INDEX

Explanations

capitalized words that may represent names or titles

oai_token-act-pair · gpt-3.5-turbo

instances of "Col" followed by a number, indicating references to sections or categories

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

cffff

-0.92

wered

-0.85

natureconservancy

-0.82

shapeshifter

-0.75

 forth

-0.73

ELF

-0.65

butt

-0.64

 Werewolf

-0.63

ãĥ¼ãĥĨ

-0.62

visor

-0.62

POSITIVE LOGITS

onel

1.17

onial

1.12

ossal

1.03

lections

1.02

ombo

1.00

ossus

0.96

Col

0.90

iseum

0.89

umbo

0.88

ophon

0.87

Activations Density 0.010%

capitalized words that may represent names or titles

instances of "Col" followed by a number, indicating references to sections or categories

No Comments

No Known Activations