INDEX

Explanations

mentions of police or government operations where they search or seize items

oai_token-act-pair · gpt-3.5-turbo

terms related to police or military actions, particularly raids

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

gran

-0.71

 Alps

-0.71

pure

-0.69

present

-0.66

faith

-0.65

bearing

-0.65

guy

-0.64

peer

-0.64

å¦

-0.64

uber

-0.64

POSITIVE LOGITS

 raided

0.97

ãĥĥãĥĪ

0.93

ccoli

0.85

ĵĺ

0.78

ãĤ¤ãĥĪ

0.76

Ń·

0.76

apons

0.75

ashtra

0.75

ategory

0.73

ataka

0.70

Activations Density 0.011%

mentions of police or government operations where they search or seize items

terms related to police or military actions, particularly raids

No Comments

No Known Activations