'and' or ',' inside metric pairs

What this neuron does

Fires on the joining word or comma in pairs of test measures, like 'sensitivity and specificity' or 'sensitivity, and'. Turned up, the model drifted toward 'score' and 'recall' and talked about evaluation numbers.

Written by Claude from this neuron's activations and steering outputs. Confidence: medium. Think it is wrong? Flag it (opens a prefilled issue on GitHub).

Fires most on

The text windows where this neuron's activation was highest. Amber marks positive activation; the outlined token is the peak. Hover a token for its strength.

1. #1 peak

   ⏎Mining Frequent Patterns – Frequent Itemset Mining Methods. Classification: Classification by Decision Tree Induction – Bayesian Classification – Rule based Classification - Prediction – Accuracy and Error Measures⏎Cluster Analysis – Types

2. #2 peak

   in three who did not. This demonstrated a 50% positive predictive value, a negative predictive value of 89%, a sensitivity 70% and 75% specificity. Conclusion: Use of CCHR would reduce unnecessary use

3. #3 peak

   2011) have informed these revised guidelines. More information on HPV and cervical cancer.⏎DCEG research provided the initial assessment of the specificity, sensitivity, and appropriate applications of the first-generation HIV antibody testing

4. #4 peak

   with very low false-positive rates (~0.2%). In other words, these tests are about 99.0% sensitive and 99.8% specific.⏎With performance like that one might expect these to be considered diagnostic tests.

5. #5 peak

   the vast majority of elbow fractures. A posterior fat pad sign also has value in detecting occult elbow fractures, where it has high sensitivity (>90%) but only intermediate

6. #6 peak

   recombineering in E. coli". PLoS ONE 8 (2): e57075. . . http://www.plosone.org/article/info%3Adoi%2F10.1371

Steering

Push this neuron up or down while the model writes, and watch what changes. The outputs were generated ahead of time, so nothing runs in your browser.

Most negative on

Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.

1. #1 peak

   Deep Learning?⏎Deep learning is a form of machine learning that uses neural networks with many “deep” layers between the input and output nodes. By training a network on a large data set, a model is created that can be used to make accurate

2. #2 peak

   of the task, such as identifying certain skin cancer symptoms from digital images. This is based on neural networks similar to the human brain, which are fed large amounts of data.⏎A person can however understand an image and commit this to memory from just

3. #3 peak

   Unsupervised machine learning⏎Unsupervised learning is a machine learning technique, in which there is no need to supervise the model. Instead, the model works on its own to discover information. For some use cases, unsupervised learning can

4. #4 peak

   from diagnosing medical conditions to driving cars.⏎One of the most exciting aspects of AI is its ability to learn and adapt. Machine learning algorithms can analyze vast amounts of data to identify patterns and make predictions, and they improve with experience. This capability

Which neurons in earlier layers feed this one, and which later ones it feeds. Amber links excite, blue links suppress; longer bars are stronger.

Activation histogram

How often this neuron fires at each strength. Almost always it does close to nothing (the tall bars in the middle); the snippets above are the rare moments it fired hard, amber for positive and blue for negative.

Logit lens

The words this neuron nudges the model toward (amber) or away from (blue) when it fires. A hint, not a label: many neurons work through later layers instead.