Michael Wilson

When running the `generate` method, the logits are obtained like this: ``` logits = self.model.forward(tokens[:, prev_pos:cur_pos], prev_pos) ``` Initially, `prev_pos=0`, so the first step will return the predictions based on...

### System Info transformers version: 4.27.4 python version: 3.8.8 ### Who can help? @ArthurZucker @younesbelkada ### Information - [ ] The official example scripts - [X] My own modified scripts...

In `model.Transformer.forward`, the following line says it'll only compute the logits for the last position in h: ```python output = self.output(h[:, -1, :]) # only compute last logits ``` I'm...

Thank you for making this package! Apologies if this is out of place, but I had a thought about the memory usage of bridge sampler when running multiple repetitions on...

Parallelize sampling for gen_samples and q22 with multiple repetitions

3

Sampling from a multivariate normal distribution can take a long time. When running bridge sampling with multiple repetitions, sampling to construct `gen_samples` and `q22` currently takes place serially, even if...

Reliability of running multiple repetitions of bridge_sampler when logml cannot be determined within maxiter?

1

I'm fitting some rather complicated models using `brms`, and have been finding difficulty getting `bridge_sampler` to run without getting the warning message about being unable to determine logml within maxiter....

Converting OPT-175B tokenizer to HF format?

2

## ❓ Questions and Help #### What is your question? I've downloaded the weights for OPT-175B using the URL I got after filling out the Google form. I've also got...

It would be helpful to be able to change the name of the `job_jobid_status.tsv` file. (In particular, I would find it useful to be able to allow it to use...