INDEX

Explanations

changes over time

np_max-act · gemini-2.0-flash

This neuron preferentially lights up on mid-frequency content words (e.g. common nouns, verbs, and adjectives) rather than on very common function words or extremely rare tokens.

oai_token-act-pair · o4-mini Triggered by @xinyanhu8

New Auto-Interp

Configuration

andyrdt/saes-llama-3.1-8b-instruct/resid_post_layer_11/trainer_1

Dataset (Dashboard)

Various

Features

131,072

Data Type

float32

Hook Name

blocks.11.hook_resid_post

Architecture

standard

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

clang

-0.07

 independently

-0.06

ीछ

-0.06

 diplomats

-0.06

NavigationItemSelectedListener

-0.06

PB

-0.06

compat

-0.06

 phishing

-0.06

 vista

-0.06

 sweets

-0.06

POSITIVE LOGITS

 VALUE

0.07

mando

0.07

acr

0.06

.expand

0.06

 rendered

0.06

añ

0.06

[source

0.06

ikipedia

0.06

 oldu

0.06

μερα

0.06

Activations Density 0.133%

changes over time

This neuron preferentially lights up on mid-frequency content words (e.g. common nouns, verbs, and adjectives) rather than on very common function words or extremely rare tokens.

No Comments

No Known Activations