OPT with quantizable MatMuls (#85)

Author: natuan

src/transformers/models/opt/modeling_opt.py

@@ -120,6 +120,46 @@ def forward(self, attention_mask: torch.LongTensor, past_key_values_length: int
         return super().forward(positions + self.offset)
 
 
+class BMMLeftInput_QK(nn.Identity):
+    ...
+
+
+class BMMRightInput_QK(nn.Identity):
+    ...
+
+
+class BMMOutput_QK(nn.Identity):
+    ...
+
+
+class BMMLeftInput_PV(nn.Identity):
+    ...
+
+
+class BMMRightInput_PV(nn.Identity):
+    ...
+
+
+class BMMOutput_PV(nn.Identity):
+    ...
+
+
+class QuantizableBatchMatMul(nn.Module):
+    """
+    Wrapper around torch.bmm with distinct inputs/output class
+    instances that could be quantized through SparseML recipe
+    """
+
+    def __init__(self, left_input_cls, right_input_cls, output_cls):
+        super().__init__()
+        self.left_input = left_input_cls()
+        self.right_input = right_input_cls()
+        self.output = output_cls()
+
+    def forward(self, a: torch.Tensor, b: torch.Tensor):
+        return self.output(torch.bmm(self.left_input(a), self.right_input(b)))
+
+
 class OPTAttention(nn.Module):
     """Multi-headed attention from 'Attention Is All You Need' paper"""
 
@@ -150,6 +190,9 @@ def __init__(
         self.q_proj = nn.Linear(embed_dim, embed_dim, bias=bias)
         self.out_proj = nn.Linear(embed_dim, embed_dim, bias=bias)
 
+        self.attn_weights_bmm = QuantizableBatchMatMul(BMMLeftInput_QK, BMMRightInput_QK, BMMOutput_QK)
+        self.attn_output_bmm = QuantizableBatchMatMul(BMMLeftInput_PV, BMMRightInput_PV, BMMOutput_PV)
+
     def _shape(self, tensor: torch.Tensor, seq_len: int, bsz: int):
         return tensor.view(bsz, seq_len, self.num_heads, self.head_dim).transpose(1, 2).contiguous()
 
@@ -208,7 +251,7 @@ def forward(
         value_states = value_states.view(*proj_shape)
 
         src_len = key_states.size(1)
-        attn_weights = torch.bmm(query_states, key_states.transpose(1, 2))
+        attn_weights = self.attn_weights_bmm(query_states, key_states.transpose(1, 2))
 
         if attn_weights.size() != (bsz * self.num_heads, tgt_len, src_len):
             raise ValueError(
@@ -254,7 +297,7 @@ def forward(
 
         attn_probs = nn.functional.dropout(attn_weights, p=self.dropout, training=self.training)
 
-        attn_output = torch.bmm(attn_probs, value_states)
+        attn_output = self.attn_output_bmm(attn_probs, value_states)
 
         if attn_output.size() != (bsz * self.num_heads, tgt_len, self.head_dim):
             raise ValueError(