INDEX

Explanations

phrases related to remote control or manipulation

oai_token-act-pair · gpt-3.5-turbo

references to remote access and control

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

erald

-0.80

glers

-0.78

ī

-0.75

rooms

-0.73

rament

-0.71

pieces

-0.70

============

-0.69

iful

-0.68

oria

-0.68

qs

-0.66

POSITIVE LOGITS

 remotely

0.87

earable

0.82

 sensed

0.79

pil

0.76

 detonated

0.74

 identifiable

0.71

TPPStreamerBot

0.69

accessible

0.68

 charged

0.68

 hacked

0.68

Activations Density 0.021%

phrases related to remote control or manipulation

references to remote access and control

No Comments

No Known Activations