INDEX

Explanations

apologies or expressions of regret

oai_token-act-pair · gpt-3.5-turbo

instances of the word "sorry" or variations of it

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

tnc

-0.74

rouse

-0.71

krit

-0.68

arnaev

-0.67

irrel

-0.67

eele

-0.67

tein

-0.67

Ranked

-0.66

minecraft

-0.64

 helicop

-0.64

POSITIVE LOGITS

 sorry

1.11

 excuse

0.89

Sorry

0.87

sorry

0.86

 Sorry

0.84

GES

0.83

giving

0.78

 apologies

0.75

 guys

0.74

 pardon

0.71

Activations Density 0.012%

apologies or expressions of regret

instances of the word "sorry" or variations of it

No Comments

No Known Activations