INDEX

Explanations

the pronoun 'it' referring to a previously mentioned subject

oai_token-act-pair · gpt-3.5-turbo

the word "it" in various 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

Priv

-0.61

evil

-0.59

hips

-0.59

 Highest

-0.58

quist

-0.58

=>

-0.58

 Buck

-0.58

TED

-0.58

 Unlimited

-0.58

 ears

-0.57

POSITIVE LOGITS

unes

1.08

asca

1.03

chy

1.00

ueller

0.96

self

0.95

alian

0.93

zbollah

0.91

zik

0.87

chwitz

0.86

ÃĥÃĤ

0.85

Activations Density 0.249%

the pronoun 'it' referring to a previously mentioned subject

the word "it" in various contexts

No Comments

No Known Activations