INDEX

Explanations

places or entities related to government and authority

oai_token-act-pair · gpt-3.5-turbo

references to governmental and institutional entities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

gradient

-0.69

 DRAG

-0.68

luster

-0.63

 Dover

-0.62

eatures

-0.61

Dur

-0.60

ãĥĥãĥĪ

-0.60

ivities

-0.60

 Stain

-0.59

POR

-0.59

POSITIVE LOGITS

who

1.13

who

1.08

 whom

1.07

whose

0.87

 illegally

0.81

 whose

0.78

 swear

0.75

 flock

0.75

 interested

0.74

 subscribing

0.70

Activations Density 0.419%

places or entities related to government and authority

references to governmental and institutional entities

No Comments

No Known Activations