INDEX

Explanations

phrases indicating that something is either too extreme, too dark, or too overwhelming to handle

oai_token-act-pair · gpt-3.5-turbo

phrases indicating difficulty or problematic situations

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

ugal

-0.71

ICO

-0.69

kind

-0.67

ãĤ¼

-0.67

ATIONAL

-0.66

ARA

-0.66

ãĥ©ãĥ³

-0.66

ance

-0.64

issance

-0.64

oi

-0.63

POSITIVE LOGITS

 adequately

0.71

 anymore

0.71

ð

0.69

 admit

0.68

 Siber

0.68

 submer

0.66

 Spoon

0.66

Bod

0.65

 boast

0.65

mop

0.65

Activations Density 0.081%

phrases indicating that something is either too extreme, too dark, or too overwhelming to handle

phrases indicating difficulty or problematic situations

No Comments

No Known Activations