Simon Levine

Hello, Was wondering if you might provide some literature or background on your AlgebraicLoss implementation as I can't seem to find anything by that name. Thanks.

change TextTokenizer 2DConvolution to 1D

1

Hello, It seems to make more intuitive sense to use 1D convolutions here over the embedding with a channel size equal to the word embedding dimension, rather than the edge-case...

Issue with Tokenizer (fast) splitting `<mask>` into constituent added special tokens despite mask token in vocab and in special tokens map

10

### System Info - `transformers` version: 4.25.1 - Platform: Linux-5.4.0-91-generic-x86_64-with-glibc2.31 - Python version: 3.10.6 - Huggingface_hub version: 0.11.0.rc0 - PyTorch version (GPU?): 1.13.0+cu117 (True) - Tensorflow version (GPU?): not installed...

https://github.com/RosettaCommons/RFdiffusion/blob/main/rfdiffusion/inference/model_runners.py#L673 ``` msa_prev, pair_prev, px0, state_prev, alpha, logits, plddt = self.model(msa_masked, msa_full, seq_in, xt_in, idx_pdb, t1d=t1d, t2d=t2d, xyz_t=xyz_t, alpha_t=alpha_t, msa_prev = None, pair_prev = None, state_prev = None, t=torch.tensor(t), return_infer=True,...

Hi, I'm trying to use this framework at the batch level within another training framework. Is there a way to make it possible to try using a recorded output before...

Make use_kernels positional arg in PairformerNoSeqModule to avoid torch issues when use_reentrant=False is not passed.