Nayoung Ko comments

Results 4 comments of


                                            Nayoung Ko

Different result in disjoint 15-5s setting.

> Hey! Probably you get different results due to a different batch size... This setting is particularly challenging due to the non-i.i.d. data, so probably decreasing the BS hampers the...

Different result in disjoint 15-5s setting.

> I actually never tried with a lower batch size. The main issue is using a low batch size in the 15-1 increasing the non i.i.d.-ness of data (you may...

Asking about the class token.

> We used the feature embedding from the VIT encoder without using the class token. Alternatively, we initialized a class embedding in the decoder in > > https://github.com/zbwxp/SegVit/blob/85616388cd8983e2d4daafd971540b210e956666/decode_heads/atm_head.py#L239-L240 Thank you...

Minillm: can you share a detail operation to reproduce the result?

> I am checking to see if I am doing this correctly. > > 1. I have downloaded the LLaMA2-13B from HuggingFace (https://huggingface.co/meta-llama/Llama-2-13b-hf). > 2. I have generated a weight...