attention op & Sequential with OrderedDict

python/jittor/__init__.py

@@ -512,11 +512,10 @@ class Module:
             end = 0
             for k in key_:
                 if isinstance(v, nn.Sequential):
-                    if ori_int(k) >= len(v.layers):
+                    v = v[k]
-                        end = 1
+                    if v is None:
+                        end=1
                         break
-                    else:
-                        v = v[ori_int(k)]
                 else:
                     if hasattr(v, k):
                         v = getattr(v, k)

python/jittor/attention.py

@@ -0,0 +1,175 @@
+# ***************************************************************
+# Copyright (c) 2020 Jittor. Authors:
+#     Guowei Yang <471184555@qq.com>
+#     Dun Liang <randonlang@gmail.com>. 
+#
+# All Rights Reserved.
+# This file is subject to the terms and conditions defined in
+# file 'LICENSE.txt', which is part of this source code package.
+# ***************************************************************
+
+import jittor as jt
+from jittor import init, Module, nn
+import numpy as np
+import math
+
+class MultiheadAttention(Module):
+    def __init__(
+        self,
+        embed_dim,
+        num_heads,
+        kdim=None,
+        vdim=None,
+        dropout=0.0,
+        bias=True,
+        add_bias_kv=False,
+        add_zero_attn=False,
+        self_attention=False,
+        encoder_decoder_attention=False,
+        q_noise=0.0,
+        qn_block_size=8,
+    ):
+        super().__init__()
+        self.embed_dim = embed_dim
+        self.kdim = kdim if kdim is not None else embed_dim
+        self.vdim = vdim if vdim is not None else embed_dim
+        self.qkv_same_dim = self.kdim == embed_dim and self.vdim == embed_dim
+
+        self.num_heads = num_heads
+        assert dropout==0, "TODO: dropout>0"
+
+        self.head_dim = embed_dim // num_heads
+        assert (self.head_dim * num_heads == self.embed_dim), "embed_dim must be divisible by num_heads"
+        self.scaling = self.head_dim ** -0.5
+
+        self.self_attention = self_attention
+        self.encoder_decoder_attention = encoder_decoder_attention
+
+        assert not self.self_attention or self.qkv_same_dim, ("Self-attention requires query, key and " "value to be of the same size")
+
+        #TODO: quant_noise
+        self.k_proj = nn.Linear(self.kdim, embed_dim, bias=bias)
+        self.v_proj = nn.Linear(self.vdim, embed_dim, bias=bias)
+        self.q_proj = nn.Linear(embed_dim, embed_dim, bias=bias)
+
+        self.out_proj = nn.Linear(embed_dim, embed_dim, bias=bias)
+
+        assert not add_bias_kv, "TODO: add_bias_kv=True"
+        self.bias_k = self.bias_v = None
+
+        self.add_zero_attn = add_zero_attn
+
+        self.reset_parameters()
+
+        self.onnx_trace = False
+        self.tpu = False
+        
+    def reset_parameters(self):
+        if self.qkv_same_dim:
+            # Empirically observed the convergence to be much better with
+            # the scaled initialization
+            init.xavier_uniform_(self.k_proj.weight, gain=1 / math.sqrt(2))
+            init.xavier_uniform_(self.v_proj.weight, gain=1 / math.sqrt(2))
+            init.xavier_uniform_(self.q_proj.weight, gain=1 / math.sqrt(2))
+        else:
+            init.xavier_uniform_(self.k_proj.weight)
+            init.xavier_uniform_(self.v_proj.weight)
+            init.xavier_uniform_(self.q_proj.weight)
+
+        # init.xavier_uniform_(self.out_proj.weight)
+        if self.out_proj.bias is not None:
+            init.constant_(self.out_proj.bias, 0.)
+        if self.bias_k is not None:
+            init.xavier_normal_(self.bias_k)
+        if self.bias_v is not None:
+            init.xavier_normal_(self.bias_v)
+
+    def execute(
+        self,
+        query,
+        key = None,
+        value = None,
+        key_padding_mask = None,
+        incremental_state = None,
+        need_weights = True,
+        static_kv = False,
+        attn_mask = None,
+        before_softmax = False,
+        need_head_weights = False,
+    ):
+        if need_head_weights:
+            need_weights = True
+
+        tgt_len, bsz, embed_dim = query.shape
+        assert embed_dim == self.embed_dim
+        assert list(query.shape) == [tgt_len, bsz, embed_dim]
+
+        assert incremental_state is None, "TODO: incremental_state is not None"
+        saved_state = None
+
+        if self.self_attention:
+            q = self.q_proj(query)
+            k = self.k_proj(query)
+            v = self.v_proj(query)
+        elif self.encoder_decoder_attention:
+            # encoder-decoder attention
+            q = self.q_proj(query)
+            if key is None:
+                assert value is None
+                k = v = None
+            else:
+                k = self.k_proj(key)
+                v = self.v_proj(key)
+        else:
+            assert key is not None and value is not None
+            q = self.q_proj(query)
+            k = self.k_proj(key)
+            v = self.v_proj(value)
+        q = q*self.scaling
+
+        assert self.bias_k is None, "TODO: self.bias_k is not None:"
+
+        q = q.view(tgt_len, bsz * self.num_heads, self.head_dim).transpose(1, 0, 2)
+        if k is not None:
+            k = k.view(-1, bsz * self.num_heads, self.head_dim).transpose(1, 0, 2)
+        if v is not None:
+            v = v.view(-1, bsz * self.num_heads, self.head_dim).transpose(1, 0, 2)
+
+        assert saved_state is None, "TODO: saved_state is not None"
+        assert k is not None
+        src_len = k.shape[1]
+
+        assert key_padding_mask is None, "TODO: key_padding_mask is not None"
+        assert not self.add_zero_attn, "TODO: self.add_zero_attn=True"
+
+        attn_weights = nn.bmm(q, k.transpose(0, 2, 1))
+
+        assert list(attn_weights.shape) == [bsz * self.num_heads, tgt_len, src_len]
+
+        assert attn_mask is None, "TODO: attn_mask is not None"
+        assert key_padding_mask is None, "TODO: key_padding_mask is not None"
+        
+        if before_softmax:
+            return attn_weights, v
+        
+        attn_weights_float = nn.softmax(attn_weights, dim=-1)
+        attn_weights = attn_weights_float.type_as(attn_weights)
+
+        assert v is not None
+        attn = nn.bmm(attn_weights, v)
+        assert list(attn.shape) == [bsz * self.num_heads, tgt_len, self.head_dim]
+        if self.onnx_trace and attn.shape[1] == 1:
+            # when ONNX tracing a single decoder step (sequence length == 1)
+            # the transpose is a no-op copy before view, thus unnecessary
+            attn = attn.view(tgt_len, bsz, embed_dim)
+        else:
+            attn = attn.transpose(1, 0, 2).view(tgt_len, bsz, embed_dim)
+        attn = self.out_proj(attn)
+        attn_weights = None
+        if need_weights:
+            attn_weights = attn_weights_float.view(bsz, self.num_heads, tgt_len, src_len).transpose(1, 0, 2, 3)
+            if not need_head_weights:
+                # average attention weights over heads
+                attn_weights = attn_weights.mean(dims=[0])
+
+        return attn, attn_weights

