INDEX

Explanations

phrases related to authoritative statements or decisions

oai_token-act-pair · gpt-3.5-turbo

instances of the word "the"

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

lihood

-0.91

-0.90

NetMessage

-0.85

iencies

-0.77

mania

-0.69

hyde

-0.69

icides

-0.68

rums

-0.66

omics

-0.65

casts

-0.64

POSITIVE LOGITS

 statement

1.32

 narrator

1.29

 report

1.13

 spokesperson

1.06

 spokesman

1.04

 spokeswoman

1.00

 authors

1.00

 memo

0.99

 document

0.99

 letter

0.95

Activations Density 0.112%

phrases related to authoritative statements or decisions

instances of the word "the"

No Comments

No Known Activations