Tanmay Gupta
Tanmay Gupta
I would like to second @mattf1n 's feature request. This would be super useful for vision-language modeling where we often want to feed a concatenation of image and text features...
@sanchit-gandhi I am traveling currently but gave it a quick try yesterday. Specifically, I was trying to update `FlaxT5ForConditionalGeneration` to accept `input_embed`. It looks like `FlaxGenerationMixin` in `generation_flax_utils.py` would also...
@sanchit-gandhi here's the [PR](https://github.com/huggingface/transformers/pull/18613) you requested. I was actually able to get it to work with minimal modifications to `generation_flax_utils.py`. @mattf1n a similar solution might work for GPT-2 as well?
Could you share an example of how that works in jsonnet? Also, is there something similar for yaml? I prefer yaml slightly because it has less syntax/brackets etc cluttering the...
I would like to second @isunitha98selvan's feature request for being able to index multiple returned values. In the following example, `InferenceStep` needs to refer to individual outputs from `ModelCreationStep`. I...
That would work for me!
Thanks, setting the version number worked!
I removed that directory
@dirkgr any updates on this bug? Even though version numbers are great, sometimes I want to recache a step because of a change somewhere else in the code that affects...
Hi @alibabadoufu, That's a great experiment suggestion and is certainly worth trying! Just one thing to note is that this "random-substitution" baseline while not involving context in substitute sampling, will...