Merge branch 'mpcusack-mpcusack/jitsave'

model.py

@@ -49,7 +49,7 @@ def reshape_for_broadcast(freqs_cis: torch.Tensor, x: torch.Tensor):
     assert 0 <= 1 < ndim
     assert freqs_cis.shape == (x.shape[1], x.shape[-1])
     shape = [d if i == 1 or i == ndim - 1 else 1 for i, d in enumerate(x.shape)]
-    return freqs_cis.view(*shape)
+    return freqs_cis.view(shape)
 
 def apply_rotary_emb(
     xq: torch.Tensor,
@@ -59,8 +59,8 @@ def apply_rotary_emb(
 ) -> Tuple[torch.Tensor, torch.Tensor]:
 
     # reshape xq and xk to match the complex representation
-    xq_r, xq_i = xq.float().reshape(*xq.shape[:-1], -1, 2).unbind(-1)
-    xk_r, xk_i = xk.float().reshape(*xk.shape[:-1], -1, 2).unbind(-1)
+    xq_r, xq_i = xq.float().reshape(xq.shape[:-1] + (-1, 2)).unbind(-1)
+    xk_r, xk_i = xk.float().reshape(xk.shape[:-1] + (-1, 2)).unbind(-1)
     
     # reshape freqs_cos and freqs_sin for broadcasting
     freqs_cos = reshape_for_broadcast(freqs_cos, xq_r)
@@ -142,10 +142,11 @@ class Attention(nn.Module):
 
         # flash implementation
         if self.flash:
-            output = torch.nn.functional.scaled_dot_product_attention(xq, xk, xv, attn_mask=None, dropout_p=self.dropout if self.training else 0, is_causal=True)
+            output = torch.nn.functional.scaled_dot_product_attention(xq, xk, xv, attn_mask=None, dropout_p=self.dropout if self.training else 0.0, is_causal=True)
         else:
             # manual implementation
             scores = torch.matmul(xq, xk.transpose(2, 3)) / math.sqrt(self.head_dim)
+            assert hasattr(self, 'mask')
             scores = scores + self.mask[:, :, :seqlen, :seqlen]   # (bs, n_local_heads, seqlen, cache_len + seqlen)
             scores = F.softmax(scores.float(), dim=-1).type_as(xq)
             scores = self.attn_dropout(scores)
@@ -198,6 +199,8 @@ class TransformerBlock(nn.Module):
 
 
 class Transformer(nn.Module):
+    last_loss: Optional[torch.Tensor]
+
     def __init__(self, params: ModelArgs):
         super().__init__()
         self.params = params
@@ -227,6 +230,9 @@ class Transformer(nn.Module):
             if pn.endswith('w3.weight') or pn.endswith('wo.weight'):
                 torch.nn.init.normal_(p, mean=0.0, std=0.02/math.sqrt(2 * params.n_layers))
 
+        # Initialize attribute for the loss of the last forward call. This will be set if the forward is called with a targets tensor.
+        self.last_loss = None
+
     def _init_weights(self, module):
         if isinstance(module, nn.Linear):
             torch.nn.init.normal_(module.weight, mean=0.0, std=0.02)
@@ -235,7 +241,7 @@ class Transformer(nn.Module):
         elif isinstance(module, nn.Embedding):
             torch.nn.init.normal_(module.weight, mean=0.0, std=0.02)
 
-    def forward(self, tokens, targets=None):
+    def forward(self, tokens: torch.Tensor, targets: Optional[torch.Tensor] = None) -> torch.Tensor:
         _bsz, seqlen = tokens.shape
         h = self.tok_embeddings(tokens)
         h = self.dropout(h)
@@ -249,13 +255,13 @@ class Transformer(nn.Module):
         if targets is not None:
             # if we are given some desired targets also calculate the loss
             logits = self.output(h)
-            loss = F.cross_entropy(logits.view(-1, logits.size(-1)), targets.view(-1), ignore_index=-1)
+            self.last_loss = F.cross_entropy(logits.view(-1, logits.size(-1)), targets.view(-1), ignore_index=-1)
         else:
             # inference-time mini-optimization: only forward the output on the very last position
             logits = self.output(h[:, [-1], :]) # note: using list [-1] to preserve the time dim
-            loss = None
+            self.last_loss = None
 
-        return logits, loss
+        return logits
 
     def configure_optimizers(self, weight_decay, learning_rate, betas, device_type):
         # start with all of the candidate parameters

save_torchscript.py

@@ -0,0 +1,66 @@
+#!/usr/bin/env python
+"""Saves the model as a TorchScript.
+
+Usage examples:
+    ./save_torchscript.py
+    ./save_torchscript.py --dim=300
+    ./save_torchscript.py --gzip_output=True --zero_params=True
+
+The resulting file can be loaded in C++ code and then used for training or
+inference with:
+    #include <torch/script.h>
+    torch::jit::Module module = torch::jit::load("model.pt")
+
+Note that the serialized model includes the initial parameters and with the default
+ModelArgs the file is 59M and gzips down to 55M. If you want to serialize/distribute
+the model parameters separately you can zero out the parameters before saving it and
+it will gzip down to 780K.
+"""
+import gzip
+import os
+import shutil
+from inspect import signature
+
+import torch
+
+from model import ModelArgs, Transformer
+
+# Model args config
+dim = 288
+n_layers = 6
+n_heads = 6
+n_kv_heads = n_heads
+multiple_of = 32
+max_seq_len = 256
+dropout = 0.0
+vocab_size = 32000
+norm_eps = 1e-5
+# Save config
+model_path = "model.pt"
+zero_params = False
+gzip_output = False
+# Allow config overrides
+exec(open("configurator.py").read())
+
+
+def main() -> None:
+    model_args = {k: globals()[k] for k in signature(ModelArgs).parameters}
+    model = Transformer(ModelArgs(**model_args))
+
+    # If requested zero params before saving the model. This is useful in
+    # conjunction with gzip_output.
+    if zero_params:
+        for p in model.parameters():
+            p.detach().zero_()
+
+    torch.jit.save(torch.jit.script(model), model_path)
+
+    if gzip_output:
+        with open(model_path, "rb") as f_in:
+            with gzip.open(f"{model_path}.gz", "wb") as f_out:
+                shutil.copyfileobj(f_in, f_out)
+        os.unlink(model_path)
+
+
+if __name__ == "__main__":
+    main()

train.py

@@ -211,7 +211,8 @@ def estimate_loss():
         for k in range(eval_iters):
             X, Y = next(batch_iter)
             with ctx:
-                logits, loss = model(X, Y)
+                logits = model(X, Y)
+                loss = model.last_loss
             losses[k] = loss.item()
         out[split] = losses.mean()
     model.train()
@@ -294,7 +295,8 @@ while True:
             # looking at the source of that context manager, it just toggles this variable
             model.require_backward_grad_sync = micro_step == gradient_accumulation_steps - 1
         with ctx:
-            logits, loss = model(X, Y)
+            logits = model(X, Y)
+            loss = model.last_loss
             loss = loss / gradient_accumulation_steps
         # immediately async prefetch next batch while model is doing the forward pass on the GPU
         X, Y = next(train_batch_iter)