nl13 hours ago
If you are going to go to the bother of fine tuning for trivial problems like subject classification then I think you'll find Scikit Learn with a SGDClassifier on 2-grams will do probably just as well and be under 1MB for the trained classifier.
You can train it in under a minute, and it will work perfectly well on embedded devices.
Small LLMs are good choices for text classification in two cases:
- If you next to provide in-context examples and classifier based on them.
- Your classification goes beyond simple subject-type classifiers. For example, multiple choice question answering is classification where small LLM will work but traditional ML methods won't/
zubiauran hour ago
A small transformer like BERT or variants is a better fit. It only takes a few examples, which can be generated synthetically using an LLM.
Trains quickly and classifies speedily on modern hardware.
Had a lot of fun doing stuff like this years ago, before LLMs were a thing.
djsjajah12 hours ago
Not with 800 examples. If you are going to consider an ngram model, I think you are better off getting a frontier llm to write you an absurd regex.
nl8 hours ago
Hmm maybe. Turns out the author trained a logistic-regression classifier on the embeddings too, but didn't report the results:
https://github.com/thelgevold/fine-tuned-classifier/blob/mai...
IanCal6 hours ago
I would also recommend the approach of using an llm to create the examples, and then train from there.
You can even get fancy and do things like active learning with the llm taking the role of the human annotator and sending in trial statements (and you can use a cheap one for larger gen and a more expensive one for the classification).
I’d be interested in seeing how well LLMs work with writing things like code for what snorkel AI used to have (there was open source code a while back that I assume is still around somewhere, you wrote code that was a low quality set of classifiers and it trained a model around those)
brokensegue10 hours ago
there are models between 2-grams and 600m param models that would be good options. i don't expect a 2-gram to do very well here. also i'm not sure why this model isn't a fine choice if it solves their problem
throwa3562629 hours ago
What would you suggest instead?
stephantul6 hours ago
A non-autoregressive transformer trained with a classification objective.
deepsquirrelnet12 hours ago
If you want to go deeper on language models, try these project ideas:
- Zero-shot encoders like tasksource or GliNER
- Natural language inference: https://huggingface.co/blog/dleemiller/nli-xenc-ways-to-use
- GRPO training
- GEPA prompt tuning Qwen 0.6B (or GEPA, then GRPO)
- Use an embedding model and train a classifier (MLP, logistic, svm)
- Use a larger LLM to generate a synthetic dataset (beware of lack of diversity, mine "seed text" from real sources first)
- Synthetically generate "hard examples" where more than one category may be valid and DPO tune your preferred responses
throwaw127 hours ago
may I ask where did you get the list? I am looking for ways to get involved in going little more deeper on LLMs (I have very high level understanding, but my direct work doesn't involve them, hence I am not familiar with deeper details)
mickael-kerjean12 hours ago
If you are interested in small language model to fine tune, gemma3:270m is quite interesting for its size
nextaccountic11 hours ago
> The model invents new categories (e.g. apartments) and doesn’t stick to the provided list of allowed categories
Can this specific failure mode be solved by providing a grammar that the output must adhere to? (Not sure if Qwen has this feature, it's used for eg. to ensure the output is parseable json)
nl11 hours ago
It can.
It's something that is implemented by the thing that runs the model - eg Llama.cpp - rather than the model itself.
Note that it is hard to make work if you turn thinking on because the grammar gets complicated quickly (I don't recall if Qwen 0.6B can do thinking).
aesthesia9 hours ago
Thinking shouldn't be too hard to deal with---just let the model generate freely until it hits a </think> token, then do constrained decoding, right?
stymaar7 hours ago
Sure, but does llama-cpp support that?
mijoharas2 hours ago
This was my thought as well. I'm surprised that it's not being used here (afaict)
thomascountz8 hours ago
Yes, you can use constrained decoding like logit masking to force all invalid tokens in the vocabulary to -inf, and effectively be removed from selection. I believe llama.cpp exposes this by accepting a formatted grammar.
