Academic Connect is now available as a mobile app.download links inside (20)

Great question.I went through something very similar in my second year.

The key insight for me was that LoRA is actually quite well-suited for NER tasks, especially in low-resource settings. I would recommend:

1. Use LoRA with r=8 or r=16.don't go higher for 8K samples
2. Apply LoRA to attention layers only, not the feed-forward layers
3. Use a cosine learning rate schedule with warm-up (10% of steps)

For Telugu-English code-mixed NER specifically, you might also look at MuRIL.it is pretrained on Indian language data and often outperforms XLM-R on Indic tasks even with less fine-tuning data.

[Moderator] Pinning this thread as it contains highly useful information for new PhD students.

The short answer: it depends heavily on the journal and field, but here is a general framework.

If your null result is well-powered and the hypothesis was reasonable: Absolutely publish it. Null results in well-designed studies are as valuable as positive results. Journals like PLOS ONE explicitly welcome them.

If your study is underpowered: Be transparent about it. Report effect sizes and confidence intervals, not just p-values. A small, underpowered study with honest reporting is still a contribution.

What kills papers is when limitations are obvious but the authors have not addressed them.

I'd push back slightly on the idea that industry collaboration is universally beneficial for academic research. The incentive misalignment is real.

Industry partners want deliverables on a schedule. Academic research is often exploratory and unpredictable. When the two clash, it is usually the researcher who has to compromise.either by rushing results, constraining publication, or steering away from results that are inconvenient for the sponsor.

This doesn't mean industry collaboration is bad. But the terms matter enormously. IP rights, publication rights, and the ability to pursue negative results should be negotiated upfront and protected in writing.

Check whether your institution has a Springer/Elsevier waiver agreement.many NITs and IITs do.

I am in a very similar situation. Would you be willing to share the outline of your PMRF proposal? Not the content.just the section headings and approximate word allocation.

Respectfully, I think this framing misses the systemic dimension. Individual researchers cannot solve a collective action problem. If you submit to open-access journals and your colleague submits to Nature, your colleague gets the promotion. Until evaluation criteria change at the institutional level.which requires policy intervention.individual choices have minimal impact.

If you are not already using a reading group format with your students, I would strongly recommend it. Forces everyone to articulate what they read.

FDR (Benjamini-Hochberg) is widely accepted and preferable to Bonferroni when you have many tests and limited power. The key is to be explicit about which correction you used and why.

On replication: No independent cohort is a genuine limitation but not a dealbreaker for rare disease research, where cohort availability is a known constraint. Many high-quality papers acknowledge this and still get published in good journals. Frame it as future work.

Power analysis: Yes, report it. Showing you are aware of the limitation is better than not mentioning it.

The Fulbright SoP is very different from a typical PhD SoP. It needs to focus on cultural exchange and diplomacy, not just research.

Strongly agree with the point about choosing venues carefully. One paper in NeurIPS > five papers in obscure workshops.

This is exactly what I needed to hear. One clarifying question: you mentioned FedProx handles non-IID data better.does that mean I should switch to FedProx by default, or only if I observe poor convergence with FedAvg first?

This is a known issue with XLM-R on code-mixed data. The MuRIL preprint has a section specifically comparing these.

Strongly agree with the point about choosing venues carefully. One paper in NeurIPS > five papers in obscure workshops.

I have a slightly different take from my experience in industry research. The reproducibility crisis is real but unevenly distributed. Fields with strong engineering culture (computational biology, ML with benchmarks) have actually improved significantly in the last 5 years. The bigger problem is in fields where data sharing is structurally difficult.clinical medicine, behavioral economics.

The ML community's move toward open code and reproducibility checklists has been genuinely effective.

I found that writing a paper abstract first (before the paper) and then reverse-engineering the paper from the abstract helped with clarity enormously.

As a senior faculty member (25 years in the system), I want to offer a different perspective. Yes, the pressures are real. But I have also seen the other side: students who produce excellent work, publish honestly, and build reputations over time. The system is imperfect but not entirely broken.

What has actually helped my students: choosing venues carefully (a few strong papers rather than many weak ones), developing genuine collaborations rather than transactional coauthorships.

Could you clarify what you mean by results-blind review? I have heard of it but never understood how it works in practice. Wouldn't reviewers need to see results to evaluate whether the methodology is sound?

I think you are conflating two separate problems here. The publish-or-perish culture creates pressure, yes.but the solution is not to criticize researchers who are navigating an unfair system. The real problem is with evaluation committees who treat publication count as a proxy for research quality. Fix the evaluation, and the incentives change.

Also worth noting: open review has been implemented in some fields and has its own problems. It can disadvantage early-career researchers who fear retaliation from senior colleagues they have reviewed critically.

This is really helpful context. One more question: for the ethics committee approval, does the institution where the data is collected or the institution where the analysis is performed need to be the primary ethics approver?

The Fulbright SoP is very different from a typical PhD SoP. It needs to focus on cultural exchange and diplomacy, not just research.

Just to make sure I understand the Zotero workflow correctly.is the .bib file synced to Overleaf automatically every time you add a new paper, or do you need to trigger an export manually?

Thank you for this detailed answer! A quick follow-up: when you mention cosine learning rate with warm-up, are you warming up the LoRA parameters specifically or the entire model including the frozen backbone?