too many values to unpack

[pannous]

~/ai/ py nanoblt.py
INFO:root:Model parameters: 15.41M
Traceback (most recent call last):
  File "/Users/me/dev/ai/nanoblt.py", line 675, in <module>
    train(model, config, data)
    ~~~~~^^^^^^^^^^^^^^^^^^^^^
  File "/Users/me/dev/ai/nanoblt.py", line 634, in train
    logits, loss = model(x, y)
                   ~~~~~^^^^^^
  File "/opt/homebrew/lib/python3.13/site-packages/torch/nn/modules/module.py", line 1740, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
           ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
  File "/opt/homebrew/lib/python3.13/site-packages/torch/nn/modules/module.py", line 1751, in _call_impl
    return forward_call(*args, **kwargs)
  File "/Users/me/dev/ai/nanoblt.py", line 477, in forward
    dec_out = self.local_decoder(byte_emb, global_embs)
  File "/opt/homebrew/lib/python3.13/site-packages/torch/nn/modules/module.py", line 1740, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
           ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
  File "/opt/homebrew/lib/python3.13/site-packages/torch/nn/modules/module.py", line 1751, in _call_impl
    return forward_call(*args, **kwargs)
  File "/Users/me/dev/ai/nanoblt.py", line 434, in forward
    B, T = tgt_tokens.shape
    ^^^^
ValueError: too many values to unpack (expected 2)

        print(tgt_tokens.shape)
        # torch.Size([4, 1024, 128])
        print(patch_memory.shape)
        # torch.Size([4, 32, 256])

[HaileyStorm]

Hey, thanks for checking this project out -- it's definitely a work in progress. Haven't touched it in a bit but will be back. If you end up playing with it more and fix things up, feel free to submit a PR :)

…

On Fri, Jan 31, 2025, 3:35 AM pannous ***@***.***> wrote: ~/ai/ py nanoblt.py INFO:root:Model parameters: 15.41M Traceback (most recent call last): File "/Users/me/dev/ai/nanoblt.py", line 675, in <module> train(model, config, data) ~~~~~^^^^^^^^^^^^^^^^^^^^^ File "/Users/me/dev/ai/nanoblt.py", line 634, in train logits, loss = model(x, y) ~~~~~^^^^^^ File "/opt/homebrew/lib/python3.13/site-packages/torch/nn/modules/module.py", line 1740, in _wrapped_call_impl return self._call_impl(*args, **kwargs) ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^ File "/opt/homebrew/lib/python3.13/site-packages/torch/nn/modules/module.py", line 1751, in _call_impl return forward_call(*args, **kwargs) File "/Users/me/dev/ai/nanoblt.py", line 477, in forward dec_out = self.local_decoder(byte_emb, global_embs) File "/opt/homebrew/lib/python3.13/site-packages/torch/nn/modules/module.py", line 1740, in _wrapped_call_impl return self._call_impl(*args, **kwargs) ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^ File "/opt/homebrew/lib/python3.13/site-packages/torch/nn/modules/module.py", line 1751, in _call_impl return forward_call(*args, **kwargs) File "/Users/me/dev/ai/nanoblt.py", line 434, in forward B, T = tgt_tokens.shape ^^^^ ValueError: too many values to unpack (expected 2) print(tgt_tokens.shape) # torch.Size([4, 1024, 128]) print(patch_memory.shape) # torch.Size([4, 32, 256]) — Reply to this email directly, view it on GitHub <#1>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ANIHYIF2QJGU3BHSFWCLA5L2NNGW5AVCNFSM6AAAAABWHIXV42VHI2DSMVQWIX3LMV43ASLTON2WKOZSHAZDGMBRGIYDSNQ> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>