doubtfuluser9 hours ago
But why using an encoder model instead of a BERT based model? For a pure classification that should be easier to train and work quite well
zwaps10 hours ago
Has anyone compared recently doing something like ModernBERT plus classifier vs. full or lora FT of a small LM like qwen?
kamranjon8 hours ago
I have! I recently compared Gemma 1b to ModernBERT Large for a binary classification task and ModernBERT was the clear winner. It learned faster and performed the task better by a significant margin by the end of training. It seems the bidirectional encoder only architecture works really well for classification tasks, and I think it is related to being bidirectional whereas decoder only models like Gemma (or Qwen) can only “look backwards”. I used a mixture of FFT and LoRA as well as a mixture of CE Loss and SupCon Loss.
pj_mukh9 hours ago
“As an example, the question “When did we replace our pool pump?” will be mapped to a category called “pool” before querying the Index database.”
Cool write up! Really appreciate it but incidentally how does this categorization help you get better retrieval results?
mettamage9 hours ago
Categorization allows for retrieval strategy
pj_mukhan hour ago
because you have a different vector store for each categorization?
What if the question crosses categories?
electroglyph11 hours ago
existing embedding models like alibaba's modernbert tune or one of the jina v5s would probably map query to category automatically. (i.e. store embeddings of each category and calculate cosine sim for each incoming query vs. categories and pick the closest)
also, you could stick a classifier head on a BERT model as another option.
jszymborski13 hours ago
I think the Qwen 0.6B is so cool. It is super fast and as illustrated here it has a clear niche, esp. when fine-tuned.
I'm also interested in it as a student for distillation.
abhashanand150110 hours ago
Do small language models run on cpus or you still need a gpus to run them?
wongarsu2 hours ago
Anything below one billion parameters you can run on the CPU at acceptable speed
For larger sizes you still can, it just becomes slower and slower. For a simple classification task (small input, tiny output, and you can constrain output to a couple tokens) you could even run something like a 4B or 8B model on the CPU
a967 hours ago
I guess that technically depends on the software used to run the model, but in general it's always been possible to run on a CPU (and may even be possible to run on TPU or something else). It's just been slower. Likewise GPU RAM vs system RAM and the bandwidths involved can make hard bottlenecks.
GPU and VRAM (or fast unified RAM) is generally the option that is both available and performant, but especially really small models also run quite well on CPU and system RAM.
avadodin5 hours ago
iGPUs are often slower or only as fast as CPUs when it comes to LLM text generation.
The advantage is mainly in memory bandwidth. External GPUs' internal memory is slightly faster than DDR attached to your CPU.
Other types of "AI" models do make use of the extra compute in GPUs but not LLMs.
throwa3562629 hours ago
Are 0.6b models useful without fine tuning?
Half of the times I ask qwen 0.6b "what is 1 + 2?" it ends up in a thinking loop of "but wait, the user is asking me to ..."
kamranjon8 hours ago
A fun thing I do with Qwen 3.5 0.8b is to take a screenshot of the Hackernews homepage and ask it to give me a JSON representation of the data and it does surprisingly well. With a well structured prompt I think it could be made to be pretty reliable tool for that type of task out of the box.
armcat6 hours ago
I mean it's always nice to play around with sLLM finetuning, but for practical purposes I would always start with a lazy learner using embeddings (something like a small Stella model), pre-embed the topics/categories, embed the question, perform a kNN using cosine distance. You can use an LLM to "expand" the topics before embedding to make them more contextual. This is usually super fast and super simple and gives you a nice baseline. Then I would add a classification head after embedding layer (with maybe some dropout + 2-3 MLP layers) and train my own classifier, and compare that to lazy learner. Only after that would I start finetuning an LLM.
danielhanchen7 hours ago
Very cool write-up and GitHub repo!
crimsoneer6 hours ago
Tangentially related, but the UK Gov Incubator for AI has quite a nifty LLM driven classification pipeline for survey answers.
cold_harboran hour ago
[dead]
VaporJournalAPP10 hours ago
[flagged]
mlpicker9 hours ago
[flagged]
lastdrop8 hours ago
[dead]