INDEX

Explanations

verbal expressions indicating disapproval

oai_token-act-pair · gpt-3.5-turbo

expressions of approval or disapproval

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

aunder

-0.94

eworks

-0.77

ocument

-0.70

ixtape

-0.68

hern

-0.66

phabet

-0.62

Clear

-0.62

rez

-0.62

ixt

-0.62

 diarr

-0.60

POSITIVE LOGITS

0.98

 passionately

0.82

 uncond

0.73

 whatsoever

0.73

ļé

0.71

 homosexuals

0.71

Īè

0.69

freedom

0.69

 seeing

0.68

 unres

0.67

Activations Density 0.334%

verbal expressions indicating disapproval

expressions of approval or disapproval

No Comments

No Known Activations