[Misc][LoRA] Support Rank Stabilized LoRA (RSLoRA) (vllm-project#6909)

Signed-off-by: Jee Jee Li <[email protected]>
Co-authored-by: Jee Jee Li <[email protected]>

Author: 2 people

tests/lora/test_lora_manager.py

@@ -1,4 +1,5 @@
 import json
+import math
 import os
 from typing import Dict, List
 
@@ -50,6 +51,18 @@ def test_peft_helper(sql_lora_files):
         "embed_tokens",
         "lm_head",
     ]
+    scaling = peft_helper.lora_alpha / peft_helper.r
+    assert abs(peft_helper.vllm_lora_scaling_factor - scaling) < 1e-3
+
+    # test RSLoRA
+    config = dict(r=8,
+                  lora_alpha=16,
+                  target_modules=["gate_proj"],
+                  use_rslora=True)
+    peft_helper = PEFTHelper.from_dict(config)
+
+    scaling = peft_helper.lora_alpha / math.sqrt(peft_helper.r)
+    assert abs(peft_helper.vllm_lora_scaling_factor - scaling) < 1e-3
 
     expected_error = "vLLM only supports modules_to_save being None."
     with pytest.raises(ValueError, match=expected_error):
@@ -60,13 +73,6 @@ def test_peft_helper(sql_lora_files):
             modules_to_save=["lm_head"],
         )
         PEFTHelper.from_dict(config)
-    expected_error = "vLLM does not yet support RSLoRA."
-    with pytest.raises(ValueError, match=expected_error):
-        config = dict(r=8,
-                      lora_alpha=16,
-                      target_modules=["gate_proj"],
-                      use_rslora=True)
-        PEFTHelper.from_dict(config)
 
     expected_error = "vLLM does not yet support DoRA."
     with pytest.raises(ValueError, match=expected_error):

vllm/lora/lora.py

@@ -67,15 +67,9 @@ def from_config(
         peft_helper: PEFTHelper,
         embeddings_tensor: Optional[torch.Tensor] = None,
     ) -> "LoRALayerWeights":
-        return cls(
-            module_name,
-            peft_helper.r,
-            peft_helper.lora_alpha,
-            None,
-            None,
-            None,
-            embeddings_tensor,
-        )
+        return cls(module_name, peft_helper.r, peft_helper.lora_alpha, None,
+                   None, None, embeddings_tensor,
+                   peft_helper.vllm_lora_scaling_factor)
 
     @classmethod
     def create_dummy_lora_weights(

vllm/lora/models.py

@@ -173,7 +173,7 @@ def from_lora_tensors(
         return cls(lora_model_id,
                    peft_helper.r,
                    loras,
-                   scaling_factor=peft_helper.vllm_scaling_factor)
+                   scaling_factor=peft_helper.vllm_long_context_scaling_factor)
 
     @classmethod
     def from_local_checkpoint(

vllm/lora/peft_helper.py

@@ -4,6 +4,8 @@
 from dataclasses import MISSING, dataclass, field, fields
 from typing import Literal, Optional, Union
 
+from vllm.utils import print_info_once
+
 
 @dataclass
 class PEFTHelper:
@@ -14,21 +16,22 @@ class PEFTHelper:
 
     bias: Literal["none", "all", "lora_only"] = field(default="none")
     modules_to_save: Optional[list[str]] = field(default=None)
+    # True to use Rank-Stabilized LoRA (rsLoRA, see: https://arxiv.org/abs/2312.03732)
     use_rslora: bool = field(default=False)
+    # True to use Weight-Decomposed Low-Rank Adaptation (DoRA, see: https://arxiv.org/abs/2402.09353)
     use_dora: bool = field(default=False)
-    # long lora field
+    # long context lora field
     context_length: int = field(default=0)
     # Extra vllm field, start with 'vllm_' to avoid conflict
+    vllm_lora_scaling_factor: float = field(default=1.0)
     vllm_max_position_embeddings: Optional[int] = field(default=False)
-    vllm_scaling_factor: Optional[float] = field(default=None)
+    vllm_long_context_scaling_factor: Optional[float] = field(default=None)
 
     def _validate_features(self):
         error_msg = []
 
         if self.modules_to_save:
             error_msg.append("vLLM only supports modules_to_save being None.")
-        if self.use_rslora:
-            error_msg.append("vLLM does not yet support RSLoRA.")
 
         if self.use_dora:
             error_msg.append("vLLM does not yet support DoRA.")
@@ -38,10 +41,15 @@ def _validate_features(self):
 
     def __post_init__(self):
         self._validate_features()
+        if self.use_rslora:
+            print_info_once("Loading LoRA weights trained with rsLoRA.")
+            self.vllm_lora_scaling_factor = self.lora_alpha / math.sqrt(self.r)
+        else:
+            self.vllm_lora_scaling_factor = self.lora_alpha / self.r
         if self.context_length:
             if self.vllm_max_position_embeddings is None:
                 self.vllm_max_position_embeddings = self.context_length
-            self.vllm_scaling_factor = float(
+            self.vllm_long_context_scaling_factor = float(
                 math.ceil(self.context_length /
                           self.vllm_max_position_embeddings))