INDEX

Explanations

phrases related to offering apologies

oai_token-act-pair · gpt-3.5-turbo

expressions of apology

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

nesota

-0.77

insula

-0.71

iltration

-0.71

irrel

-0.70

eele

-0.68

 cryptoc

-0.66

arnaev

-0.66

tnc

-0.66

Ranked

-0.66

 infiltration

-0.64

POSITIVE LOGITS

 sorry

0.86

faced

0.78

fully

0.75

GES

0.72

BLE

0.71

 excuse

0.71

 Guilty

0.69

sorry

0.66

giving

0.66

face

0.66

Activations Density 0.012%

phrases related to offering apologies

expressions of apology

No Comments

No Known Activations