transfer-learning-conv-ai

transfer-learning-conv-ai copied to clipboard

6 GB VRAM?

1

Hello, is it possible to run the 117M GPT-2 model with 6 GB VRAM using FP16?

Dampfinchen

The lists of arguments in the README do not correctly describe the actual arguments used by the training and interaction scripts. This change simply updates the README so it matches...

micahcantor

Using GPT2-xl in train and interact

1

Hi, i realized that there are a new gpt2-xl pretrained model appears in list here: https://huggingface.co/transformers/pretrained_models.html Is it possible to train with this model and then interact with result ?

GraphGrailAi

Error while saving models

1

Hi, Thank you for sharing us the code. I am using this repo to fine-tune GPT2 model on my conversation dataset. I am using pytorch_transformers to import AdamW, OpenAIGPTDoubleHeadsModel, OpenAIGPTTokenizer,...

Rasipuram

tb_logger Error: got an unexpected keyword argument 'another_engine'

2

Hi, I'm trying to run train.py on Google Colab. interact.py works fine, but when trying to fine-tune, "!python train.py" (with or without arguments) stops here: > INFO:train.py:Build inputs and labels...

RVirmoors

Traceback (most recent call last): File "./interact.py", line 16, in from train import SPECIAL_TOKENS, build_input_from_segments, add_special_tokens_ File "/home/vikas/Desktop/chatbot_thesis/approaches/Hugging_face/train.py", line 14, in from ignite.engine import Engine, Events File "/home/vikas/anaconda3/envs/hugging/lib/python3.6/site-packages/ignite/__init__.py", line 2,...

VikasRajashekar

personalities = [dialog["personality"] for dataset in dataset.values() for dialog in dataset], TypeError: list indices must be integers or slices, not str

1

This error does make sense, "personality is a str not int and thus can't be used as an index. Do we have any fix for this?

RisabBiswas

Is it possible to do the auxilary task multiple times for each instance?

3

Hi, Is it possible to define multiple mc_token for each instance and do an auxilary classification for each of them? If yes, can you guide me? Thanks

fabrahman

How are the distractors made in the dataset?

3

I want to use my own custom dataset with this project, but I don't understand how the distractors were made in the original dataset to get a grasp on how...

Cakeszy

Logical error in train.py

1

Hi Iam facing error for dataset_name, dataset in personachat.items(): #num_candidates = len(dataset[0]["utterances"][0]["candidates"]) num_candidates=2 if args.num_candidates > 0 and dataset_name == 'train': num_candidates = min(args.num_candidates, num_candidates) for dialog in dataset: persona...

monisha08041998