Baptiste Roziere comments

Results 21 comments of


                                            Baptiste Roziere

"&" inside a macro isn't recognized

Hi. I'm experiencing what I think is a related issue: With ``` println!("{}", var1 + var2); ``` the "," and "+" are not in the ast.

Problem with unrecognized argument

Hi, Is this line the command you run in a terminal? ``` %run codegen_sources/preprocessing/preprocess data/test_dataset 20 python obfuscation 8 500 200 400 roberta False data/bpe/cpp-java-python/vocab False data/bpe/cpp-java-python --1 ``` I...

Code clustering

Hi. You can use our released DOBF models to do that (those models are using the roberta tokenizer and architecture). You can start with the DOBF + DAE version for...

UncompletedJobError: Job 195 (task: 0) with path /home/sushantk/anaconda3/codeGen/data/test_dataset/log/195_0_result.pkl has not produced any output (state: FINISHED) No output/error stream produced ! Check: /home/sushantk/anaconda3/codeGen/data/test_dataset/log/195_0_log.out

Hi, Are the logs (.out and .err) files empty? Are some `.tok` files generated or does it fail directly? You may want to try using a local parallelism (`--local_parallelism 4`)...

Failed to generate create self-training dataset per transcoder-st doc

Hi, Yes it should have been something that works on any cluster. Sorry about that. This should fix it: https://github.com/facebookresearch/CodeGen/commit/14a2f9983b26b56dca4c3821b007387cbe1fd93e

problem with transcoder

Hi, Your code contains `enc_path, map_location=lambda storage, loc: storage.cuda(params.local_rank))` on line 181 in __init__.py while the current version is different and this line is not in the file. https://github.com/facebookresearch/CodeGen/blob/9720a8bdba18552cc499975e1f1fb6a7eca74612/codegen_sources/model/src/model/__init__.py#L181 Could...

Baptiste Roziere