Nguyen Van Nha

i'm not found vocab in model DCTSS for japanese, you can share with me, thank.

large number of users and items

2

i have 2M user and 300k items, ram is not enough, How do you implement????

output wrong

1

i try with ls_bert.when i print last hidden state of bert model, all equal to 0. output class of hf model và ls model are different 

Sample code training and dataset of ColBert V2

7

hope you release the training code of colbert v2

Can use character input ?

2

i want train this model with japanese, i can use character as input Thank

i read this paper and try running this code but evaluate on test2017 only achieve MAP = 83% . i use glove embeddings -300 dimension...

Do you have plan to release instruction model LLAMA-I?

[Chatllama] error when load dataset when use deepspeed

5

hi, when I use deepspeed , I encountered this error: [2023-03-09 10:46:33,647] [INFO] [logging.py:77:log_dist] [Rank 0] DeepSpeed Flops Profiler Enabled: False Traceback (most recent call last): File "/datahdd/nhanv/Projects/NLP/chatllama/artifacts/main.py", line 50,...

[Chatllama] Should I add a <end_of_text> at the end of sentence?

3

When I train a actor model with bloom-560M. I realize that , the model generates the text repeated at the end. The model always generates enough words predefined by max_length...

I don't understand the difference of weighted_token_accuracy and weighted cross_entropy loss. Can you explain?