INDEX

Explanations

It appears that neuron 4 did not activate for any tokens in the provided examples, suggesting it seems to be looking for a pattern or feature not present in the text samples

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

 rocks

-0.79

inas

-0.72

 rall

-0.70

 scaling

-0.68

 inline

-0.64

 nutshell

-0.64

gem

-0.62

Interstitial

-0.62

 clad

-0.61

pel

-0.60

POSITIVE LOGITS

etsk

0.85

uberty

0.83

olitan

0.78

Seb

0.77

Sov

0.73

 Everybody

0.73

Neb

0.73

 Flavoring

0.71

 Nobody

0.69

ĨĴ

0.69

Activations Density 0.000%

No Known Activations

This feature has no known activations.