INDEX

Explanations

mentions of individuals who have held previous positions or statuses

oai_token-act-pair · gpt-3.5-turbo

the term "ex-" as a prefix relating to former positions or statuses

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

kson

-0.81

apest

-0.73

Zup

-0.73

eday

-0.69

 dances

-0.69

utical

-0.69

oglu

-0.68

atche

-0.66

 Twain

-0.65

 intervals

-0.65

POSITIVE LOGITS

girlfriend

0.96

communication

0.92

sent

0.92

imposed

0.91

president

0.90

nom

0.90

commun

0.90

vict

0.88

chance

0.88

serving

0.86

Activations Density 0.047%

mentions of individuals who have held previous positions or statuses

the term "ex-" as a prefix relating to former positions or statuses

No Comments

No Known Activations