include HQQLinear in find target_linear

src/axolotl/utils/models.py

@@ -854,13 +854,13 @@ class ModelLoader:
             self.cfg.adapter in ["qlora", "lora"]
             and hasattr(self.model_config, "quantization_config")
             and self.model_config.quantization_config["quant_method"]
-            in ["gptq", "awq", "bitsandbytes", "hqq"]
+            in ["gptq", "awq", "bitsandbytes"]
+            and not self.cfg.hqq
         ):
             quant_config_class_dict = {
                 "gptq": GPTQConfig,
                 "awq": AwqConfig,
                 "bitsandbytes": BitsAndBytesConfig,
-                "hqq": HqqConfig,
             }
 
             quant_config_class = quant_config_class_dict[
@@ -904,7 +904,7 @@ class ModelLoader:
                 **bnb_config,
             )
 
-        elif self.cfg.hqq:
+        if self.cfg.hqq:
             from axolotl.utils.schemas.quant import get_hqq_quant_config_kwargs
 
             self.model_kwargs["quantization_config"] = HqqConfig(
@@ -1471,7 +1471,16 @@ def load_llama_adapter(model, cfg):
 
 
 def find_all_linear_names(model):
-    cls = (bnb.nn.Linear4bit, bnb.nn.Linear8bitLt, torch.nn.Linear)
+    from hqq.core.peft import HQQLinearLoRA
+    from hqq.core.quantize import HQQLinear
+
+    cls = (
+        bnb.nn.Linear4bit,
+        bnb.nn.Linear8bitLt,
+        torch.nn.Linear,
+        HQQLinear,
+        HQQLinearLoRA,
+    )
     lora_module_names = set()
     for name, module in model.named_modules():
         if (