INDEX

Explanations

names of individuals or characters

oai_token-act-pair · gpt-3.5-turbo

names or references related to individuals involved in legal or criminal contexts

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

catentry

-0.81

-0.77

soType

-0.73

AppData

-0.66

å¤©

-0.66

ãĢĲ

-0.64

 [...]

-0.63

[&

-0.62

 Flavoring

-0.60

POSITIVE LOGITS

hower

0.84

espie

0.83

anwhile

0.78

sonian

0.78

hyde

0.77

anyahu

0.76

ierrez

0.75

ashtra

0.74

kefeller

0.73

enegger

0.72

Activations Density 0.198%

names of individuals or characters

names or references related to individuals involved in legal or criminal contexts

No Comments

No Known Activations