INDEX

Explanations

names of high-ranking positions or specific roles that hold authority

oai_token-act-pair · gpt-3.5-turbo

references to specific roles or titles within organizations or government

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

Īè

-0.68

ridges

-0.65

urtles

-0.60

 à¤

-0.59

 Leafs

-0.58

VIDIA

-0.56

;;

-0.56

Generic

-0.56

sequence

-0.56

¥µ

-0.55

POSITIVE LOGITS

's

0.82

hood

0.76

 liking

0.71

 playbook

0.71

ÃŃs

0.70

rative

0.69

owned

0.68

 accountable

0.68

\'

0.67

 owned

0.64

Activations Density 0.881%

names of high-ranking positions or specific roles that hold authority

references to specific roles or titles within organizations or government

No Comments

No Known Activations