Wasi Ahmad

IndexError: tuple index out of range

3

Getting the following error. Any suggestion to solve it? > 2018-06-04 02:49:35,663 [MainThread ] [INFO ] Writing summaries. > 2018-06-04 02:49:35,664 [MainThread ] [INFO ] Processing summaries. Saving system files...

Is re_ranking implementation wrong?

5

Either my understanding is wrong or the implementation of re-ranking is wrong. The [EvaluateRetrieval.rerank](https://github.com/beir-cellar/beir/blob/c3334fd5b336dba03c5e3e605a82fcfb1bdf667d/beir/retrieval/evaluation.py#L25) looks like: ``` def rerank(self, corpus: Dict[str, Dict[str, str]], queries: Dict[str, str], results: Dict[str, Dict[str, float]],...

Hi, the full data is shared but the train/valid/test split is not provided. Do you plan to share it soon? Without the split (in-domain, out-of-domain), we won't be able to...

If I have understood the part of the [repetition model](https://github.com/ari-holtzman/l2w/blob/master/word_rep/reprnn.py#L41) correctly, there is no such restriction of using a fixed window of size `k` as mentioned in the paper: "In...

*Issue #, if available:* #7 *Description of changes:* By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your...

Hello, We have come across your work on Code Retrieval. Can you benchmark our code embedding models, [CodeSage](https://huggingface.co/codesage)? Thank you.

How synthetic data were generated?

1

Hello, In the technical report, it was mentioned that CodeQwen1.5 was used to generate synthetic data, but there is no further detail. Is it possible to elaborate? For example, what...

Why DeepSeek-Coder-v2 236B is not trained with FIM objective?

3

The paper mentions that DeepSeek-Coder-v2 236B is trained by only utilizing the Next-Token-Prediction objective. No FIM objective is used. Is there any reason not to use FIM?

How does StarCoder-15b generates concept, instruction, and responses?

3

Is StarCoder-15b an instruct model itself? If not, how does it is utilized in concept, instruction, and response generation?

A couple of questions. 1. What is the maximum length you considered for models to generate during inference? 2. For FIM tasks, typically models generate a special symbol to indicate...