INDEX

Explanations

terms related to official approval or permission

oai_token-act-pair · gpt-3.5-turbo

terms related to regulatory approvals and clearances

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Esper

-0.68

Graph

-0.65

vir

-0.65

 advers

-0.64

 passionately

-0.63

 accus

-0.63

hens

-0.62

epend

-0.61

Pug

-0.60

{*

-0.60

POSITIVE LOGITS

 clearance

4.40

 approvals

1.39

 cleared

1.35

 clearing

1.30

 Clear

1.11

 clears

1.11

 approval

1.05

 turnover

1.02

 leakage

1.02

 waivers

1.00

Activations Density 0.010%

terms related to official approval or permission

terms related to regulatory approvals and clearances

No Comments

No Known Activations