INDEX

Explanations

phrases or descriptions indicating a degree of something being slightly off or problematic

oai_token-act-pair · gpt-3.5-turbo

descriptions of things that are slightly problematic or of moderate concern

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

apons

-0.89

hire

-0.87

 Flavoring

-0.84

itivity

-0.83

itars

-0.80

anwhile

-0.80

andise

-0.78

orsi

-0.77

 Siber

-0.77

utenberg

-0.77

POSITIVE LOGITS

 confused

1.12

 rusty

1.09

 confusing

1.07

bit

1.06

 tricky

1.02

 misunderstood

1.01

 clumsy

1.01

 awkward

1.00

 prick

0.98

 unclear

0.97

Activations Density 0.043%

phrases or descriptions indicating a degree of something being slightly off or problematic

descriptions of things that are slightly problematic or of moderate concern

No Comments

No Known Activations