fix cpu nf4 (bitsandbytes-foundation#1432)

Signed-off-by: jiqing-feng <[email protected]>

Author: jiqing-feng

bitsandbytes/autograd/_functions.py

@@ -579,7 +579,8 @@ def matmul_4bit(
     assert quant_state is not None
     if A.device.type in ("cpu", "xpu") and A.requires_grad == False:
         if getattr(quant_state, "ipex", False):
-            out = F.gemv_4bit(A, B.t(), out, state=quant_state)
+            B = B.t() if len(B.shape) == 2 else B
+            out = F.gemv_4bit(A, B, out, state=quant_state)
             if bias is not None:
                 out += bias
             return out

bitsandbytes/nn/modules.py

@@ -508,7 +508,8 @@ def forward(self, x: torch.Tensor):
             x = x.to(self.compute_dtype)
 
         bias = None if self.bias is None else self.bias.to(self.compute_dtype)
-        out = bnb.matmul_4bit(x, self.weight.t(), bias=bias, quant_state=self.weight.quant_state)
+        weight = self.weight.t() if len(self.weight.shape) == 2 else self.weight
+        out = bnb.matmul_4bit(x, weight, bias=bias, quant_state=self.weight.quant_state)
 
         out = out.to(inp_dtype)