python/jittor/init.py

@@ -55,3 +55,30 @@ def relu_invariant_gauss(shape, dtype, mode="fan_in"):
 
 def relu_invariant_gauss_(var, mode="fan_in"):
     var.assign(relu_invariant_gauss(tuple(var.shape), var.dtype, mode))
+
+#TODO: bound = gain * math.sqrt(6.0/fan) ??
+def xavier_uniform(shape, dtype, gain=1.0):
+    assert len(shape)>1
+
+    matsize=1
+    for i in shape[2:]:
+        matsize *= i
+    fan = (shape[1] * matsize) + (shape[0] * matsize)
+    bound = gain * math.sqrt(1.0/fan)
+    return uniform(shape, dtype, -bound, bound)
+
+def xavier_uniform_(var, gain=1.0):
+    var.assign(xavier_uniform(tuple(var.shape), var.dtype, gain))
+
+def xavier_gauss(shape, dtype, gain=1.0):
+    assert len(shape)>1
+    
+    matsize=1
+    for i in shape[2:]:
+        matsize *= i
+    fan = (shape[1] * matsize) + (shape[0] * matsize)
+    std = gain * math.sqrt(2.0/fan)
+    return gauss(shape, dtype, 0, std)
+
+def xavier_gauss_(var, gain=1.0):
+    var.assign(xavier_gauss(tuple(var.shape), var.dtype, gain))

