INDEX

Explanations

phrases that indicate alignment or consistency with a specific standard, guideline, or previous action

oai_token-act-pair · gpt-3.5-turbo

references to compliance or alignment with established standards or guidelines

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

CVE

-0.84

ilt

-0.76

ils

-0.74

soever

-0.73

ilts

-0.68

lvl

-0.66

itars

-0.64

loss

-0.63

 Horror

-0.62

 livest

-0.61

POSITIVE LOGITS

backer

0.86

arity

0.76

 vein

0.73

OGR

0.68

 below

0.68

meal

0.67

stad

0.66

dated

0.65

ups

0.64

 lineage

0.63

Activations Density 0.021%

phrases that indicate alignment or consistency with a specific standard, guideline, or previous action

references to compliance or alignment with established standards or guidelines

No Comments

No Known Activations