AttributeError: Can't get attribute 'conditional_trans_classification' on <module '__mp_main__' from '~/CogLTX/run_20news.py'>
Did anyone met the same problem? Thanks!!
I ran into this problem, but I simply deleted the function because config.init_relevance is null and will not be used by the code below
I ran into this problem, but I simply deleted the function because config.init_relevance is null and will not be used by the code below
I followed solutions from https://github.com/Sleepychord/CogLTX/issues/3 and https://github.com/Sleepychord/CogLTX/issues/2, yet it throws another exception
When training introspector, specifically outputs = self.roberta( input_ids, attention_mask=attention_mask, token_type_ids=token_type_ids, position_ids=position_ids, head_mask=head_mask, inputs_embeds=inputs_embeds, ) at model.py Line31-Line38, it throws:
RuntimeError: CUDA error: CUBLAS_STATUS_EXECUTION_FAILED when calling cublasSgemm( handle, opa, opb, m, n, k, &alpha, a, lda, b, ldb, &beta, c, ldc)