INDEX

Explanations

fully/ently

np_max-act · gemini-2.0-flash

This neuron detects manner-oriented instruction words—especially adverbs and verbs like “gently,” “carefully,” “lean,” or “dry”—that qualify how to perform each step.

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 infect

-0.06

Wer

-0.06

�

-0.06

 stup

-0.06

Mos

-0.06

 unrelated

-0.06

 škola

-0.06

Incorrect

-0.05

ers

-0.05

 дер

-0.05

POSITIVE LOGITS

%D

0.07

yh

0.06

,,,

0.06

irmware

0.06

UMMY

0.06

`;↵↵

0.06

 }↵↵↵

0.06

 Vladimir

0.06

ilio

0.06

 chained

0.06

Activations Density 0.011%

fully/ently

This neuron detects manner-oriented instruction words—especially adverbs and verbs like “gently,” “carefully,” “lean,” or “dry”—that qualify how to perform each step.

No Comments

No Known Activations

fully/ently

This neuron detects manner-oriented instruction words—especially adverbs and verbs like “gently,” “carefully,” “lean,” or “dry”—that qualify how to perform each step.

No Comments

No Known Activations