INDEX

Explanations

The neuron is looking for instances of the word "im" with values ranging from 9 to 10, specifically in the context of text sequences

oai_token-act-pair · gpt-3.5-turbo

instances of the substring "im"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

NetMessage

-0.86

yards

-0.73

 corners

-0.68

 Chrys

-0.66

 renown

-0.66

 Parenthood

-0.66

 bachelor

-0.66

Downloadha

-0.65

 mysteries

-0.65

 grease

-0.63

POSITIVE LOGITS

im

1.34

agine

1.19

perial

1.15

mediately

1.14

aterial

1.12

imi

1.12

ple

1.12

plementation

1.09

pson

1.08

mediate

1.07

Activations Density 0.013%

The neuron is looking for instances of the word "im" with values ranging from 9 to 10, specifically in the context of text sequences

instances of the substring "im"

No Comments

No Known Activations