INDEX

Explanations

adjectives denoting something that cannot be changed or avoided

oai_token-act-pair · gpt-3.5-turbo

the term "able" in various contexts, indicating potential or capability

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

erva

-1.06

krit

-0.96

itect

-0.84

icion

-0.80

urer

-0.79

hetically

-0.78

umat

-0.78

anson

-0.77

ffield

-0.77

avers

-0.75

POSITIVE LOGITS

 Haram

0.84

 Danger

0.77

 Isles

0.74

âĶĢâĶĢ

0.71

 flaw

0.70

 Islands

0.68

 fools

0.68

 Samar

0.67

 Skies

0.66

cffffcc

0.66

Activations Density 0.045%

adjectives denoting something that cannot be changed or avoided

the term "able" in various contexts, indicating potential or capability

No Comments

No Known Activations