PRISMΔDB
SAE MODEL #2

FEATURE 2161

/ Medical/Scientific Discourse with Ethical/Social Implications
Refresh Stats

This page shows the detailed analysis of a specific feature in the Sparse Autoencoder (SAE). It includes the semantic interpretation generated by the LLM, the top activating documents that trigger this feature, and statistical metrics like density and activation distribution.

Semantic Interpretation
gemma3:12b #3

The positive examples consistently involve medical or scientific research, often with a focus on disease, treatment, or genetic factors. Crucially, these studies frequently intersect with ethical, social, or philosophical considerations. Examples include bioethics, patient experiences, the impact of disease on quality of life, and the moral implications of medical interventions. The negative examples, conversely, deal with purely technical or scientific topics (materials science, DFT calculations, gene editing) lacking this ethical/social dimension. The common thread is a discussion of a medical or scientific topic that also involves a consideration of its impact on individuals or society, or raises ethical questions.

STATISTICS & DISTRIBUTION
Statistics Explained

Density
Fraction of documents where this feature activates at least once.
Higher density = feature appears frequently across the dataset.

Peak Activation
Maximum activation value observed for this feature over all documents.

Activation Histogram
Distribution of all activation values for this feature. Each bar represents a bin (range) of values, and its height shows how many documents fall in that range.

Density
0.00800
Peak Act
3.77
0.0 Max
Global Context
TOP ACTIVATING CONTEXTS
DOC #161 ANALYZE
ACT: 3.7661
Eye on religion: considering the influence of Buddhist and Shinto thought on contemporary Japanese bioethics. Religious traditions can play a significant role in the shaping of bioethical thought. In Japan, traditional Buddhist and Shinto thought continue to influence contemporary bioethical perspectives. To better define this relationship, this p…
DOC #902 ANALYZE
ACT: 2.5355
The process of exiting vegetarianism: an exploratory study. The experience, reasons, and contexts associated with leaving vegetarianism were explored. Interviews were conducted with a convenience sample of 19 ex-vegetarians and 15 continuing vegetarians. Exiting vegetarianism is similar to the process of leaving other important individual identiti…
DOC #12 ANALYZE
ACT: 2.5130
Common allelic variants of exons 10, 12, and 33 of the thyroglobulin gene are not associated with autoimmune thyroid disease in the United Kingdom. Thyroglobulin (Tg) is a major autoantigen for autoimmune thyroid disease (AITD). The Tg gene (Tg) has been mapped to chromosome 8q24, which has recently been linked in two independent studies to AITD. …
DOC #790 ANALYZE
ACT: 2.3914
Depression and anxiety correlate with disease-related characteristics and quality of life in Chinese patients with gout: a case-control study. This study aims to evaluate the prevalence of depression and anxiety and investigate the potential risk factors for depression and anxiety in Chinese gout patients. A self-report survey was administered to …
DOC #484 ANALYZE
ACT: 2.1313
Annual risk of tuberculous infection in rural areas of Junagadh district. Paucity of epidemiological data on tuberculosis in India prompted the National Tuberculosis Institute, Bangalore to embark upon a nation-wide survey to estimate the Annual Risk of Tuberculosis Infection in different parts of the country. The survey in Junagadh district, one …
DOC #669 ANALYZE
ACT: 1.9378
Is there a role for TPN in terminally ill patients with bowel obstruction? There is controversy regarding the use of total parenteral nutrition (TPN) in individuals with metastatic malignancies. The objective of this study was to determine whether a subgroup of patients with intestinal obstruction would benefit from support with TPN. A retrospecti…
TOPOLOGY
CORRELATIONS
W – Weight-space · similarity between decoder vectors (features that point in similar directions in the embedding space).
D – Data / co-activation · features that tend to fire together on the same documents (co-occurrence in the dataset).