Update.

[picoclvr.git] / mygpt.py

diff --git a/mygpt.py b/mygpt.py
index b4446c6..c93010a 100755 (executable)
--- a/mygpt.py
+++ b/mygpt.py
@@ -5,6 +5,11 @@
 
 # Written by Francois Fleuret <francois@fleuret.org>
 
+# This is an implementation from scratch of a "GPT", that is a model
+# composed of several causal self-attention blocks. It is equipped
+# with a caching mechanism for keys and values to avoid a O(N^3) cost
+# for auto-regression.
+
 import math
 
 import torch
@@ -14,19 +19,6 @@ from torch.nn import functional as F
 
 ######################################################################
 
-
-class WithResidual(nn.Module):
-    def __init__(self, *f):
-        super().__init__()
-        self.f = f[0] if len(f) == 1 else nn.Sequential(*f)
-
-    def forward(self, bs):
-        bs.x = bs.x + self.f(bs).x
-        return bs
-
-
-######################################################################
-
 # A BracketedSequence is a BxTx... tensor with a first and a nb time
 # steps to compute.
 
@@ -78,6 +70,19 @@ class CacheWrapper(nn.Module):
 ##############################
 
 
+class WithResidual(nn.Module):
+    def __init__(self, *f):
+        super().__init__()
+        self.f = f[0] if len(f) == 1 else nn.Sequential(*f)
+
+    def forward(self, bs):
+        bs.x = bs.x + self.f(bs).x
+        return bs
+
+
+##############################
+
+
 class AddPositionalEncoding(nn.Module):
     def __init__(self, len_max):
         super().__init__()
@@ -253,6 +258,30 @@ class MyGPT(nn.Module):
         bs = self.readout(bs)
         return bs
 
+    # ar_mask is a tensor with 0s and 1s, of same shape as input, with
+    # 1s where tokens should be generated. The others are kept
+    # unchanged.
+
+    def masked_inplace_autoregression(
+        self, input, ar_mask, forbidden_tokens=None, deterministic_synthesis=False
+    ):
+        to_generate = (ar_mask.sum(0) > 0).nonzero()
+        if to_generate.min() > 0:
+            self(
+                BracketedSequence(input, 0, to_generate.min())
+            )  # Needed to initialize the model's cache
+        for s in range(to_generate.min(), to_generate.max() + 1):
+            output = self(BracketedSequence(input, s, 1)).x
+            logits = output[:, s]
+            if forbidden_tokens is not None:
+                logits = logits.masked_fill(forbidden_tokens, float("-inf"))
+            if deterministic_synthesis:
+                t_next = logits.argmax(1)
+            else:
+                dist = torch.distributions.categorical.Categorical(logits=logits)
+                t_next = dist.sample()
+            input[:, s] = ar_mask[:, s] * t_next + (1 - ar_mask[:, s]) * input[:, s]
+
 
 ######################################################################
 
@@ -281,8 +310,6 @@ if __name__ == "__main__":
         z = model(BracketedSequence(x, s, 1))
         y2[:, s] = z.x[:, s]
 
-    # print(y1.max(dim = 2).values)
-    # print(y2.max(dim = 2).values)
     print(f"error={((y1 - y2).norm() / (y1.norm() + y2.norm())).item()}")
 
 ######################################################################