[HN Gopher] Lightweight Safety Classification Using Pruned Langu...
___________________________________________________________________
Lightweight Safety Classification Using Pruned Language Models
Layer Enhanced Classification (LEC) is a novel technique that
outperforms current industry leaders like GPT-4o, LlamaGuards 1 and
8B, and deBERTa v3 Prompt Injection v2 for content safety and
prompt injection tasks. We prove that the intermediate hidden
layers in transformers are robust feature extractors for text
classification. On content safety, LEC models achieved a 0.96 F1
score vs GPT-4o's 0.82 and Llama Guard 8B's 0.71.The LEC models
were able to outperform the other models with only 15 training
examples for binary classification and 50 examples for multi-class
classification across 66 categories. On prompt injection,LEC
models achieved a 0.98 F1 score vs GPT-4o's 0.92 and deBERTa v3
Prompt Injection v2's 0.73. LEC models were able to outperform
deBERTa with only 5 training examples and GPT-4o with only 55
training examples. Read the full paper and our approach here:
https://arxiv.org/abs/2412.13435
Author : sandijean90
Score : 16 points
Date : 2024-12-19 17:49 UTC (5 hours ago)
(HTM) web link (arxiv.org)
(TXT) w3m dump (arxiv.org)
| bberenberg wrote:
| Are these models available for us to try out?
| gdiamos wrote:
| This is really easy to set up - and is much more accurate than
| asking the LLM to predict True/False.
|
| Just feed the outputs of an embedding API into logistic
| regression, e.g. from sklearn. import voyageai
| vo = voyageai.Client() # This will automatically use the
| environment variable VOYAGE_API_KEY. # Alternatively, you
| can use vo = voyageai.Client(api_key="<your secret key>")
| texts = ["Sample text 1", "Sample text 2"] result =
| vo.embed(texts, model="voyage-2", input_type="document")
| print(result.embeddings[0])
|
| https://scikit-learn.org/1.5/modules/generated/sklearn.linea...
| flatpepsi17 wrote:
| I'd pay good money for a local LLM with no "content safety" at
| all.
___________________________________________________________________
(page generated 2024-12-19 23:02 UTC)