Cassidy Laidlaw

Results 9 issues of Cassidy Laidlaw

Here is a set of images generated by your `elastic` attack on a random sample of CIFAR-10 images against a robust model at eps4 (1): I have no idea what...

## How to reproduce the behaviour I would expect the two sentences below to be tokenized the same way. However, in the second, the special cases for "won't" and "can't"...

enhancement
lang / en
feat / tokenizer

I know somebody else has an open PR for DeepFool in PyTorch but I think my implementation is better in a few ways: * It correctly handles stopping on batches—the...

In the SPSAAdam class in `cleverhans/attacks/spsa.py`, the gradient is calculated with the following code: delta = self._delta delta_x = self._get_delta(x, delta) delta_x = tf.concat([delta_x, -delta_x], axis=0) loss_vals = tf.reshape( loss_fn(x...

Steps to reproduce: * Set up Django project with push_notifications app installed * Connect to MySQL database * Run `ALTER DATABASE database_name CHARACTER SET = utf8mb4 COLLATE = utf8mb4_unicode_ci;` *...

I don't know if this is on the roadmap for this project, but it would be nice! Even without rewriting everything, it might be possible to easily support python 3...

## Why are these changes needed? In newer versions of python, type aliases (e.g., where you do TrainerConfigDict = dict) have to be explicitly defined as a TypeAlias (e.g., TrainerConfigDict:...

After training a reward model according to the instructions, I tried to run BoN sampling via the following command: python src/bon/run_bon_pipeline.py models/rm-pythia-44m_seed0 --seeds 0 However, I got the following traceback:...

Hi, thanks for making your code available! I tried to run the reward model training script: accelerate launch --config_file configs/accelerate_config.yaml src/reward_modeling/training/trainer_rm.py --configs defaults_rm rm-pythia-44m --rng_seed 0 It failed with this...