B. GOR
Looks like suitable solution for voting app or even DAO's. Is it?
The actual error: ``` 0%| | 0/3000 [00:00
It happens after warning: ``` /usr/local/lib/python3.10/dist-packages/trl/trainer/sft_trainer.py:284: UserWarning: You passed a tokenizer with `padding_side` not equal to `right` to the SFTTrainer. This might lead to some unexpected behaviour due to overflow...
This is full script I am running, following tutorial step by step: ``` from transformers import AutoModelForCausalLM, AutoTokenizer import torch from typing import Dict, Iterator, Any from argilla.feedback import TrainingTask...
Okay so how do I change data_collator when passing it to config?
My temporary workaround works, skipping some rows in my dataset, but I'm not sure why I'm doing that and what's correct approach if I want to use all of the...
> hi @7flash , I believe we are using the internal Datasets implementation of train test split to do this so it might be an internal bug. Could you check...
I mean with vector embeddings? @baptisteArno
It's also called semantic router can be done like this ```python message_to_intent = { 'how much money i have?': 'show balance', 'hi there': 'show greetings' } for message in message_to_intent.keys():...
does it work now?