torch.LongTensor` of shape `(sequence_length)`, *optional*):
                 Indices depicting the position of the input sequence tokens in the sequence.
+            kwargs (`dict`, *optional*):
+                Arbitrary kwargs to be ignored, used for FSDP and other methods that injects code
+                into the model
         """
 
         residual = hidden_states

diff --git a/tests/fsdp/test_fsdp.py b/tests/fsdp/test_fsdp.py
@@ -14,6 +14,7 @@
 
 import itertools
 import os
+import subprocess
 import unittest
 from copy import deepcopy
 from functools import partial
@@ -31,6 +32,7 @@
     require_accelerate,
     require_fsdp,
     require_torch_accelerator,
+    require_torch_gpu,
     require_torch_multi_accelerator,
     slow,
     torch_device,
@@ -276,6 +278,20 @@ def test_training_and_can_resume_normally(self, state_dict_type):
             if "learning_rate" in log:
                 self.assertAlmostEqual(log["learning_rate"], log1["learning_rate"], delta=1e-5)
 
+    @require_torch_multi_accelerator
+    @slow
+    @require_torch_gpu
+    @require_fsdp
+    def test_fsdp_cpu_offloading(self):
+        try:
+            subprocess.run(
+                "accelerate launch utils/testing_scripts/fsdp_cpu_offloading.py --config utils/testing_scripts/dummy_fsdp_config.yml",
+                shell=True,
+                check=True,
+            )
+        except:  # noqa
+            raise AssertionError("CPU offloading failed with FSDP!")
+
     def run_cmd_and_get_logs(self, use_accelerate, sharding_strategy, launcher, script, args, output_dir):
         if not use_accelerate:
             fsdp_args = [