INDEX

Explanations

the word "Uncle" along with various names or descriptions following it in the text

oai_token-act-pair · gpt-3.5-turbo

occurrences of familial titles, specifically variations of "uncle" and "aunt."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

mberg

-0.85

isers

-0.74

anwhile

-0.66

acters

-0.63

istas

-0.62

EVA

-0.61

veyard

-0.61

ization

-0.60

 largeDownload

-0.60

rd

-0.59

POSITIVE LOGITS

liest

1.08

 uncle

0.98

hetically

0.91

hest

0.91

aned

0.85

 nephew

0.84

heses

0.80

rieved

0.79

hetics

0.79

 Vernon

0.78

Activations Density 0.028%

the word "Uncle" along with various names or descriptions following it in the text

occurrences of familial titles, specifically variations of "uncle" and "aunt."

No Comments

No Known Activations