python/jittor/nn.py

@@ -13,6 +13,7 @@
 import jittor as jt
 from jittor import init, Module
 import numpy as np
+import collections
 import math
 from jittor.pool import Pool, pool, AdaptiveAvgPool2d
 from jittor.optim import *
@@ -780,13 +781,17 @@ class Upsample(Module):
 
 class Sequential(Module):
     def __init__(self, *args):
-        self.layers = []
+        self.layers = collections.OrderedDict()
         for mod in args:
-            self.append(mod)
+            if isinstance(mod, collections.OrderedDict):
+                for k, m in mod.items():
+                    self.add_module(k, m)
+            else:
+                self.append(mod)
     def __getitem__(self, idx):
         return self.layers[idx]
     def execute(self, x):
-        for layer in self.layers:
+        for k, layer in self.layers.items():
             x = layer(x)
         return x
     def dfs(self, parents, k, callback, callback_leave):
@@ -794,7 +799,7 @@ class Sequential(Module):
         ret = callback(parents, k, self, n_children)
         if ret == False:
             return
-        for k,v in enumerate(self.layers):
+        for k,v in self.layers.items():
             parents.append(self)
             v.dfs(parents, k, callback, callback_leave)
             parents.pop()
@@ -803,6 +808,10 @@ class Sequential(Module):
     def append(self, mod):
         assert callable(mod), f"Module <{type(mod)}> is not callable"
         assert not isinstance(mod, type), f"Module is not a type"
-        self.layers.append(mod)
+        self.layers[len(self.layers)]=mod
+    def add_module(self, name, mod):
+        assert callable(mod), f"Module <{type(mod)}> is not callable"
+        assert not isinstance(mod, type), f"Module is not a type"
+        self.layers[name]=mod
 
 ModuleList = Sequential

python/jittor/test/test_attention.py

@@ -0,0 +1,65 @@
+
+# ***************************************************************
+# Copyright (c) 2020 Jittor. Authors: 
+#     Guowei Yang <471184555@qq.com>
+#     Dun Liang <randonlang@gmail.com>. 
+# All Rights Reserved.
+# This file is subject to the terms and conditions defined in
+# file 'LICENSE.txt', which is part of this source code package.
+# ***************************************************************
+import unittest
+import jittor as jt
+import jittor.attention as jtatt
+import numpy as np
+
+skip_this_test = False
+
+try:
+    jt.dirty_fix_pytorch_runtime_error()
+    import torch
+    import torch.nn as tnn
+    import fairseq
+except:
+    torch = None
+    tnn = None
+    skip_this_test = True
+    
+def check_equal(q,k,v,tatt,jatt):
+    tq=torch.from_numpy(q)
+    jq=jt.array(q)
+    tk=torch.from_numpy(k)
+    jk=jt.array(k)
+    tv=torch.from_numpy(v)
+    jv=jt.array(v)
+
+    jatt.load_parameters(tatt.state_dict())
+    ty, tw = tatt(tq, tk, tv)
+    jy, jw = jatt(jq, jk, jv)
+    assert np.allclose(ty.detach().numpy(), jy.numpy(), rtol=1e-3)
+    assert np.allclose(tw.detach().numpy(), jw.numpy(), rtol=1e-3)
+
+@unittest.skipIf(skip_this_test, "No Torch found")
+class TestAttention(unittest.TestCase):
+    def test_attention(self):
+        q=np.random.rand(4,8,16).astype(np.float32)
+        k=np.random.rand(4,8,16).astype(np.float32)
+        v=np.random.rand(4,8,16).astype(np.float32)
+
+        tatt=fairseq.modules.multihead_attention.MultiheadAttention(16,1)
+        jatt=jt.attention.MultiheadAttention(16,1)
+        check_equal(q,k,v,tatt,jatt)
+        
+        tatt=fairseq.modules.multihead_attention.MultiheadAttention(16,4)
+        jatt=jt.attention.MultiheadAttention(16,4)
+        check_equal(q,k,v,tatt,jatt)
+        
+        tatt=fairseq.modules.multihead_attention.MultiheadAttention(16,1,self_attention=True)
+        jatt=jt.attention.MultiheadAttention(16,1,self_attention=True)
+        check_equal(q,q,q,tatt,jatt)
+        
+        tatt=fairseq.modules.multihead_attention.MultiheadAttention(16,4,self_attention=True)
+        jatt=jt.attention.MultiheadAttention(16,4,self_attention=True)
+        check_equal(q,q,q,tatt,jatt)
+
+if __name__ == "__main__":
+    unittest.main()