INDEX

Explanations

phrases related to sources of information or authority

oai_token-act-pair · gpt-3.5-turbo

phrases that convey reliance on evidence or data

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

icably

-0.71

ilyn

-0.70

assies

-0.70

"]=>

-0.69

ovember

-0.68

icer

-0.67

anke

-0.67

mos

-0.66

etsk

-0.66

ipeg

-0.66

POSITIVE LOGITS

 assumption

1.08

 premise

0.98

 principles

0.96

 principle

0.92

 criteria

0.87

 whims

0.87

 assumptions

0.86

 methodology

0.83

 standpoint

0.80

 hypothesis

0.78

Activations Density 0.313%

phrases related to sources of information or authority

phrases that convey reliance on evidence or data

No Comments

No Known Activations