这是indexloc提供的服务，不要输入任何密码
Skip to content

FIX Prefix tuning with Qwen3 issue #2883

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 4 commits into from
Nov 5, 2025
Merged

FIX Prefix tuning with Qwen3 issue #2883

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
befd225
FIX Prefix tuning with Qwen3 issue
BenjaminBossan Oct 31, 2025
a28321d
Fix bug
BenjaminBossan Oct 31, 2025
08d0ca4
Fix for model_config.head_dim = None
BenjaminBossan Oct 31, 2025
7e7b1df
Correct fix...
BenjaminBossan Oct 31, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 0 additions & 3 deletions src/peft/mapping_func.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -25,7 +25,6 @@
from .mixed_model import PeftMixedModel
from .peft_model import PeftModel
from .tuners.tuners_utils import BaseTuner, BaseTunerLayer
from .utils import _prepare_prompt_learning_config


def get_peft_model(
Expand Down Expand Up @@ -120,8 +119,6 @@ def get_peft_model(
low_cpu_mem_usage=low_cpu_mem_usage,
)

if peft_config.is_prompt_learning:
peft_config = _prepare_prompt_learning_config(peft_config, model_config)
Copy link
Member Author

@BenjaminBossan BenjaminBossan Oct 31, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

While debugging this issue, I noticed that _prepare_prompt_learning_config is being called twice, here and in peft_model.py:

peft/src/peft/peft_model.py

Line 1019 in d43b315

peft_config = _prepare_prompt_learning_config(peft_config, dict_config)

This is not a huge deal, as the results should be the same and because the function is quick. Still, let's remove the redundant call here.

return MODEL_TYPE_TO_PEFT_MODEL_MAPPING[peft_config.task_type](
model,
peft_config,
Expand Down
8 changes: 6 additions & 2 deletions src/peft/utils/other.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1105,9 +1105,13 @@ def _prepare_prompt_learning_config(peft_config, model_config):
peft_config.num_attention_heads = num_attention_heads

# For grouped-query attention, see #1901.
if peft_config.peft_type == "PREFIX_TUNING" and "num_key_value_heads" in model_config:
if (peft_config.peft_type == "PREFIX_TUNING") and ("num_key_value_heads" in model_config):
num_key_value_heads = model_config["num_key_value_heads"]
peft_config.token_dim = peft_config.token_dim // peft_config.num_attention_heads * num_key_value_heads
if model_config.get("head_dim", None) is not None:
head_dim = model_config["head_dim"]
else:
head_dim = peft_config.token_dim // peft_config.num_attention_heads
peft_config.token_dim = head_dim * num_key_value_heads
peft_config.num_attention_heads = num_key_value_heads

if getattr(peft_config, "encoder_hidden_size", None) is None:
Expand Down
26 changes: 19 additions & 7 deletions tests/test_decoder_models.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -665,15 +665,27 @@ def test_lora_layer_replication(self):
self._test_prepare_for_training(model_id, LoraConfig, config_kwargs.copy())
self._test_generate(model_id, LoraConfig, config_kwargs.copy())

def test_prompt_learning_with_grouped_query_attention(self):
def test_prefix_tuning_qwen2_with_grouped_query_attention(self):
# See 1901, fixes a bug with handling GQA
model_id = "peft-internal-testing/tiny-dummy-qwen2"
base_model = AutoModelForCausalLM.from_pretrained(model_id)
peft_config = PrefixTuningConfig(num_virtual_tokens=10, task_type="CAUSAL_LM")
model = get_peft_model(base_model, peft_config)
x = torch.tensor([[1, 2, 3]])
# does not raise
model(x)
with hub_online_once(model_id):
base_model = AutoModelForCausalLM.from_pretrained(model_id)
peft_config = PrefixTuningConfig(num_virtual_tokens=10, task_type="CAUSAL_LM")
model = get_peft_model(base_model, peft_config)
x = torch.tensor([[1, 2, 3]])
# does not raise
model(x)

def test_prefix_tuning_qwen3_with_grouped_query_attention(self):
# See 2881, fixes a bug with handling GQA
model_id = "trl-internal-testing/tiny-Qwen3ForCausalLM"
with hub_online_once(model_id):
base_model = AutoModelForCausalLM.from_pretrained(model_id)
peft_config = PrefixTuningConfig(num_virtual_tokens=10, task_type="CAUSAL_LM")
model = get_peft_model(base_model, peft_config)
x = torch.tensor([[1, 2, 3]])
# does not raise
model(x)

def test_prefix_tuning_mistral(self):
# See issue 869, 1962
Expand Down
Loading