How to get topk output from FM (context aware recommender model).

[reemim]

What is the correct configuration for an FM (context-aware recommender) to produce the top k results. Below is my code and configuration file.

CONFIG

dataset config : General Recommendation

USER_ID_FIELD: user_id
ITEM_ID_FIELD: item_id
load_col:
inter: ['user_id', 'item_id', 'rating', 'timestamp']
user: ['user_id', 'age', 'gender', 'occupation']
item: ['item_id', 'release_year', 'class']

model config

embedding_size: 10

Training and evaluation config

epochs: 10
train_batch_size: 4096
eval_batch_size: 4096
train_neg_sample_args:
distribution: uniform
sample_num: 1
alpha: 1.0
dynamic: False
candidate_num: 0
eval_args:
split: {'RS':[0.8,0.1,0.1]}
order: RO
group_by: ~
mode: labeled
metrics: ['Recall', 'MRR', 'NDCG', 'Hit', 'Precision']
topk: 10
valid_metric: MRR@10
metric_decimal_place: 4

---

CODE

from logging import getLogger
from recbole.config import Config
from recbole.data import create_dataset, data_preparation
from recbole.model.context_aware_recommender import FM
from recbole.trainer import Trainer
from recbole.utils import init_seed, init_logger
from recbole.quick_start.quick_start import load_data_and_model
from recbole.quick_start import run_recbole

if name == 'main':

# configurations initialization
config = Config(model='FM', dataset='ml-100k', config_file_list=['/home/uceerm5/RS_testing/test_context.yaml'])

# init random seed
init_seed(config['seed'], config['reproducibility'])

# logger initialization
init_logger(config)
logger = getLogger()

# write config info into log
logger.info(config)

# dataset creating and filtering
dataset = create_dataset(config)
logger.info(dataset)

# dataset splitting
train_data, valid_data, test_data = data_preparation(config, dataset)

# model loading and initialization
model = FM(config, train_data.dataset).to(config['device'])
logger.info(model)

# trainer loading and initialization
trainer = Trainer(config, model)

# model training
best_valid_score, best_valid_result = trainer.fit(train_data, valid_data)

# model evaluation
test_result = trainer.evaluate(test_data)
print(test_result)

---

When the last three lines of the config file were changed to the below the code was working properly but the results were not top k.

train_neg_sample_args: ~
metrics: ['AUC', 'LogLoss']
valid_metric: AUC

[VladHardAndSoftWare]

You can use code from paragraph 4 (https://www.kaggle.com/code/astrung/recbole-lstm-sequential-for-recomendation-tutorial)