search

github / CodeSearchNet

Datasets, tools, and benchmarks for representation learning of code.
https://arxiv.org/abs/1909.09436
MIT License
2.18k stars 385 forks source link

Benchmark Submission: mallamanis #201

Closed mallamanis closed 4 years ago

mallamanis commented 4 years ago

This pull request represents a submission to the codesearchnet benchmark.

It is the main communication channel for you and the reviewers.

Wandb run results for review

mallamanis commented 4 years ago

Update to the baseline model, due to the n_tree issue. This improves from the previous run from 0.113 to 0.239

(related to #198 )