-
-
Hi, when I attempted to replicate the experimental results on the SCOPe benchmark, I found significant inconsistencies. When using biencoder's forward_left function during do_retrieval and biencoder's…
-
Hi there,
I tried to reproduce the results of DPR baseline on the full mode dataset. But somehow, from the code, I see that in this baseline, you first generate the embeddings (around 18K chunks) …
-
I have the following model using sequences to predict the next item:
```python
class Model(tfrs.models.Model):
def __init__(self):
super().__init__()
self.query_model = tf…
-
I followed your conversion here and your explanations it really help me a lot on how to implement semantic search.
https://github.com/huggingface/transformers/issues/876
I wanted your suggestio…
-
Dear developers,
Could you please share the hyper-parameters used for the table retrieval model to replicate the results [here](https://github.com/google-research/tapas/blob/master/DENSE_TABLE_RETR…
-
지금은 TF-IDF 기반으로 질문과 지문의 빈도 수를 파악하고 있는데 학습 모델기반으로 수정할 수 있을 것 같아 보입니다.
우선, 모델을 비교해보려면 TF-IDF의 성능을 먼저 검증해야할 것 같습니다.
-
I am working on [LUCENE-8374](https://issues.apache.org/jira/browse/LUCENE-8374) and I would like to have independent verification of my findings. I have looked through `luceneutil` and it does not se…
tokee updated
5 years ago
-
Hello,
I am seeing some behavior I can't explain. We have a query which tries to pull relevant documents from a DB given two embeddings per document (and at runtime, we get two embeddings for this …
-
Hi,
I found a weird thing that if using the multilingual-bert e.g: bert-base-multilingual-uncased, it seems like the grad_cache doesn't work. I know it sounds weird, changing different bert models…