INDEX

Explanations

phrases related to hijacking and attackers

oai_token-act-pair · gpt-3.5-turbo

terms related to hijacking and its perpetrators

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

 Rite

-0.84

-0.83

 Angels

-0.68

 Habit

-0.63

ãģ®éŃĶ

-0.62

 Piercing

-0.62

 Patron

-0.61

 Blessed

-0.60

ç¥ŀ

-0.59

 Teachers

-0.59

POSITIVE LOGITS

hij

1.16

ackers

0.98

inx

0.92

 hijacked

0.90

intosh

0.90

acking

0.87

acker

0.86

ulence

0.86

eering

0.83

ibl

0.81

Activations Density 0.019%

phrases related to hijacking and attackers

terms related to hijacking and its perpetrators

No Comments

No Known Activations