INDEX

Explanations

strings with the word "it" followed by a number

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "it."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 schooling

-0.71

hips

-0.69

 destro

-0.68

«ĺ

-0.66

 dise

-0.63

 prevail

-0.63

¥ŀ

-0.62

 prevailed

-0.60

 parting

-0.60

 therap

-0.59

POSITIVE LOGITS

anium

1.06

amins

1.01

herer

1.01

unes

0.98

self

0.97

chell

0.94

ople

0.92

rogen

0.92

atis

0.91

hers

0.90

Activations Density 0.034%

strings with the word "it" followed by a number

occurrences of the word "it."

No Comments

No Known Activations