Pietro Lesci

Hi there, I am trying to train an NER tagger on my own data. In the docs there does not seem to be worked out examples of using flair to...

🐛 LocScaleReparam and enumeration with NUTS

4

## Context Hi there, While trying to reproduce the [annotators.py](https://github.com/pyro-ppl/numpyro/blob/master/examples/annotation.py) example - originally written in numpyro - in pyro, I've come across what might be a bug. Discussion on pyro...

Return pytorch tensor for mini-imagenet labels?

2

https://github.com/learnables/learn2learn/blob/06893e847693a0227d5f35a6e065e6161bb08201/learn2learn/vision/datasets/mini_imagenet.py#L111 Currently, when loading mini-imagenet the inputs are returned as pytorch tensors while the labels as numpy arrays. Since the user will likely use both in a training loop, does...

Support sequence tagging evaluation metrics (NLP)

6

## 🚀 Feature Support for sequence tagging evaluation metrics _à la_ [`seqeval`](https://github.com/chakki-works/seqeval). That is, support the evaluation of the performance of chunking tasks such as named-entity recognition, part-of-speech tagging, semantic...

Hi there, Is there an easy way to convert the MultiWoZ 2.1 dataset to the same format as the 2.2? Best, Pietro

### 🐛 Describe the bug ## Description Hi there, I would like to get per-sample gradients (application: transformer classifier for text). in a preliminary test, I noticed that when computing...

Add training loss data

1

Task description: "Collect all loss values into CSV files from WandB and -- if needed -- log files". The most important file is `pythia_runs.tsv` in which I manually collect the...

[Pythia on Pile-Dedup] Training for ~1.5 epochs: how to identify the repeated sequences (i.e., the additional .5 epoch)?

3

Hi there, The deduplicated dataset has fewer sequences and to keep a consistent token count with the non-deduplicated version the models are trained for ~1.5 epochs (as discussed in the...

Support for text representation transforms?

3

Hi there, First of all, thanks for creating this amazing library - investing time and money to support it, and making it open source. I discovered it at JuliaCon 2021...

Hi there, I was wondering whether there is an easy way to ask the tokeniser trainer to return the counts (or frequency) of the pair in the moment the merge...