curiousily

/

alpaca-bitcoin-tweets-sentiment

like 4

Venelin ,

(1) Firstly thanks for this great work. I found this: https://www.mlexpert.io/machine-learning/tutorials/alpaca-fine-tuning#data . I think I followed the code and data exactly from your tutorial. But today I verified output from generate.py, what i got is:

(2) I got the repeat information below when I call generate.py by loading my fine-tuned model after following your above tutorial (https://huggingface.co./linpang/alpaca-bitcoin-tweets-sentiment/tree/main) :

Instruction:

Determine the sentiment

Input:

I am experimenting whether I can live only with bit coins donated. Please cooperate

Response:

output= Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request.

(3) But when I load your model: 'curiousily/alpaca-bitcoin-tweets-sentiment', I got correct response (prompter.get_response= Positive) :

prompt= Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request.

Instruction:

Determine the sentiment

Input:

I am experimenting whether I can live only with bit coins donated. Please cooperate

Response:

output= Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request.

Instruction:

Determine the sentiment

Input:

I am experimenting whether I can live only with bit coins donated. Please cooperate

Response:

Positive
prompter.get_response= Positive

(4) It seems your adapter_model.bin is much larger than mine (16.8M >>443 bytes).
(5) Can you please let me know what the problem might be? I checked log file on my side, the log shows training and val_loss are descreasing.

Please help and look forward to hearing from you soon!

Thanks a lot!