Am I responsible for not reproducibility of collaborators’ result?

It sounds like you are in a difficult position, but, yes, if your name is on a publication or report, people will assume that you are responsible for everything in it.

It may be that you have been caught by your own carelessness, of course, accepting something you weren't sure about because they "seem more relevant to me". It is hard to be definite here with so little information, but perhaps you need to be more careful.

But the situation now needs some resolution. This is something you need to work out with your advisor and perhaps your collaborators. I think the only way to really solve it is to have a general meeting where you discuss the situation. Someone, somewhere, seems to have been making improper assumptions. Perhaps there is shared blame here. But if all fingers are now pointing at you, then you have to work to find a general resolution as the situation isn't likely to change without your action.

Your attitude in such a meeting should be "this is what happened... this is what I assumed... this is why I acted as I did", being entirely truthful. Some of the blame may fall to you, but better that it be worked out than that it not be.

This is a grey area, and there are people on both sides of the fence. Take a look at the Schön scandal, especially the "responsibilities of coauthors" section of the Bell Labs investigative committee report. Some summaries:

• Co-authors share a significant part of the credit.
• If amazing results are reported, then the simple presence of a well-known researcher as a co-author gives credibility to the paper, even before it is reproduced.
• However if a co-author's work turns out to be wrong, inaccurate, or (even worse) fraudulent, everyone else's name on the paper is suspect. Negative reputational damage is virtually assured.
• ... nonetheless, at least in this particular case, all the coauthors were exonerated of misconduct.

Notably, the committee agrees that it's an "extraordinarily difficult" question, and they don't feel qualified to make a specific judgment. Therefore, you'll have to draw your own conclusions about what it means to be a coauthor, and make your own decisions on whether or not you care about being able to reproduce your collaborator's results.