OpenPipe
AI
fine-tuning guide
SHOW HN2023
How HN took it
12 positive, 24 neutral or asking, 1 mixed, 8 critical, out of 45 comments about the project. Thread score +0.04.
Performance: 1 for, 5 againstIdea: 2 for, 2 againstDocs: 4 for, 0 againstPricing: 1 for, 1 againstExecution: 1 for, 0 against
The thread, comment by comment
Each row is a top-level comment and its replies, in HN's order. Select a cell to open that comment on HN.
positive neutral or mixed critical about another commenter or something else the maker
What HN said
What people liked
Informative and well-organized for newcomers
From 3 comments
Genuinely informative reply for those (few) of us on HN who don’t know the details of LLMs, thanks
davidwritesbugs on HNThank you for posting this. I had to go look for your HuggingFace data sets to find the labeled variety you produced with GPT-4, but other than that, everything was easy to follow.
zten on HNThis looks very helpful! I'm just starting out in the ML/LLM space and have an opportunity to work on this at $dayjob, bookmarking as this looks like an excellent resource. Thank…
accrual on HN
Objections
50x cost claim depends on full GPU utilization, which is unrealistic
From 2 comments
> Fine-tuning has one huge advantage though: it is far more effective at guiding a model's behavior than prompting, so you can often get away with a much smaller model. That gets…
minimaxir on HNI think the cost calculation here does not reflect the actual scenario where most people face. In real world scenario, we don't get inputs queued up to millions and wait for the…
szesiongteo on HN
Fine-tuned 7B models are weaker outside training data
From 2 comments
Fine-tuned low parameter LLM's are superficially good but the cracks are obvious if you test them on anything that isn't very strictly tied to the training data. IMO GPT-4 is…
atleastoptimal on HNA 7b model will work for very specific cases, but it will have a hard time drawing parallels between synonims, so you'll need to be extremely careful in building your fine tuning…
avereveard on HN
Claim to replace GPT-4 is inflated given its mixture-of-experts design
From 1 comment
"to replace GPT-3.5/4" Very inflated statement when it comes to GPT4 since it is a MoE model with 8 separate models each an expert in one area, and you can't replace all 8 models…
facu17y on HN
Requests
Guidance on fine-tuning for larger models and codellama
From 2 comments
I have been trying to figure out how to fine tune codellama. Will the llama2 examples work for codellama as well?
braindead_in on HNWhat are hardware requirements for larger models? What can I fine-tune on Nvidia A100? Will it be possible to work with 70b for example?
indeyets on HN
Questions
Fine-tuning versus RAG for private document QA
From 1 comment
I am a little bit confused whether I need fine-tuning or RAG for my use case? My use case is this: I have some private data (say 1000 word documents), I want a QA capability on…
loganathanspr on HN
Whether input/output pairs are required
From 1 comment
"You do this by training an existing model on example input/output pairs that demonstrate the task you want your fine-tuned model to learn." Are fine-tuning datasets required to…
rrherr on HN
Alternatives named
GPT-3.5 and GPT-4 APIs
From 2 comments
For translation jobs, I've experimented with Llama 2 70B (running on Replicate) v/s GPT-3.5; For about 1000 input tokens (and resulting 1000 output tokens), to my surprise,…
ronyfadel on HN> Fine-tuning has one huge advantage though: it is far more effective at guiding a model's behavior than prompting, so you can often get away with a much smaller model. That gets…
minimaxir on HN
DeBERTa for pure classification
From 1 comment
Depends on your use case. If you're doing pure classification then there are smaller encoder-only models like DeBERTa that might get you better performance with a much smaller…
kcorbitt on HN