Patrick Fernandes

icon indicating a website link coderpat.github.io icon indicating an email [email protected]

icon company CMU & IST icon location Lisbon, Portugal icon location PhD student @neulab & @deep-spin Previously @Google @Unbabel @microsoft

Results 10 issues of Patrick Fernandes

Tests are missing for all classes

Documentation is missing for some classes

Type annotations missing in some parts of the code

Sharding a model checkpoint for deepspeed usage

3 comment

Hey! I'm using a custom version of this repo to run BLOOM-175B with DeepSpeed and it works great, thank you for this! I was thinking of exploring using large models...

[MT] Add guide for collecting parallel corpora

[MT] Add guide for manual evaluation

[WIP] Add first release for MuDA

[WIP] Add multilingual tagger

Add Multi Encoder option

[BUG] NaNs when unfreezing vision encoder in the multi-modal example

5 comment

**Describe the bug** In the multimodal example in https://github.com/NVIDIA/Megatron-LM/tree/main/examples/multimodal , wehn we run the scripts for either pretraining or finetuning, if we unfreeze the vision encoder we get immediate NaNs...

stale