INDEX

Explanations

keywords related to very specific and detailed information or attributes

oai_token-act-pair · gpt-3.5-turbo

the word "specific" and its variants, indicating an emphasis on specificity in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Psychiat

-0.76

 Rudd

-0.72

 Dickinson

-0.71

 Oaks

-0.71

=-=-=-=-=-=-=-=-

-0.71

 Roads

-0.68

 Dinosaur

-0.67

 Alive

-0.67

Haw

-0.64

OWER

-0.64

POSITIVE LOGITS

ities

1.13

theless

1.02

ivity

0.96

ancial

0.95

specific

0.94

yip

0.93

etheless

0.92

arily

0.90

comings

0.88

ividual

0.86

Activations Density 0.007%

keywords related to very specific and detailed information or attributes

the word "specific" and its variants, indicating an emphasis on specificity in various contexts

No Comments

No Known Activations