INDEX

Explanations

geographical locations or entities from different regions, potentially related to news or events

oai_token-act-pair · gpt-3.5-turbo

geographical locations and proper nouns

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

 Arlington

-0.99

 Allan

-0.98

 Fairfax

-0.89

 Ambrose

-0.89

 Celt

-0.86

 Alexandria

-0.84

 Unicorn

-0.82

 Stall

-0.81

Ung

-0.78

 Goldstein

-0.77

POSITIVE LOGITS

1.43

ij

1.37

ji

1.25

ja

1.25

1.21

jj

1.18

jri

1.14

je

1.14

Js

1.13

jas

1.12

Activations Density 0.355%

geographical locations or entities from different regions, potentially related to news or events

geographical locations and proper nouns

No Comments

No Known Activations