INDEX

Explanations

contractions of the form "could not" or "is not."

oai_token-act-pair · gpt-3.5-turbo

the negation of potential actions or possibilities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 newcom

-0.95

 corrid

-0.79

aditional

-0.78

 pione

-0.78

 princ

-0.78

ortunately

-0.78

 satell

-0.74

ò

-0.74

 gobl

-0.73

ThumbnailImage

-0.73

POSITIVE LOGITS

't

1.31

Â´

0.86

nw

0.78

âĤ¬

0.77

âĶĢâĶĢâĶĢâĶĢ

0.77

won

0.77

saf

0.77

ï¿½

0.75

\'

0.74

Activations Density 0.031%

contractions of the form "could not" or "is not."

the negation of potential actions or possibilities

No Comments

No Known Activations