INDEX

Explanations

It appears that Neuron 4 does not respond to any specific content in the provided excerpts, as indicated by the absence of any non-zero activation values in the activations given. Without any non-zero activations to analyze, it is not possible to determine what Neuron 4 is looking for

oai_token-act-pair · gpt-4-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.2.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

pire

-0.72

 largeDownload

-0.69

ihilation

-0.68

ption

-0.66

conservancy

-0.65

wic

-0.64

merce

-0.64

ument

-0.63

IPP

-0.62

 Eternity

-0.62

POSITIVE LOGITS

 referen

0.75

 Lank

0.75

acus

0.67

 Leban

0.63

QUI

0.62

 questioning

0.62

boa

0.61

 compar

0.60

bal

0.60

 Vall

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.