INDEX

Explanations

the word "on" in contexts related to making plans or decisions

oai_token-act-pair · gpt-3.5-turbo

phrases that express intentions or plans

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

arta

-0.78

omsky

-0.77

ESA

-0.76

chin

-0.75

Pac

-0.74

MRI

-0.74

Mine

-0.73

externalActionCode

-0.71

displayText

-0.70

SUP

-0.70

POSITIVE LOGITS

 keeping

1.13

 staying

1.12

 seeing

1.07

 preserving

1.07

 sticking

1.07

 getting

1.04

 delivering

1.03

 acquiring

1.02

 having

1.02

 avoiding

1.02

Activations Density 0.096%

the word "on" in contexts related to making plans or decisions

phrases that express intentions or plans

No Comments

No Known Activations