🎯

Focusing

Thilina Rajapakse ThilinaRajapakse

🎯

Focusing

AI researcher, serial procrastinator, avid reader, fantasy and Sci-Fi geek, and fan of the Oxford comma.

497 followers · 2 following

University of Amsterdam
The Netherlands
https://www.linkedin.com/in/t-rajapakse/
@t_rajapakse

View GitHub Profile

Recently created

Least recently created

Recently updated

Least recently updated

ThilinaRajapakse / retrieval_results.csv

Created November 12, 2021 23:34

	Metric	Value
	MRR@1	0.8084420567920184
	MRR@2	0.8511128165771297
	MRR@3	0.861857252494244
	MRR@5	0.8678357636224099
	MRR@10	0.8696916151981385
	Top 1 Accuracy	0.8084420567920184
	Top 2 Accuracy	0.8937835763622409
	Top 3 Accuracy	0.9260168841135841
	Top 5 Accuracy	0.9516500383729855

ThilinaRajapakse / eval_retrieval.py

Created November 12, 2021 23:23

	import logging

	from simpletransformers.retrieval import RetrievalModel, RetrievalArgs


	logging.basicConfig(level=logging.INFO)
	transformers_logger = logging.getLogger("transformers")
	transformers_logger.setLevel(logging.WARNING)

ThilinaRajapakse / train_retrieval.py

Created November 12, 2021 14:12

	import logging

	from simpletransformers.retrieval import RetrievalModel, RetrievalArgs


	logging.basicConfig(level=logging.INFO)
	transformers_logger = logging.getLogger("transformers")
	transformers_logger.setLevel(logging.WARNING)

ThilinaRajapakse / multi_run.py

Created May 20, 2021 15:30

	import logging

	import pandas as pd

	from simpletransformers.classification import (
	MultiLabelClassificationModel,
	MultiLabelClassificationArgs,
	)

	logging.basicConfig(level=logging.INFO)

ThilinaRajapakse / multiroberta.py

Last active May 20, 2021 15:05

	import torch
	from torch import nn
	from torch.nn import BCEWithLogitsLoss, CrossEntropyLoss, MSELoss
	from transformers import (
	RobertaModel
	)


	class RobertaForMultiLabelSequenceClassification(BertPreTrainedModel):
	"""

ThilinaRajapakse / mt5_scores.csv

Created January 3, 2021 20:19

	model	English to Sinhalese	Sinhalese to English
	mT5	10.3	24.4
	Tatoeba	9.2	22.1

ThilinaRajapakse / eval_mt5_3.py

Created January 3, 2021 20:14

	# Predict
	sinhala_preds = model.predict(to_sinhala)

	eng_sin_bleu = sacrebleu.corpus_bleu(sinhala_preds, sinhala_truth)
	print("--------------------------")
	print("English to Sinhalese: ", eng_sin_bleu.score)

	english_preds = model.predict(to_english)

	sin_eng_bleu = sacrebleu.corpus_bleu(english_preds, english_truth)

ThilinaRajapakse / eval_mt5_2.py

Last active January 3, 2021 20:11

	eval_df = pd.read_csv("data/eval.tsv", sep="\t").astype(str)

	sinhala_truth = [eval_df.loc[eval_df["prefix"] == "translate english to sinhala"]["target_text"].tolist()]
	to_sinhala = eval_df.loc[eval_df["prefix"] == "translate english to sinhala"]["input_text"].tolist()

	english_truth = [eval_df.loc[eval_df["prefix"] == "translate sinhala to english"]["target_text"].tolist()]
	to_english = eval_df.loc[eval_df["prefix"] == "translate sinhala to english"]["input_text"].tolist()

ThilinaRajapakse / eval_mt5_1.py

Last active January 3, 2021 20:08

	import logging
	import sacrebleu
	import pandas as pd
	from simpletransformers.t5 import T5Model, T5Args


	logging.basicConfig(level=logging.INFO)
	transformers_logger = logging.getLogger("transformers")
	transformers_logger.setLevel(logging.WARNING)

ThilinaRajapakse / train_mt5_4.py

Created January 3, 2021 19:35

	# Train the model
	model.train_model(train_df, eval_data=eval_df)

NewerOlder