X-Git-Url: https://fleuret.org/cgi-bin/gitweb/gitweb.cgi?a=blobdiff_plain;f=mygpt.py;h=3a48cdbb793160ea9c88875d4f353b6a89555477;hb=3dd98b99909b2bca323673263874e2abb39ac10c;hp=d8fd227f63c39a70dded3c55f3c230c3a9d58862;hpb=3e4af6d54fb3d7bd6794035cb79e30ecdcadeb6f;p=mygptrnn.git diff --git a/mygpt.py b/mygpt.py index d8fd227..3a48cdb 100755 --- a/mygpt.py +++ b/mygpt.py @@ -190,6 +190,8 @@ class DumbRec(nn.Module): nb_lines, attention_dropout=0.0, len_max=1e5, + logger=print, + **kwargs, ): super().__init__() @@ -319,6 +321,8 @@ class KVRec(nn.Module): nb_lines, attention_dropout=0.0, len_max=1e5, + logger=print, + **kwargs, ): super().__init__() @@ -471,34 +475,61 @@ class Caterpillar(nn.Module): caterpillar_height, attention_dropout=0.0, len_max=1e5, + logger=print, + **kwargs, ): super().__init__() warnings.warn("Caterpillar", RuntimeWarning) - def randw(*d): - return nn.Parameter(torch.randn(*d) / math.sqrt(d[-1])) + def randw(*d, amplitude=None): + if amplitude is None: + amplitude = 1 / math.sqrt(d[-1]) + return nn.Parameter(amplitude * torch.randn(*d)) self.caterpillar_length = caterpillar_length self.caterpillar_height = caterpillar_height self.attention_dropout = attention_dropout - self.proba_gate_dropout = 0.0 + ###################################################################### + # sup_args + + x = kwargs.get("gate_dropout") + if x is None: + self.proba_gate_dropout = 0.0 + else: + self.proba_gate_dropout = float(x) + + logger(f"self.proba_gate_dropout {self.proba_gate_dropout}") + + x = kwargs.get("default_bg") + if x is None: + default_bg = -math.log(caterpillar_height - 1) + else: + default_bg = float(x) + + logger(f"default_bg {default_bg}") + + ###################################################################### self.w_G = randw(nb_heads, caterpillar_height, dim_model) - self.b_G = nn.Parameter( - torch.full( - (nb_heads, caterpillar_height), -math.log(caterpillar_height - 1) - ) - ) + self.b_G = nn.Parameter(torch.full((nb_heads, caterpillar_height), default_bg)) self.w_K = randw(nb_heads, dim_qk, dim_model) self.w_V = randw(nb_heads, dim_v, dim_model) self.w_Q = randw(nb_heads, dim_qk, dim_model) self.w_O = randw(dim_v * nb_heads, dim_model) - self.init_K_rec = randw(caterpillar_height, caterpillar_length, dim_qk) - self.init_V_rec = randw(caterpillar_height, caterpillar_length, dim_v) + self.init_K_rec = randw( + caterpillar_height, + caterpillar_length, + dim_qk, + ) + self.init_V_rec = randw( + caterpillar_height, + caterpillar_length, + dim_v, + ) def reset_inner_loss(self): self.acc_attention = 0 @@ -520,22 +551,22 @@ class Caterpillar(nn.Module): DV = self.w_V.size(1) DK = self.w_K.size(1) DM = self.w_O.size(1) - CH = self.caterpillar_height - CL = self.caterpillar_length + R = self.caterpillar_height + L = self.caterpillar_length assert ( - t0 >= CL and (t1 - t0) % CL == 0 + t0 >= L and (t1 - t0) % L == 0 ), f"bs.first should be greater than caterpillar_length, and bs.nb should be a multiple of caterpillar_length" # We cache values to deal efficiently with auto-regression if bs.init_cache: - self.rec_V = X.new_zeros(N, CH, T, DV) - self.rec_K = X.new_zeros(N, CH, T, DK) + self.rec_V = X.new_zeros(N, R, T, DV) + self.rec_K = X.new_zeros(N, R, T, DK) # We start the recurrent sequences with optimizable # initial values. No idea if it helps. - self.rec_V[:, :, t0 - CL : t0] = self.init_V_rec[None, :, :, :] - self.rec_K[:, :, t0 - CL : t0] = self.init_K_rec[None, :, :, :] + self.rec_V[:, :, t0 - L : t0] = self.init_V_rec[None, :, :, :] + self.rec_K[:, :, t0 - L : t0] = self.init_K_rec[None, :, :, :] self.cache_Y = X.new_zeros(N, T, DM) @@ -546,49 +577,84 @@ class Caterpillar(nn.Module): # Compute the recurrent state # This is the Gating sequence that modulates the storing of - # the new key and value in the CH pairs of the current - # stack. There are CH independent gating values, which means + # the new key and value in the R pairs of the current + # stack. There are R independent gating values, which means # that the current K/V may be stored in multiple pairs of the # recurrent state, or not at all. G = ( - torch.einsum("ntc,hec->nhet", X, self.w_G) + self.b_G[None, :, :, None] + torch.einsum("ntc,hrc->nhrt", X, self.w_G) + self.b_G[None, :, :, None] ).sigmoid() + # warnings.warn("softmax gating", RuntimeWarning) + + # G = ( + # torch.einsum("ntc,hrc->nhrt", X, self.w_G) + self.b_G[None, :, :, None] + # ).softmax(dim=2) + + ###################################################################### + # The "flashbacks" + + if self.training and self.proba_gate_dropout > 0.0: + # This is a better implementation of "flashbacks". + + # G is NxHxExT where e is the caterpillar's row. + + warnings.warn("gate dropout", RuntimeWarning) + + kill = ( + torch.rand(G.size(), device=G.device) <= self.proba_gate_dropout + ).float() + + alpha = G / (1 - self.proba_gate_dropout) + + G = alpha * (1 - kill) + + ###################################################################### # Clip the gating to avoid values greater than 1 when several # heads hit the same row G = G / G.sum(1, keepdim=True).clamp(min=1) + ###################################################################### + # Roll the gating indexes + + # warnings.warn("rotating barrel", RuntimeWarning) + + # r_barrel = torch.arange(R, device=G.device)[None, None, :, None] + # t_barrel = torch.arange(t1 - t0, device=G.device)[None, None, None, :] + # r_barrel = (r_barrel + (t_barrel + t0) // L) % R + # G = G.gather(dim=2, index=r_barrel.expand_as(G)) + # We prepare the arguments for the parallel scan A = 1 - G.sum(1) - gated_V = torch.einsum("nhet,nhtd->netd", G, V) - gated_K = torch.einsum("nhet,nhtd->netd", G, K) - # We start from cached values, which matters in inference + # warnings.warn("harmonic recurrence", RuntimeWarning) + # har = torch.arange(t0, t1, device = G.device).float() + 1 + # A = har / (har + 1) + # G = G / har - init_rec_V = self.rec_V[:, :, t0 - CL : t0] - init_rec_K = self.rec_K[:, :, t0 - CL : t0] + gated_V = torch.einsum("nhrt,nhtd->nrtd", G, V) + gated_K = torch.einsum("nhrt,nhtd->nrtd", G, K) - ###################################################################### + # We start from cached values, which matters in inference - if self.training and self.proba_gate_dropout > 0.0: - warnings.warn("gate dropout", RuntimeWarning) - epsilon = 0.5 + init_rec_V = self.rec_V[:, :, t0 - L : t0] + init_rec_K = self.rec_K[:, :, t0 - L : t0] ################################################################# # Associative scan # Here there is a trick: Since the stack at position t is - # computed by updating that at position t-CL, the parallel - # scan operates with a period of CL. To do so we split the - # sequence indexing in two axes, the second of size CL, and + # computed by updating that at position t-L, the parallel + # scan operates with a period of L. To do so we split the + # sequence indexing in two axes, the second of size L, and # run the parallel scan using the first as the sequence index. - A = A.unflatten(2, (-1, CL)) - gated_V = gated_V.unflatten(2, (-1, CL)) - gated_K = gated_K.unflatten(2, (-1, CL)) + A = A.unflatten(2, (-1, L)) + gated_V = gated_V.unflatten(2, (-1, L)) + gated_K = gated_K.unflatten(2, (-1, L)) next_V = pscan_dim(A, gated_V, init_rec_V, dim=2) next_K = pscan_dim(A, gated_K, init_rec_K, dim=2) @@ -606,14 +672,14 @@ class Caterpillar(nn.Module): # the column in the caterpillar windowed_V = moving_window( - self.rec_V[:, :, t0 - CL + 1 : t1], dim=2, win_dim=3, win_size=CL + self.rec_V[:, :, t0 - L + 1 : t1], dim=2, win_dim=3, win_size=L ) windowed_K = moving_window( - self.rec_K[:, :, t0 - CL + 1 : t1], dim=2, win_dim=3, win_size=CL + self.rec_K[:, :, t0 - L + 1 : t1], dim=2, win_dim=3, win_size=L ) - # We have an attention score for each of the CHxCL values + # We have an attention score for each of the RxL values ar = torch.einsum( "nhtd,nftld->nhtfl", @@ -655,6 +721,8 @@ class QKVAttention(nn.Module): nb_heads=1, causal=False, attention_dropout=0.0, + logger=print, + **kwargs, ): super().__init__() @@ -746,6 +814,8 @@ class MyGPT(nn.Module): dropout=0.0, len_max=1e5, attention_layer="kvrec", + logger=print, + **kwargs, ): super().__init__() @@ -782,6 +852,8 @@ class MyGPT(nn.Module): nb_heads=nb_heads, causal=causal, attention_dropout=dropout, + logger=logger, + **kwargs, ) elif attention_layer == "dumbrec": return DumbRec( @@ -791,6 +863,8 @@ class MyGPT(nn.Module): nb_heads=nb_heads, nb_lines=nb_lines, attention_dropout=dropout, + logger=logger, + **kwargs, ) elif attention_layer == "kvrec": return KVRec( @@ -800,6 +874,8 @@ class MyGPT(nn.Module): nb_heads=nb_heads, nb_lines=nb_lines, attention_dropout=dropout, + logger=logger, + **kwargs, ) elif attention_layer == "caterpillar": return Caterpillar( @@ -810,6 +886,8 @@ class MyGPT(nn.Module): caterpillar_length=self.caterpillar_length, caterpillar_height=self.caterpillar_height, attention_dropout=dropout, + logger=logger, + **kwargs, ) else: raise ValueError(f"Unknown attention type {attention_layer}.")