INDEX

Explanations

names of individuals and their colleagues

oai_token-act-pair · gpt-3.5-turbo

references to specific individuals or researchers in scientific contexts

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

Interstitial

-0.83

venge

-0.75

interstitial

-0.74

 treason

-0.72

 oath

-0.71

 punishable

-0.70

Appearances

-0.68

stab

-0.68

 bombed

-0.67

thumbnails

-0.67

POSITIVE LOGITS

 researcher

1.01

 Aviv

0.99

 Krish

0.94

 Heather

0.91

 Melanie

0.90

 biologist

0.90

 Karin

0.90

Zhu

0.90

 Zhang

0.89

earcher

0.89

Activations Density 0.931%

names of individuals and their colleagues

references to specific individuals or researchers in scientific contexts

No Comments

No Known Activations